The AI infrastructure cycle continues producing measurable financial confirmation across multiple layers of the stack. This week, the strongest signals came from optical networking, memory, custom silicon, and the physical power and cooling infrastructure required to support continued data-center expansion.

The important change is increasingly visible in earnings: companies exposed to the bottlenecks are not simply talking about future demand. Revenue is accelerating, backlogs remain elevated, capacity is being expanded, and customers continue committing capital years ahead of deployment.

At the same time, valuations across several AI infrastructure names have risen dramatically. The next phase will increasingly separate companies converting scarcity into earnings from those benefiting primarily from expectations.

New This Week

Optical Networking: Ciena Delivers Another Confirmation

Ciena provided one of the clearest new signals this week.

Fiscal third-quarter revenue increased 37% year over year to $1.67 billion, while adjusted EPS reached $2.11, ahead of expectations. Management specifically cited AI-driven network investment and raised full-year revenue guidance to approximately $6.42 billion.

This follows strong August results across the optical group.

Lumentum reported fiscal Q4 revenue of $1.01 billion, with non-GAAP gross margin of 50.4%, and guided the September quarter to $1.225–$1.275 billion of revenue. Management said 1.6T cloud modules, optical circuit switching, CPO lasers and other next-generation products are beginning to layer into growth.

Coherent reported Q4 revenue of $2.05 billion, up 34%, while non-GAAP EPS increased to $1.74. Management also highlighted expanding production capacity to address accelerating customer demand.

Applied Optoelectronics delivered its fifth consecutive quarter of record revenue and said demand for 800G and 1.6T products is expected to outpace production capacity through mid-2027.

Investment signal: Photonics has moved decisively from a projected bottleneck into measurable revenue and capacity constraints.

CRI Watchlist: LITE • COHR • AAOI • CIEN • CRDO

Memory Shortage Is Broadening

Memory may be becoming an even stronger bottleneck than previously expected.

Dell recently highlighted DRAM and NAND availability as constraints on AI-server supply. Industry estimates now point toward substantial additional pricing increases as suppliers prioritize AI-related memory production.

More importantly, SK hynix now expects the memory shortage could persist through 2030, despite continuing industry capacity expansion. Its new Indiana AI-memory facility represents a $4 billion investment and is expected to begin HBM4E production in 2029.

Micron separately announced plans to invest $10 billion over the next decade in an AI-memory research facility in Boise.

Micron’s latest reported quarter already showed extraordinary economics in its data-center businesses, including substantial revenue growth and margin expansion, while HBM4 entered high-volume shipments.

Investment signal: The constraint is expanding beyond HBM. Capacity diverted toward AI memory is tightening conventional DRAM and NAND as well.

CRI Watchlist: MU • STX • WDC • LRCX • AMAT • KLAC

Custom Silicon & Networking

Marvell’s latest earnings reinforced another important layer of the infrastructure cycle.

Fiscal Q2 revenue reached a record $2.739 billion, up 37%, while Data Center revenue increased 46% year over year. Management described AI bookings as exceptionally strong and again increased its fiscal 2027 and 2028 revenue outlooks.

The stock nevertheless sold off as investors questioned how quickly recently announced custom-silicon agreements would convert into revenue. That distinction matters.

Broadcom provided another confirmation this week, raising its outlook for AI chip revenue as hyperscalers continue increasing infrastructure commitments.

Investment signal: AI demand remains strong, but the market is becoming less willing to pay immediately for revenue expected several years into the future.

CRI Watchlist: MRVL • AVGO • CRDO

Power, Transformers & Electrical Infrastructure

The physical power layer remains one of the strongest areas of the entire AI capital cycle.

Recent global reporting shows transformer, power-management and cooling suppliers expanding manufacturing as data-center infrastructure investment accelerates. McKinsey estimates cited by Reuters point toward nearly $7 trillion of global data-center investment through 2030.

The financial evidence from U.S. electrical suppliers remains compelling.

Powell Industries’ latest quarter produced:

$934 million new orders — up 158%
$2.4 billion backlog — up 69%
3.0x book-to-bill

The company also secured a data-center order exceeding $400 million.

Eaton reported Electrical Americas orders up 41% on a rolling twelve-month basis, while backlog increased 33%. Electrical Global backlog increased 103% year over year, with the combined electrical businesses maintaining a 1.2 book-to-bill ratio. Eaton also raised full-year organic growth guidance.

Investment signal: Transformer, switchgear and electrical-distribution demand remains supported by actual orders and backlog rather than forecasts alone.

CRI Watchlist: POWL • ETN • HUBB • NVT • GEV • PWR

Cooling Infrastructure

Cooling remains another strongly confirmed bottleneck.

Vertiv’s latest reported quarter showed revenue increasing 24% to $3.27 billion, adjusted operating profit rising 51%, and adjusted free cash flow increasing 234%. Management raised full-year guidance across its major financial metrics.

Modine’s data-center revenue increased 90% year over year in its latest quarter following its previously announced $4 billion long-term cooling-capacity agreement.

An important development further down the infrastructure stack came from Eaton and Trane. The companies introduced an integrated AI data-center power and cooling design that they say can improve energy efficiency by up to 15%, reduce installation costs by as much as 30%, and reduce copper usage by up to 80%.

Investment signal: Cooling is increasingly merging with electrical architecture rather than remaining a separate HVAC market.

CRI Watchlist: VRT • MOD • TT • ETN • AAON

Commodities & Materials

The physical intensity of AI infrastructure is becoming easier to quantify.

A recent academic analysis of 20 minerals used across AI data-center infrastructure found that power infrastructure—not semiconductors—accounts for most modeled mineral demand. Copper represented 83% of total mineral mass, with grid transmission and distribution responsible for 64% of modeled copper requirements.

That strengthens the longer-duration CRI view that AI capital spending ultimately reaches far beyond servers.

Investment signal: Copper exposure remains a slower-moving derivative of the AI cycle, but the infrastructure buildout continues strengthening the structural demand case.

CRI Watchlist: FCX • SCCO

Capital & Backlog Check

🔴 Electrical Infrastructure

Signal: Very Strong

Powell: $2.4B backlog, +69% YoY
Eaton: Electrical Global backlog +103%

CRI Watchlist: POWL • ETN • HUBB • NVT

🟢 Optical Networking

Signal: Accelerating

Ciena: revenue +37%
Coherent: revenue +34%
Lumentum: strong forward guidance
AAOI: demand exceeding production capacity

CRI Watchlist: LITE • COHR • AAOI • CIEN

🔴 Memory

Signal: Tightening

SK hynix sees shortage potentially lasting through 2030 while Micron and SK hynix commit billions to additional U.S. investment.

CRI Watchlist: MU • LRCX • AMAT • KLAC

🟢 Cooling

Signal: Strong

Vertiv raised guidance while Modine continues reporting rapid data-center growth.

CRI Watchlist: VRT • MOD • TT

🟡 Custom Silicon

Signal: Strong demand / valuation-sensitive

Marvell and Broadcom continue showing strong AI demand, but investors are becoming increasingly focused on the timing of revenue conversion.

CRI Watchlist: AVGO • MRVL

Emerging Constraint Watchlist

🔴 Transformer & Switchgear Capacity

Status: Still constrained

Powell and Eaton backlog data continue confirming demand.

Watch: POWL • ETN • HUBB • NVT

🔴 Memory Availability

Status: Worsening

AI demand is increasingly affecting DRAM and NAND availability in addition to HBM.

Watch: MU • LRCX • AMAT

🟠 Optical Manufacturing Capacity

Status: Tight

AAOI explicitly expects demand to exceed production capacity through mid-2027, while larger optical suppliers continue expanding manufacturing.

Watch: AAOI • LITE • COHR

🟠 AI Power Density

Status: Emerging

Electrical and cooling architectures are increasingly being designed together as rack power rises.

Watch: VRT • ETN • TT • MOD

🟠 Copper Availability

Status: Long-duration pressure

Grid infrastructure remains significantly more material-intensive than the chips sitting inside the data center.

Watch: FCX • SCCO

What We’re Watching Next

The most important near-term confirmation will be whether strong demand continues converting into higher earnings rather than simply larger capital commitments.

We are watching:

• Micron’s September 30 earnings and memory pricing commentary

• Additional optical capacity expansion

• Powell and Eaton backlog conversion

• Vertiv and Modine margin performance as capacity expands

• Transformer manufacturing investment

• Hyperscaler power commitments

• Copper project development

• Marvell’s October 6 Investor Day for more detail on custom silicon

• Evidence that AI infrastructure spending remains strong into 2027

Final CRI Read

This week’s evidence strengthens the AI bottleneck thesis while also changing its character.

Photonics is now delivering earnings confirmation. Memory shortages appear to be broadening. Electrical infrastructure backlog remains exceptional. Cooling companies continue converting AI demand into revenue and cash flow.

The investment question is therefore becoming more selective.

The strongest opportunities are increasingly companies where scarcity is already visible through:

Orders • Backlog • Capacity constraints • Revenue growth • Margin expansion • Capital commitments • Raised guidance

For this week’s CRI ranking, the strongest financial confirmation is:

1. Electrical infrastructure
2. Optical networking and photonics
3. Memory and HBM
4. Cooling and thermal management
5. Custom silicon and networking
6. Copper and physical materials

The AI buildout remains intact. But the next phase may reward investors who follow where the money is actually being spent—and where suppliers still cannot build capacity fast enough.

The MicroCap Discovery model became substantially more active this week.

The screen evaluated 341 companies, with 80 names reaching the Promotion Watchlist, another 64 sitting in Accumulation Watch, and 197 remaining in Early Monitor.

Perhaps most notable: zero companies landed in the Reject / No Flow bucket.

But the opportunity isn’t evenly distributed.

Healthcare/Biotech has become the overwhelming leader at the top of the screen. Beneath it, however, several less-developed rotations are appearing in AI Software/Data, Materials/Commodities, Energy/Power and Financials.

That second layer may ultimately be more interesting for discovery.

The Weekly Dashboard

341 names screened

80 — Promotion Watchlist

64 — Accumulation Watch

197 — Early Monitor

0 — Reject / No Flow

This is a considerably healthier discovery environment than one where only a handful of stocks meet the model’s structural requirements.

But there is an important distinction:

More opportunity does not mean broader leadership.

The highest-scoring portion of the screen is heavily concentrated.

Promotion Watchlist — Biotech Takes Control

The Promotion Watchlist expanded to 80 companies.

Healthcare/Biotech alone accounts for 51 of them.

That is nearly two-thirds of the entire promotion universe.

The highest-ranked setups this week are:

🥇 KURA — Kura Oncology

CRI Score: 99

Theme: Healthcare / Biotech
Structure: Compression + Capital Flow

KURA is the highest-scoring company in the entire screen.

Trend, relative strength and capital flow all received maximum scores of 20.

The combination is particularly interesting because this isn’t simply momentum chasing. The model identifies compression occurring alongside strong capital-flow behavior.

CRI Read: Highest-priority Promotion Watchlist setup this week.

🥈 CRON — Cronos Group

CRI Score: 95

Theme: Healthcare / Biotech
Structure: Confirmed Trend Leadership

CRON also carries maximum Trend, Relative Strength and Capital Flow scores.

The stock is sitting at its modeled 52-week high while reported revenue growth in the dataset is 58.4% year over year.

CRI Read: Strong confirmed leadership rather than an early discovery setup.

🥉 OABI — OmniAb

CRI Score: 94

Theme: Healthcare / Biotech
Structure: Confirmed Trend Leadership

OABI combines maximum Trend, RS and Capital Flow scores with unusually strong reported revenue growth in the screen.

The model also shows elevated recent volume activity.

CRI Read: One of the strongest combinations of operating growth, price structure and capital flow in this week’s universe.

VIR — Vir Biotechnology

CRI Score: 94

Structure: Compression + Capital Flow

VIR also receives maximum scores for Trend, RS and Capital Flow.

The distinction from some of the other leaders is its compression structure.

CRI Read: A high-scoring setup where the model sees capital flow developing alongside tightening structure.

EOLS — Evolus

CRI Score: 93

Structure: Confirmed Trend Leadership**

Trend: 20
Relative Strength: 20
Capital Flow: 20

EOLS is another example of why Healthcare/Biotech dominates this week’s Promotion Watchlist.

REPL — Replimune Group

CRI Score: 92

Structure: Compression + Capital Flow

Maximum Trend, RS and Capital Flow scores put REPL firmly among this week’s highest-conviction technical setups.

ACHV — Achieve Life Sciences

CRI Score: 92

Structure: Compression + Capital Flow

Another compressed Biotech setup with maximum Trend, RS and Capital Flow readings.

Other High-Priority Promotion Names

Several additional companies deserve monitoring:

WEAV — Weave Communications — Score 89
Confirmed Trend Leadership with maximum Trend, RS and Capital Flow scores.

AQST — Aquestive Therapeutics — Score 89
Strong trend and accumulation characteristics despite remaining roughly 29% below its 52-week high.

XNCR — Xencor — Score 88
Strong trend and capital flow, approximately 6% below its 52-week high.

SLN — Silence Therapeutics — Score 87

IMMX — Immix Biopharma — Score 87

SGMT — Sagimet Biosciences — Score 87

ORIC — ORIC Pharmaceuticals — Score 87

OMER — Omeros — Score 87

The concentration is impossible to miss.

Biotech isn’t simply leading. It is dominating the Promotion Watchlist.

Theme Leadership

The theme summary gives us another way to see the rotation.

🟢 Healthcare / Biotech

97 names screened
51 Promotion
19 Accumulation
Average Score: 66.5

Nothing else is close.

Healthcare/Biotech also carries the strongest average capital-flow score at 17.4.

This is established leadership.

🔵 Financials

16 names
3 Promotion
4 Accumulation
Average Score: 53.0

Financials don’t have Biotech’s numbers, but their average score is actually the second highest of all themes.

Average relative-strength score is also very strong at 14.1.

This fits remarkably well with this week’s broader Sector Rotation model, where Financials moved into the #1 sector position.

The signals are beginning to reinforce one another.

🟡 Materials / Commodities

29 names
2 Promotion
9 Accumulation
Average Score: 50.2

This may be one of the more interesting early rotations.

Only two names have graduated to Promotion, but nine are already in Accumulation Watch.

That is precisely the relationship we want to see when looking for a theme before it becomes crowded.

🟡 Energy / Power

30 names
7 Promotion
2 Accumulation
Average Score: 47.8

Energy has more confirmed leadership than Materials, with seven Promotion names.

Its capital-flow score also remains relatively healthy.

This fits the broader Energy strength we’ve been tracking across the CRI models.

AI Is Building — But It Hasn’t Broken Out Yet

This is where the report gets especially interesting.

AI Software / Data

60 names screened

9 Promotion

10 Accumulation

Average Score: 42.9

There is activity here.

But AI Software/Data remains well behind Healthcare/Biotech in average score and capital-flow intensity.

The Accumulation Watch contains several notable AI Software/Data names:

SPT — Sprout Social

Score: 69

Confirmed Trend Leadership and approximately 24% below its 52-week high.

BLZE — Backblaze

Score: 68

Confirmed Trend Leadership with positive capital-flow characteristics and approximately 25% below its 52-week high.

These are closer to the type of setups MicroCap Discovery was designed to surface:

Improving structure before the stock becomes an obvious momentum trade.

AI Infrastructure Hasn’t Joined Yet

There is an important contrast.

AI Infrastructure / Semiconductors

8 names screened

0 Promotion

0 Accumulation

Average score: 37.1

AI Optical / Networking

5 names screened

0 Promotion

0 Accumulation

Average score: 29.0

That’s not a reason to abandon these themes.

It tells us the opposite of what we see in Biotech:

The microcap market isn’t confirming an AI hardware rotation yet.

This is particularly useful because our broader research remains constructive on AI infrastructure over the long term.

The MicroCap model is simply saying:

The small-company technical and capital-flow confirmation isn’t there yet.

We wait rather than force it.

Accumulation Watch — Where the Next Rotation May Be Forming

The 64-name Accumulation Watch is arguably more important for discovery than the 80 Promotion names.

Promotion tells us what is already working.

Accumulation can tell us what may work next.

Several names stand out.

ASST — Strive

Score: 69 | Financials

Confirmed Trend Leadership with strong relative strength.

The stock remains dramatically below its 52-week high, making this a very different setup from many of the fully extended Promotion names.

GROY — Gold Royalty

Score: 69 | Materials / Commodities

Structure: Strong Trend + Accumulation

Capital Flow: 15
Relative Strength: 20

Approximately 33% below its 52-week high.

This is one of the more interesting Materials setups in the screen.

ITRG — Integra Resources

Score: 69 | Materials / Commodities

Another Strong Trend + Accumulation setup.

Relative Strength is already at 20, while Capital Flow scores 15.

Approximately 39% below the 52-week high.

QMCO — Quantum

Score: 69

Confirmed Trend Leadership.

Trend and Relative Strength both score 20, while the stock remains roughly 15% below its 52-week high.

SPT — Sprout Social

Score: 69 | AI Software / Data

An interesting AI software candidate where relative strength has improved before the stock has returned to its prior high.

BLZE — Backblaze

Score: 68 | AI Software / Data

Another AI Software/Data name showing confirmed trend characteristics without being near its 52-week high.

VZLA — Vizsla Silver

Score: 67 | Materials / Commodities

Strong Trend + Accumulation.

The stock remains roughly 40% below its 52-week high, yet Relative Strength is already at the model maximum.

That’s exactly the type of divergence worth monitoring.

RES — RPC

Score: 67 | Energy / Power

Structure: Compression + Capital Flow

An Energy setup worth watching as the broader Energy theme remains structurally strong.

The Potential Next-Rotation List

Rather than chasing the highest-scoring stocks, this is the group I would watch for graduation into stronger model status:

GROY — Gold Royalty
Materials / Commodities

ITRG — Integra Resources
Materials / Commodities

SPT — Sprout Social
AI Software / Data

BLZE — Backblaze
AI Software / Data

VZLA — Vizsla Silver
Materials / Commodities

RES — RPC
Energy / Power

CLBK — Columbia Financial
Financials

PRME — Prime Medicine
Healthcare / Biotech

These aren’t necessarily the strongest stocks today.

That’s the point.

They represent places where structure and/or accumulation may be developing before full Promotion status.

What Changed in the Opportunity Set?

Three distinct layers are emerging.

Layer 1 — Established Leadership

Healthcare / Biotech

This is where the model has the most confirmation and the most Promotion names.

The opportunity is real—but increasingly obvious.

Layer 2 — Developing Rotation

Financials
Materials / Commodities
Energy / Power
AI Software / Data

These themes have enough Accumulation and Promotion activity to deserve attention without showing Biotech-level crowding.

This is where discovery becomes more interesting.

Layer 3 — Waiting for Confirmation

AI Infrastructure / Semiconductors
AI Optical / Networking

Long-term themes remain compelling.

The microcap model simply hasn’t confirmed them.

That makes them watchlists rather than trades generated by this screen.

CRI Bottom Line

The MicroCap Discovery model is sending a much stronger signal this week.

80 of 341 screened companies have reached Promotion status, while another 64 are showing Accumulation characteristics.

Healthcare/Biotech is overwhelmingly dominant, accounting for 51 Promotion names and producing the highest average theme score.

But I would not focus exclusively on the obvious leader.

The more interesting discovery signal may be occurring one layer underneath.

Financials have strong relative strength.

Materials have nine names accumulating but only two promoted.

AI Software/Data has ten Accumulation names and nine Promotions.

Energy continues producing confirmed leadership.

Meanwhile, AI semiconductors and optical/networking have yet to confirm.

That gives us a useful roadmap heading into September:

Biotech is working now.

Financials, Materials, Energy and AI Software are worth watching for broader rotation.

AI hardware remains on deck—but the microcap money flow hasn’t arrived yet.

And that last distinction is central to the purpose of this model.

We aren’t trying to predict which story eventually becomes popular.

We’re trying to detect when capital starts moving there before everyone else notices.

The broad sector rankings tell us where capital is moving. The subsector models tell us what investors are actually buying inside those sectors.

That distinction is becoming increasingly important.

This week’s data show a market that remains willing to reward structural growth, but participation is uneven. Health Care retains excellent internal breadth and momentum. Energy remains structurally strong. Communication Services continues to improve. Industrials show unusually strong leadership concentration despite weaker broad-sector performance. Consumer participation deteriorated sharply.

The message underneath the market is increasingly clear:

Owning the right theme matters more than simply owning the right sector.

The Weekly Subsector Dashboard

🟢 HEALTH CARE — Strongest Broad Internal Structure

SSM Score: 3.00 / 5
Weekly change: -0.19

Health Care remains one of the healthiest structures in the entire model.

The standout readings are:

Breadth: 0.87
Momentum: 0.88

Those are powerful readings.

The sector weakened somewhat this week, but the decline did not destroy its underlying participation. This is important because it separates Health Care from sectors where performance depends upon only a handful of companies.

The model is showing broad participation combined with strong momentum.

Theme Read

The setup continues to favor selective exposure across:

Biotechnology
Medical innovation
Life sciences
Diagnostics
Specialty pharmaceuticals
Higher-growth health-care businesses

The key is that this is not merely a defensive Health Care trade.

There is meaningful participation underneath the sector.

CRI Read: 🟢 Working

Health Care remains one of the better places to search for individual opportunities.

🟢 ENERGY — The Structural Story Remains Intact

SSM Score: 67.0 / 100
Weekly change: -1.54

Energy remains one of the strongest structural models despite cooling slightly this week.

Its internal components remain impressive:

Structural Trend: 14.3 / 20
Crack & Refining: 17.2 / 20
Commodity Acceleration: 12.0 / 20
Breadth: 17.0 / 20

The standout continues to be refining.

That matters because the Energy opportunity isn’t simply a directional bet on crude oil.

The model continues to identify strength across the energy value chain.

Theme Read

Still constructive:

Integrated energy
Refining
Midstream infrastructure
Natural gas infrastructure
LNG
Energy transportation

More selective:

Oilfield services
Higher-cost producers
Companies requiring materially higher commodity prices

This continues to fit the larger CRI thesis that energy infrastructure may ultimately matter as much as energy production itself.

AI data centers, electrification, manufacturing reshoring and grid expansion all increase the value of reliable energy supply.

CRI Read: 🟢 Working — but increasingly selective

🟢 COMMUNICATION SERVICES — Quiet Improvement

SSM Score: 2.81 / 5
Weekly change: +0.07

Communication Services isn’t generating the attention of Technology, but its internal structure continues to improve.

The strongest component is:

Momentum: 0.79

Breadth is also reasonably healthy at 0.61.

Relative strength, however, remains only moderate.

That combination suggests something important.

This is an improving rotation rather than established leadership.

Theme Read

The model’s underlying universe tracks leadership including large digital-platform, streaming, entertainment and telecom businesses.

The opportunity therefore remains selective rather than sector-wide.

CRI Read: 🟢 Improving

This is a group worth watching for continued follow-through.

🟡 INDUSTRIALS — Weak Sector, Strong Leaders

SSM Score: 2.89 / 5
Weekly change: -0.26

Industrials may be the most interesting contradiction in this week’s report.

The broad Sector Rotation Model showed significant deterioration.

But underneath that weakness:

Leadership: 0.77

and, more importantly,

Equal-Weight Leadership: 0.99

That is an extraordinary reading.

Meanwhile:

Relative Strength: only 0.20
Momentum: only 0.27

This tells us something very different from simply saying “Industrials are weak.”

There are very strong industrial winners operating inside a weakening broad sector.

That is exactly the environment where thematic stock selection becomes important.

Theme Read

The underlying Industrial workbook includes companies exposed to areas such as:

Electrical equipment
Power infrastructure
Automation
Aerospace & defense
Heavy equipment
Transportation
Industrial machinery

This aligns closely with what we’ve been seeing elsewhere.

Broad Industrials can weaken while AI power infrastructure, electrical equipment, automation and selected capital-equipment companies continue to outperform.

That divergence is important.

CRI Read: 🟡 Do not buy the sector indiscriminately. Follow the leaders.

🔵 REAL ESTATE — Improving Internals, Weak Relative Strength

SSM Score: 2.17 / 5
Weekly change: +0.27

Real Estate improved this week, but the model isn’t signaling broad leadership.

The biggest problem remains:

Relative Strength: 0.06

That is extremely weak.

But breadth has climbed to 0.71.

This creates an interesting setup.

More Real Estate companies are participating, yet the sector itself continues to underperform the broader market.

Theme Read

The workbook tracks major Real Estate leadership including:

Data-center infrastructure
Communications infrastructure
Industrial/logistics property
Commercial property

That makes the sector particularly relevant to our infrastructure research.

The important distinction remains:

Data-center-related real estate can have a completely different structural outlook than traditional commercial real estate.

CRI Read: 🔵 Potential bottoming/improvement, but not yet leadership.

🔴 CONSUMER — Internal Participation Deteriorates

SSM Score: 1.77 / 5
Weekly change: -0.67

Consumer produced one of the clearest negative signals this week.

The weekly decline in its structural score was substantial.

The most concerning component is:

Breadth: 0.20

Equal-weight leadership is similarly weak at 0.19.

That means whatever strength remains is increasingly concentrated rather than broadly shared.

The workbook specifically tracks leadership relationships involving major discretionary and staples businesses, including Amazon, Tesla, Home Depot, Walmart, Costco and Procter & Gamble.

The message isn’t that every consumer company is weak.

It’s that broad consumer participation is poor.

CRI Read: 🔴 Not working broadly.

🔴 UTILITIES — Still Structurally Weak

SSM Score: 1.50 / 5
Weekly change: +0.27

Utilities improved slightly this week, but remain the weakest structural model in this group.

Relative strength is exceptionally poor:

0.03

Breadth is only:

0.11

Momentum:

0.17

There is one interesting contradiction.

Equal-weight leadership is relatively healthy at 0.73.

Again, that points toward individual winners existing inside a broadly weak group.

This matters tremendously for the AI power thesis.

The market is distinguishing between:

Traditional regulated utility exposure

and

companies directly benefiting from power-demand growth, transmission investment, generation additions and data-center load growth.

They are not the same investment.

CRI Read: 🔴 Utilities broadly remain weak; power infrastructure remains a stock-selection story.

⚠️ AI MODEL — Data Integrity Warning

The AI workbook reports a zero structural score across every module for August 28.

I am not interpreting that as an AI collapse.

The underlying workbook contains populated market series, but its current dashboard/export layer is returning zeros.

The same problem appears in the Financials workbook, while the Materials export is not producing a completed current score.

Those models therefore should be treated as data unavailable this week, rather than given artificial bullish or bearish classifications.

That distinction matters.

No signal is better than a false signal.

What Is Actually Working?

Pulling the models together produces a much more useful picture than looking at headline sectors alone.

🟢 Strong / Working

Health Care innovation

Broad participation and strong momentum remain intact.

Energy infrastructure & refining

The Energy model remains structurally healthy, with refining particularly strong.

Selected Industrial infrastructure

Leadership is extremely concentrated, but the strongest industrial companies remain powerful.

Communication Services

Momentum continues improving and breadth is constructive.

🟡 Emerging / Selective

Data-center real estate

Real Estate remains weak versus the market, but improving breadth deserves attention.

Industrial automation

Broad Industrials are deteriorating, but leadership concentration suggests selected structural winners remain strong.

Power infrastructure

The Utilities sector itself remains weak, reinforcing our thesis that the better opportunity may reside in the equipment and infrastructure providers rather than simply buying regulated utilities.

🔴 Not Working Broadly

Consumer

Breadth deteriorated significantly this week.

Traditional Utilities

Relative strength and breadth remain exceptionally weak.

Broad Real Estate

Participation is improving, but relative strength remains poor.

Broad Industrials

The sector is no longer a simple beta trade. Leadership has become highly concentrated.

The AI Infrastructure Read-Through

This week’s subsector data reinforce one of the most important themes we’ve been developing.

The AI buildout should not be viewed as one Technology trade.

It increasingly touches:

Semiconductors → Networking → Data centers → Cooling → Electrical equipment → Transformers → Copper → Grid infrastructure → Natural gas → Generation → Energy transportation

And the market isn’t pricing all of those layers equally.

That creates opportunity.

The Industrial model is particularly revealing.

Broad Industrials weakened while its equal-weight leadership reading approached 1.0.

In other words:

The market is aggressively differentiating between industrial companies.

That is exactly what we would expect as investors begin separating generic cyclicals from companies exposed to multi-year infrastructure capital spending.

CRI Theme Map — Heading Into September

🟢 Favor

Health Care innovation
Energy infrastructure
Refining
Midstream / natural gas infrastructure
Selected electrical equipment
Power infrastructure suppliers
Industrial automation
Selected Communication Services

🟡 Watch Closely

Data-center real estate
Materials tied to grid expansion
Semiconductors and AI infrastructure
Aerospace & defense
Industrial machinery

🔴 Exercise Caution

Broad Consumer exposure
Traditional Utilities
Broad Real Estate
Undifferentiated cyclical Industrials

CRI Bottom Line

The headline sector rankings are becoming less useful without understanding what’s happening underneath them.

This week’s subsector models show why.

Industrials can weaken while selected industrial infrastructure companies remain powerful.

Utilities can rank near the bottom while the AI power-infrastructure thesis strengthens.

Real Estate can underperform while data-center-related infrastructure remains structurally attractive.

Health Care can experience a weak week while maintaining some of the strongest breadth and momentum in the entire model.

That is the market we’re entering September with.

Not everything tied to AI will win.

Not everything tied to infrastructure will win.

And owning the right sector may no longer be enough.

The next phase of this market looks increasingly likely to reward investors who identify the right layer inside the right sector.

That’s exactly what the subsector model is designed to find.

Data through August 28, 2026

This week produced a meaningful reshuffling beneath the surface. The most important development is not simply which sector sits at No. 1. It is the sharp recovery in Technology occurring at the same time Industrials and other previously resilient cyclical groups are losing momentum.

Financials, Health Care, and Energy remain the strongest structural groups. But Technology’s move from No. 8 to No. 4 in a single week deserves attention.

Executive Read

The sector model remains constructive, but leadership is changing.

Financials moved into the No. 1 position, with a composite score of 79.9. Breadth remains exceptionally strong, with the sector’s breadth score at 95.5, while its 12-week return remains +11.5%.

Health Care holds No. 2 at a 73.7 score. Despite a -2.0% week, its underlying relative-strength score remains the strongest in the model at 100.

Energy slipped from the top position to No. 3, but this looks more like consolidation than deterioration. Four-week performance remains +5.3%, breadth is 95.5, and trend remains exceptionally strong.

And then there is Technology.

Information Technology jumped from No. 8 to No. 4, with its composite score surging from 47.2 to 61.4.

That is this week’s biggest positive rotation signal.

At the opposite end, Industrials deteriorated sharply, falling from No. 3 to No. 8.

The market isn’t simply becoming defensive or abandoning growth.

Leadership is rotating again.

The Leaders

🥇 Financials — Rank #1

Score: 79.9 | 1W: +1.1% | 4W: +2.0% | 12W: +11.5%

Financials have become the model’s strongest overall sector.

The important part isn’t the one-week gain. The sector combines:

Relative Strength: 91.7
Breadth: 95.5
Trend: 75.0
12-week return: +11.5%

Financials are classified as a Pullback Leader / Expansion sector, meaning the underlying structure remains strong even after periods of consolidation.

CRI Read: Financials remain one of the healthiest broad leadership groups in the market.

🥈 Health Care — Rank #2

Score: 73.7 | 1W: -2.0% | 4W: +5.3% | 12W: +12.4%

Health Care gave back some ground this week, but the broader signal remains strong.

Its 100 relative-strength score is the highest of any sector.

The sector also retains a Leader / Expansion classification.

That combination matters. A weak week inside an otherwise powerful intermediate trend is very different from structural deterioration.

CRI Read: Still a core leadership sector. The pullback is worth monitoring rather than interpreting as a breakdown.

🥉 Energy — Rank #3

Score: 73.3 | 1W: -1.5% | 4W: +5.3% | 12W: +9.5%

Energy surrendered the No. 1 position, but its internal structure remains unusually strong.

Its trend score is 91.7, breadth is 95.5, and the sector remains classified Leader / Expansion.

Energy’s weakness this week therefore looks much more like cooling following strong performance than an outright rotation away from the group.

CRI Read: The Energy trend remains intact.

The Biggest Change: Technology Is Coming Back

Information Technology — Rank #4

Score: 61.4 | 1W: +1.3% | 4W: +5.9% | 12W: +3.1%

Last week: Rank #8, Score 47.2

This is easily the most interesting move in the model.

Technology gained four ranking positions in one week and its composite score increased by more than 14 points.

More importantly:

Trend Score: 100
Breadth Score: 81.8
4-week return: +5.9%

The sector has moved into the model’s Improving quadrant.

There is still a weakness: volatility remains poor, and Technology has not yet re-established the broader relative-strength profile of Financials, Health Care, or Energy.

But this is exactly the type of rotation worth watching.

CRI Read: Technology has transitioned from lagging to improving. If relative strength follows trend and breadth higher, Technology could challenge the top three again.

For the AI infrastructure theme specifically, this is constructive. The model is no longer showing broad Technology deterioration even as other parts of the market maintain leadership.

Materials Quietly Strengthen

Materials — Rank #5

Score: 60.9 | 4W: +5.5% | 12W: +5.4%

Materials doesn’t have Technology’s dramatic ranking jump, but the internals continue improving.

The sector remains classified Leader / Expansion, and its breadth score improved meaningfully.

That matters for the themes we’ve been following around:

Copper
Electrical steel
Grid infrastructure materials
Transformer inputs
Industrial metals
AI power infrastructure

Materials still isn’t a dominant leadership group, but the model is moving in the right direction.

The Biggest Negative Change: Industrials

Industrials — Rank #8

Score: 44.0 | 1W: -1.7% | 4W: -1.5%

Last week: Rank #3, Score 59.5

This is the deterioration I would pay the most attention to.

Industrials fell five positions in one week.

Relative strength dropped to 33.3, breadth weakened to 40.9, and the sector moved into the Laggard / Late Cycle classification.

That doesn’t mean the industrial investment thesis is broken.

It does mean investors are becoming considerably more selective.

The distinction is particularly important for our AI infrastructure thesis. Individual electrical-equipment, power, cooling, automation and grid companies can continue outperforming even while the broad Industrial sector weakens.

CRI Read: Favor company-level structural growth rather than treating Industrials as a broad sector trade.

The Bottom of the Model

The weakest portion of the ranking now looks very different from the leadership group.

Consumer Discretionary — Rank #9: Score 37.3
Communication Services — Rank #10: Score 36.9
Real Estate — Rank #11: Score 35.5
Utilities — Rank #12: Score 20.2

Utilities are particularly weak.

The sector carries just an 8.3 relative-strength score, 9.1 breadth score, and a negative 3.7% four-week return.

Utilities remain classified Laggard / Distribution.

Real Estate also deteriorated substantially, falling from No. 7 to No. 11.

That is noteworthy given the continuing debate around lower rates eventually benefiting rate-sensitive sectors.

The market isn’t confirming that thesis yet.

What Changed This Week?

The rotation can be summarized fairly simply.

Improving: Technology, Financials, Materials and Communication Services.

Holding leadership: Health Care and Energy.

Deteriorating: Industrials, Real Estate and Utilities.

And that produces an interesting market structure.

This is not classic defensive rotation.

Health Care remains strong, but Utilities and Real Estate are near the bottom.

It isn’t pure cyclical leadership either, because Industrials just weakened dramatically.

Instead, capital appears to be concentrating around sectors with the strongest combination of earnings durability, structural growth and balance-sheet resilience.

That favors Financials, Health Care, Energy and increasingly Technology.

CRI Positioning Read-Through

For the themes we follow, this week’s model reinforces a selective rather than broad-market approach.

Highest conviction: Financials, Health Care and Energy.

Improving rapidly: Technology.

Interesting secondary opportunity: Materials.

Become more selective: Industrials.

Weakest broad setups: Real Estate and Utilities.

Within AI infrastructure specifically, I would continue separating structural beneficiaries from their parent sectors.

A transformer manufacturer, copper producer, electrical-equipment supplier, cooling company or industrial automation business doesn’t automatically become unattractive because its broad sector ranking weakens.

If anything, weakening sector breadth makes stock-level differentiation more important.

CRI Bottom Line

The most important signal this week is leadership rotation rather than market deterioration.

Financials have taken the top position. Health Care and Energy retain strong underlying structures. Materials continue improving.

But Technology’s jump from No. 8 to No. 4 is the development to watch next.

At the same time, Industrials dropping from No. 3 to No. 8 tells us not to assume every economically sensitive or AI-adjacent company will participate equally.

The market appears to be narrowing its standards.

Strong structural growth is still being rewarded. Weakening breadth is being punished.

That is an environment where sector rotation matters—but stock selection matters even more.

The updated Fear & Greed data materially changes this week’s interpretation. The prior version used the August 21 readings; the new August 28 dataset shows a sharp recovery in market internals.

The MTP regime remains RISK-ON, there is still no panic or shock signal, and the four-week analog model has moved to an UP signal. But now Fear & Greed is beginning to confirm that improving outlook rather than diverging sharply from it.

MTP Dashboard

Current Regime: RISK-ON

Stress Warning Score: 22.1
Previous: 18.8

Recovery Warning Score: 65.8
Previous: 65.8

Shock State: NORMAL

Panic Score: 0

Stress increased modestly, but remains well below levels associated with a regime transition.

The important message remains straightforward: the bull regime survived the August deterioration.

The 4-Week Model Turns Bullish

2-Week Outlook

Probability Up: 44%
Probability Down: 20%
Confidence Spread: 24 points

Signal: NEUTRAL

4-Week Outlook

Probability Up: 60%
Probability Down: 24%
Confidence Spread: 36 points

Signal: UP

This remains one of the most important developments in this week’s report.

The model isn’t yet providing strong conviction over the next couple of weeks, but the intermediate outlook has improved enough to cross the MTP’s 55% directional threshold.

Fear & Greed Just Changed Dramatically

This is where the updated data changes the story.

The Equal-Weighted Composite jumped to 50.8, up 9.4 points in one week, moving from Fear back to Neutral.

Even more impressive, the Regime-Weighted Composite surged 22 points to 54.0 and is now also Neutral.

That represents a significant improvement from August 21.

Trend: 52.0 — Neutral

1-week: +35.5
4-week: -13.9

Trend staged an enormous one-week recovery.

Last week’s Extreme Fear reading has essentially been erased.

Volatility: 73.4 — Greed

1-week: +8.5
4-week: +12.0

Volatility conditions remain supportive of risk assets.

Leadership: 27.0 — Fear

1-week: +18.5
4-week: +20.0

Leadership remains the weakest part of the system — but the direction is improving substantially.

Credit: 71.7 — Greed

1-week: -3.8
4-week: +1.9

Credit weakened slightly this week but remains firmly supportive.

The Divergence Is Starting to Close

This is perhaps the most important interpretation change from the earlier report.

Last week we had:

Extreme Fear in Trend + Extreme Fear in Leadership + strong Credit.

That created a major internal divergence.

Now:

Trend has recovered to Neutral.

Leadership has improved sharply but remains in Fear.

Volatility remains in Greed.

Credit remains in Greed.

The system is becoming much more internally consistent with the underlying Risk-On MTP regime.

That strengthens the case that the August deterioration represented a sentiment and participation reset inside the bull regime rather than the beginning of a systemic breakdown.

Leadership Is Now the Key Confirmation Signal

Leadership remains at just 27.0.

That deserves attention because a durable advance eventually needs participation to broaden.

But the weekly improvement of +18.5 points is encouraging.

A continued move through 30 and eventually toward 40–50 would significantly strengthen the bullish interpretation.

This may now be the single most important Fear & Greed component to watch.

Credit Continues to Be the Stabilizer

Credit slipped modestly from the previous week, but at 71.7 remains firmly in Greed.

This remains critical.

If the August weakness had represented the beginning of a broader financial stress event, we would expect credit conditions to deteriorate alongside market leadership.

That still hasn’t happened.

Credit therefore continues telling us that the financial system beneath equities remains relatively healthy.

Treasury Market: Functional, but Not Stress-Free

For the week ending August 28:

Total Treasury Offering: $739 billion

Average Bid-to-Cover: 2.80x

Indirect Bidder Participation: 56.7%

Direct Bidder Participation: 10.1%

Primary Dealer Takedown: 24.2%

Treasury demand remains functional.

The 2.80x bid-to-cover ratio is healthy, but dealer absorption remains elevated above the sovereign monitor’s 20% warning level.

That means the Treasury market is clearing supply, but primary dealers are still carrying a meaningful portion of it.

Sovereign Risk Remains the Bigger Macro Warning

The improvement in Fear & Greed should not obscure what is happening elsewhere.

USD/JPY: ~160.04

Still above the 155 warning threshold.

30-Year Treasury: ~5.19%

Still above the 5% warning threshold.

Primary Dealer Takedown: 24.2%

Still above the 20% monitoring threshold.

These indicators aren’t signaling immediate market failure.

But they remain the areas where stress could ultimately migrate into equities.

The long end of the Treasury curve and the yen therefore remain two of the most important macro variables to watch.

Market Performance Remains Constructive

Over the latest four-week period:

Market ROC: +3.2%

Equities vs. Bonds: +1.7%

Gold vs. Equities: +10.3%

VIX Model: 14.51

This continues to suggest something more nuanced than a conventional flight to safety.

Equities remain constructive while gold is simultaneously outperforming.

Capital appears willing to own both risk assets and hard-asset/sovereign-risk hedges.

That remains consistent with the broader CRI thesis of a capital-intensive AI infrastructure cycle occurring alongside rising sovereign and fiscal pressures.

What Would Strengthen the Bullish Case?

The setup becomes considerably stronger if:

Leadership rises above 40.

Trend remains above 50.

Credit remains above 60.

The 4-week MTP signal remains UP.

The 30-year Treasury moves back below 5%.

USD/JPY falls below 155.

If those begin occurring together, the current setup would move from a selective Risk-On environment toward a substantially healthier broad-market configuration.

What Would Change the Interpretation?

The principal warning signs remain:

Credit falls below 50.

Leadership reverses and falls below 20 again.

Trend collapses after this week’s rebound.

MTP regime moves from Risk-On toward Stress.

Panic Score rises above zero.

30-year yields accelerate materially above 5%.

USD/JPY pushes toward 165–170.

Treasury dealer absorption moves toward or above 30%.

No single signal determines the outcome.

The important signal would be several of these deteriorating together.

CRI Positioning Read-Through

This week’s updated data supports maintaining a constructive posture while remaining selective.

The strongest structural areas continue to include:

AI infrastructure

Power generation and transmission

Data-center electrical infrastructure

Cooling and thermal management

Industrial automation

Defense

Energy

Copper and strategic materials

Selective semiconductor bottlenecks

The improvement in Trend is encouraging, but Leadership at 27 argues against assuming the entire market has suddenly become healthy.

Quality still matters.

CRI Bottom Line

This week’s updated data makes the MTP setup meaningfully more constructive.

The regime remains RISK-ON.

The Panic Score remains zero.

The Shock State remains NORMAL.

The four-week model has moved to an UP signal with 60% probability of an advance versus 24% probability of decline.

And now market internals are beginning to confirm it.

Equal-Weighted Fear & Greed recovered to 50.8 Neutral.

Regime-Weighted Fear & Greed jumped to 54.0 Neutral.

Trend surged from Extreme Fear to 52.0 Neutral.

Leadership improved sharply to 27.0, although it remains the weakest link.

Credit remains strong at 71.7.

That combination strengthens the argument that August represented a fear and participation reset within an intact bull regime rather than the beginning of a larger breakdown.

The major unresolved risk has shifted away from immediate equity stress and toward the macro plumbing underneath the market:

long-duration Treasury yields, USD/JPY and Treasury absorption.

For now, the MTP message is:

Risk-On remains intact. The forward outlook is improving. Market internals are healing.

But leadership still needs to broaden — and sovereign pressure hasn’t gone away.

Executive View

The mid-week regime remains constructive but fragile.

The major indexes are holding up well despite a hotter inflation print, rising policy uncertainty, and the market waiting on Nvidia earnings after the close. That is a positive sign. However, this is not a broad “everything is working” tape. It is a selective rotation environment where leadership is concentrating around AI infrastructure, industrials, power, networking, optics, and cooling, while healthcare, consumer discretionary, communication services, and some commodity-linked areas are showing weaker participation.

The S&P 500 was essentially flat, QQQ was slightly positive, equal-weight S&P 500 outperformed the cap-weighted index, and small caps lagged modestly. That combination tells us the market is not breaking down, but risk appetite is not fully broadening either. SPY was roughly flat, QQQ was up about 0.08%, RSP was up about 0.17%, and IWM was down about 0.11%.

The biggest macro pressure point today was inflation. Reuters reported that U.S. indexes finished nearly flat to slightly lower after a hotter-than-expected inflation report, with July inflation running at 3.7% versus expectations of 3.6%, while Q2 GDP grew at 1.5%. The report also noted that the market is now assigning meaningful odds to another Fed hike at the September meeting.

The biggest market event risk remains Nvidia. Reuters noted that investors are looking to Nvidia earnings for evidence on AI demand, margins, and guidance, while options markets were pricing a roughly $280 billion potential market-cap swing after the report.

CRI read: The market is not in a defensive breakdown, but it is not a clean risk-on expansion either. This is a selective risk-on / event-risk regime where AI infrastructure leadership remains intact, but the broader tape needs confirmation from Nvidia, rates, and breadth.

Regime Health Cards

Market Structure

Status: Stable, but not aggressively bullish
Direction: Neutral to slightly positive
Signal: Indexes are holding near highs despite inflation pressure and Nvidia event risk.

The S&P 500 and Nasdaq were nearly flat, while the Dow was weaker. That is a reasonable outcome given the hotter inflation print and the market waiting for Nvidia earnings. The absence of a larger selloff is constructive, but the lack of stronger upside shows that investors are waiting for confirmation before adding broad risk.

CRI Take: Market structure remains intact. A breakdown is not visible yet, but the next move likely depends on Nvidia earnings, AI guidance, and the market’s interpretation of Fed policy risk.

Breadth and Participation

Status: Mixed
Direction: Slightly improving under the surface
Signal: Equal-weight outperformed cap-weight, but small caps lagged.

RSP outperformed SPY today, which is a modest positive breadth signal. Equal-weight strength suggests that participation was not limited entirely to mega-cap technology. However, IWM lagged, which means the broadest risk appetite is still incomplete.

Small-cap weakness matters because a durable risk-on regime usually needs participation from smaller companies, regional banks, cyclicals, and economically sensitive groups. Today’s tape showed some improvement in equal-weight participation, but not enough to call this a broad expansion.

CRI Take: Breadth is not broken, but it is not strong enough yet. The market is broadening selectively into favored themes rather than lifting all risk assets.

Leadership Concentration

Status: Still concentrated, but second-layer AI is participating
Direction: Improving
Signal: Nvidia was down ahead of earnings, while several AI infrastructure names moved higher.

This was the most important regime signal today. Nvidia fell about 1.5% ahead of earnings, but Marvell, Vertiv, Lumentum, Cisco, Micron, and GE Vernova all moved higher. That means AI infrastructure leadership did not depend entirely on Nvidia today.

That is a constructive read-through for the CRI AI infrastructure thesis. Lumentum rose more than 6%, Vertiv rose more than 3%, GE Vernova rose nearly 3%, Marvell rose about 2%, Cisco gained more than 1%, and Micron was positive. This suggests capital is still rotating into the second-layer beneficiaries: optics, networking, cooling, memory, and power infrastructure.

CRI Take: Leadership remains AI-centered, but today’s action supports the idea that the trade is broadening beneath Nvidia. That is healthier than a one-stock AI tape.

Rates and Inflation

Status: Pressure point
Direction: Deteriorating
Signal: Hotter inflation and higher Treasury yields are the main macro constraint.

The inflation print is the clearest deterioration signal in this week’s regime check. Reuters reported that July inflation was slightly hotter than expected at 3.7%, while AP noted that the 10-year Treasury yield rose to about 4.65%.

TLT was slightly lower, confirming that long-duration assets remain under pressure. Rising yields are especially important for AI and growth leadership because the market is still assigning very high valuation premiums to future AI cash flows.

CRI Take: Inflation and rates are the main macro risk. The market can absorb higher yields for a while if earnings and AI guidance remain strong, but sustained rate pressure would make leadership more fragile.

Credit Conditions

Status: Stable
Direction: Neutral
Signal: High yield and investment-grade credit were nearly flat.

HYG and JNK were essentially unchanged, while LQD was only modestly lower. That does not suggest credit stress is spreading into the equity market.

This is important because equity weakness becomes more dangerous when credit confirms deterioration. Today, credit did not confirm a risk-off move. The market remains selective, but not stressed.

CRI Take: Credit is not flashing warning signals yet. This supports staying engaged with leadership themes while monitoring rates and earnings risk.

Volatility and Event Risk

Status: Elevated event risk
Direction: Elevated into Nvidia
Signal: The market is calm at the index level, but Nvidia earnings can move the AI complex.

Reuters reported that Nvidia options implied a 5.4% post-earnings move, equal to roughly a $280 billion potential market-cap swing. That makes Nvidia the key regime event for the week.

This is the critical difference between index calm and event risk. The S&P 500 and QQQ look stable, but the AI complex could still move sharply depending on Nvidia’s data-center demand, margins, Rubin commentary, HBM availability, and guidance.

CRI Take: Volatility is concentrated around Nvidia rather than broad market stress. The reaction across second-layer AI names will matter more than Nvidia’s move alone.

Sector and Theme Rotation

Technology

Status: Positive
Direction: Improving
Signal: XLK outperformed despite Nvidia weakness.

Technology was one of the stronger sectors today, with XLK up about 0.62%. That is constructive because Nvidia itself was down, meaning the sector had support beyond a single name.

Semiconductors were more mixed. SMH was slightly negative, while SOXX was modestly positive. This suggests the market is not uniformly buying chips before Nvidia earnings, but it is not abandoning the group either.

CRI Take: Technology remains a leadership sector, but the next confirmation needs to come from earnings quality and post-Nvidia breadth.

AI Infrastructure

Status: Strong
Direction: Improving
Signal: Second-layer AI names outperformed Nvidia.

This is the strongest part of today’s regime check. AI infrastructure leadership broadened beneath Nvidia. Lumentum, Vertiv, GE Vernova, Marvell, Cisco, and Micron all finished positive, with particularly strong moves in optics, cooling, power, and networking.

This aligns directly with yesterday’s AI Bottleneck Intelligence conclusion: the market is increasingly rewarding the physical and system-level bottlenecks beneath the AI trade. Memory, optics, networking, cooling, and power are acting better than a simple “Nvidia only” market would imply.

CRI Take: The AI infrastructure thesis remains intact. The strongest confirmation is that leadership broadened even while Nvidia weakened into earnings.

Industrials

Status: Strong
Direction: Improving
Signal: XLI outperformed and power infrastructure names led.

Industrials were one of the strongest sector signals today, with XLI up about 1.09%. GE Vernova also gained nearly 3%, reinforcing the power infrastructure and electrification theme.

This is important because industrial leadership broadening alongside AI infrastructure is a healthy sign. It shows capital is not only buying software or mega-cap technology. It is also buying physical infrastructure tied to power, grid equipment, energy systems, and industrial capacity.

CRI Take: Industrials remain one of the best bridges between the AI trade and the real-economy infrastructure cycle.

Energy

Status: Positive
Direction: Stable to improving
Signal: XLE was positive despite oil-related macro noise.

Energy gained modestly, with XLE up about 0.59%. Reuters also noted that oil was moving lower earlier in the day as hopes around the Strait of Hormuz eased some geopolitical pressure.

The market’s willingness to keep energy positive while oil cooled suggests the sector may still have support from capital discipline, geopolitical risk premium, and AI-related power demand themes.

CRI Take: Energy remains useful as part of the real-assets and power-demand basket, but it is not the primary leadership engine today.

Utilities

Status: Positive
Direction: Stable
Signal: XLU gained modestly.

Utilities were positive, with XLU up about 0.42%. That can be read two ways. Part of it may be defensive demand, but in the current regime utilities also carry AI power-grid relevance.

The important distinction is whether utility strength comes from defensive rotation or from power infrastructure demand. Today’s simultaneous strength in industrials, GE Vernova, Vertiv, and utilities suggests the move is not purely defensive. It still fits the AI power theme.

CRI Take: Utilities remain a mixed but investable signal. The best interpretation is power-demand confirmation, not broad risk-off positioning.

Financials and Regional Banks

Status: Mixed
Direction: Neutral
Signal: XLF slipped, but KRE was positive.

Financials were slightly negative, while regional banks gained modestly. XLF was down about 0.12%, while KRE was up about 0.33%.

This is not a major leadership signal, but it is also not a warning sign. Regional bank participation helps offset concern that higher yields or credit stress are beginning to damage the broader market.

CRI Take: Financials are not leading, but regional bank stability supports the “no credit stress yet” regime read.

Healthcare

Status: Weak
Direction: Deteriorating
Signal: XLV was the weakest major sector proxy in today’s snapshot.

Healthcare underperformed, with XLV down about 0.98%. Reuters also noted healthcare weakness as a market drag.

This is a deterioration signal for defensive quality leadership. Healthcare weakness is not necessarily bearish for the market as a whole, but it does show that capital is not hiding broadly in traditional defensives.

CRI Take: Healthcare is not acting as a leadership group. It remains a laggard and does not currently support a defensive rotation thesis.

Consumer

Status: Weak
Direction: Deteriorating
Signal: Consumer discretionary and staples both lagged.

XLY was down about 0.68%, while XLP was down about 0.28%.

Consumer weakness is important because inflation pressure and slower spending can both weigh on consumer-linked sectors. AP reported that consumer spending slowed while inflation remained above the Fed’s target.

CRI Take: Consumer sectors are not confirming broad economic strength. This supports a more selective allocation posture rather than a broad cyclical risk-on call.

Materials and Commodities

Status: Mixed to weak
Direction: Softening
Signal: Materials were slightly positive, but copper miners and gold/silver weakened.

XLB was slightly positive, but FCX and COPX were lower. GLD and SLV also declined, while the dollar firmed modestly.

This is a soft deterioration signal for commodity leadership. The longer-term copper, grid, and electrification thesis remains intact, but today’s action does not show strong near-term accumulation in metals.

CRI Take: Materials are not leading this week. Copper and precious metals need better confirmation before increasing conviction near term.

Leadership Changes

Leadership is broadening inside AI infrastructure

The most constructive development is that AI infrastructure leadership is not limited to Nvidia. Optics, networking, cooling, power infrastructure, and memory are all participating. This supports the CRI view that the next phase of the AI trade is moving into the second-layer beneficiaries.

Industrials are confirming the physical infrastructure cycle

Industrials outperformed, and GE Vernova’s strength supports the power and electrification theme. This is important because it connects market leadership to real-world constraints: grid, turbines, power equipment, cooling, and data-center buildout.

Equal-weight participation is slightly better

RSP outperforming SPY is a positive sign. It does not fully solve the breadth problem because small caps lagged, but it shows the market is not entirely dependent on the largest cap-weighted names.

Traditional defensives are mixed

Utilities were positive, but healthcare and staples were weak. That does not look like classic defensive rotation. It looks more like selective buying of power-related utilities and infrastructure-adjacent assets.

Deterioration Signals

Inflation pressure is back in focus

The hotter inflation print raises the risk that the Fed remains tighter for longer or even considers another hike. That is the biggest macro threat to long-duration growth and high-multiple AI infrastructure names.

Nvidia event risk is unusually important

The market is calm on the surface, but Nvidia earnings can reset the entire AI complex. The key risk is not just whether Nvidia beats. The key risk is whether guidance, margins, HBM costs, financing commentary, or Rubin timing disappoints.

Small caps are not confirming broad risk-on

IWM lagged while QQQ and RSP were positive. That means investors are still selective. A healthier regime would show stronger participation from small caps, regional banks, and broader cyclicals.

Consumer sectors are weak

Consumer discretionary and staples both lagged. That is consistent with inflation pressure and slower consumer spending.

Metals and commodity leadership softened

Copper miners, gold, and silver weakened. This does not break the long-term commodity thesis, but it argues against near-term leadership in that bucket today.

Capital Rotation Read

Capital is moving toward:

AI infrastructure second-layer beneficiaries
Networking, optics, memory, cooling, and power infrastructure remain the strongest confirmation areas.

Industrials and power infrastructure
XLI strength and GEV participation confirm continued demand for physical infrastructure themes.

Selective technology
XLK strength suggests tech remains favored, but semiconductors are mixed ahead of Nvidia.

Power-linked utilities
Utility strength appears more tied to AI power demand and grid relevance than traditional defensiveness.

Capital is moving away from:

Healthcare
Healthcare remains a lagging sector and is not attracting defensive sponsorship.

Consumer discretionary and staples
Consumer weakness suggests inflation and spending concerns remain relevant.

Precious metals and copper miners
The dollar firmed, gold and silver sold off, and copper miners weakened.

Mega-cap AI concentration as the only trade
Nvidia weakness alongside strength in second-layer AI names suggests capital is not abandoning AI but is rotating under the surface.

CRI Theme Implications

AI infrastructure remains high conviction

Today’s market action supports the AI infrastructure supercycle thesis. The strongest confirmation came from the exact areas we have been tracking: optics, networking, cooling, memory, and power.

Power and electrification remain leadership themes

GE Vernova, Vertiv, industrials, and utilities all support the view that power remains one of the most important bottlenecks in the AI buildout.

Memory and bandwidth remain critical

Micron, Marvell, and Lumentum support the view that memory scarcity and data movement are still attracting capital.

Copper and materials need patience

The long-term electrification and grid buildout thesis remains intact, but today’s action in FCX and COPX was soft. This bucket needs renewed accumulation before becoming a near-term leadership call.

Defensives are not leading broadly

Healthcare and staples weakness argues against a classic risk-off rotation. Utilities are the exception, but they are increasingly tied to the power theme.

Small caps remain a watch item

Small-cap participation is still not strong enough to confirm a broad risk-on regime. This remains one of the key regime checkpoints.

1–4 Week Outlook

The base case remains selective risk-on with elevated event risk.

The market has enough strength to stay constructive, but not enough breadth to declare a broad acceleration phase. The next 1–4 weeks will depend heavily on whether Nvidia confirms the AI infrastructure thesis and whether the bond market accepts the inflation data without a major yield spike.

Bullish confirmation would look like:

Nvidia reports strong data-center demand, durable margins, credible Rubin timing, and manageable HBM cost pressure.

Second-layer AI names participate after Nvidia earnings instead of fading.

RSP continues to outperform SPY, and IWM begins to improve.

Credit remains stable, with HYG and JNK holding firm.

Industrials, power infrastructure, networking, optics, and cooling continue to attract capital.

Bearish deterioration would look like:

Nvidia beats but the stock sells off and second-layer AI names fail to participate.

Yields continue rising after the inflation print and pressure growth multiples.

Small caps and regional banks weaken.

Healthcare, staples, and utilities lead for purely defensive reasons.

Credit spreads begin to widen or high yield breaks lower.

Bottom Line

The mid-week market regime is not broken, but it is increasingly selective.

The strongest signal is that capital continues to rotate into the second layer of AI infrastructure: optics, networking, cooling, memory, power systems, and industrial infrastructure.

The biggest warning signal is that inflation and rates are again becoming a constraint, while Nvidia earnings carry unusually high event risk for the broader AI complex.

For CRI positioning, the message is clear:

Stay focused on leadership, but do not assume broad market participation. The market is rewarding companies tied to real AI infrastructure constraints, while punishing weaker consumer, healthcare, and commodity-linked groups. The next confirmation test is whether Nvidia earnings broaden the rally or expose concentration risk.

Executive View

This week’s AI infrastructure signal remains constructive, but the bottleneck map is becoming more selective. The first phase of the AI trade was dominated by GPU availability, hyperscaler capex, and Nvidia leadership. The next phase is increasingly defined by system-level constraints: HBM pricing, AI server inflation, networking density, optical bandwidth, liquid cooling, power reliability, grid stability, and the durability of debt-funded infrastructure spending.

The market is still rewarding the AI buildout. Nvidia traded higher ahead of earnings, while Cisco, Marvell, Lumentum, Micron, and Vertiv showed continued investor interest across the broader AI infrastructure stack. Nvidia closed around $213.05, up 2.19%, while Marvell rose 4.85%, Lumentum rose 6.66%, Micron rose 2.43%, Cisco rose 0.80%, and Vertiv rose modestly.

The key takeaway is that the AI trade is not weakening broadly. It is moving into a more layered phase. The best signals are coming from companies tied to memory scarcity, high-speed networking, optics, rack-scale integration, liquid cooling, power buffering, and power-ready data-center capacity.

At the same time, the deterioration signals are becoming more important. AI server costs are rising, memory supply remains tight, power systems are under stress, off-grid reliability is becoming a real execution risk, and investors are starting to question how much of the buildout depends on corporate debt, vendor financing, and aggressive future monetization assumptions.

CRI Bottleneck Heat Map

1. GPUs / Accelerators

Health: Strong demand, tight expectations
Direction: Stable to positive
CRI Read: Nvidia remains the anchor layer of the AI infrastructure trade. Demand is still strong, but investor attention is shifting toward the Rubin transition, customer financing, HBM availability, and whether hyperscaler demand can remain this aggressive.

Nvidia remains the center of the AI compute stack, and this week’s setup is dominated by expectations around its upcoming earnings and the transition toward the next major platform cycle. The market still treats compute as the anchor of the AI trade, but the bar is high. Strong demand alone may not be enough. Investors will want confidence that the next platform transition supports continued growth, that customer financing is not masking softer demand, and that margins remain durable as memory costs rise.

Investment implication: Compute remains healthy, but no longer early. The upside case depends on Nvidia proving that the next platform cycle can sustain demand and pricing power without increasing concern around financing quality.

2. Custom Silicon

Health: Improving
Direction: Positive
CRI Read: Hyperscalers and large global cloud players continue pushing proprietary chips to protect margins, improve system economics, and reduce reliance on merchant accelerators.

Rising system costs make custom silicon more attractive. When GPUs, HBM, power, and cooling all become more expensive, hyperscalers have a stronger incentive to optimize internally. Alibaba is a useful signal here. Its recent results showed heavy AI infrastructure investment pressure, but also strong AI-related cloud growth. AP reported that Alibaba’s profit dropped sharply as AI infrastructure investment increased, while AI-related services revenue grew strongly.

Alibaba’s capital raise also reinforces the scale of the global AI infrastructure race. FT reported that proceeds from Alibaba’s large equity placement are intended for full-stack AI capabilities, including AI infrastructure.

Investment implication: Custom silicon remains a positive long-term theme. Rising merchant accelerator and HBM costs strengthen the case for internal chip development, especially among hyperscalers, neoclouds, and sovereign AI platforms.

3. HBM / Memory

Health: Constrained
Direction: Worsening
CRI Read: Memory inflation is becoming one of the clearest system-cost bottlenecks for AI racks.

The most important pressure point this week is memory cost inflation. Reports indicate that some Nvidia AI server customers have been notified of price increases above 15%, with rising DRAM and HBM costs cited as a major driver. Tom’s Hardware reported that conventional DRAM contract prices were projected to rise sharply as suppliers reallocated capacity toward HBM and server products.

This matters because HBM is not a secondary input anymore. It is now one of the defining cost drivers of AI rack economics. If HBM prices rise materially, hyperscalers have three choices: absorb higher costs, push more aggressively into custom silicon and architectural efficiency, or slow and reshape deployment plans.

Micron’s strong market action reinforces that investors continue to view memory scarcity as beneficial for memory suppliers. MU traded higher on the day, confirming that capital still sees memory as one of the most important bottleneck layers.

Investment implication: HBM remains a high-conviction bottleneck. This is constructive for memory suppliers, but it raises cost and margin pressure for AI server buyers, cloud customers, and downstream infrastructure platforms.

4. Networking

Health: Strong
Direction: Positive
CRI Read: Networking remains one of the clearest second-layer AI infrastructure beneficiaries.

Cisco’s expanded Nvidia partnership was one of the strongest confirmation signals this week. Cisco announced that it is expanding its Secure AI Factory with Nvidia through a partnership with Supermicro, adding rack-scale AI computing solutions and high-density liquid- and air-cooled infrastructure for enterprises, neoclouds, and sovereign cloud customers.

This confirms a core CRI thesis: AI infrastructure winners are not limited to GPU suppliers. The system needs validated networking, secure architecture, rack-scale integration, switching, fabric, thermal design, and deployment support.

Marvell’s strong move also supports this theme. MRVL rose nearly 5%, suggesting continued capital rotation into connectivity, custom silicon, and AI networking beneficiaries.

Investment implication: Networking remains one of the healthiest second-layer AI infrastructure themes. The key watch item is whether Ethernet-based and merchant networking solutions continue to gain share as AI clusters scale.

5. Optics / Photonics

Health: Improving
Direction: Positive
CRI Read: Optical bandwidth remains a structural beneficiary of larger AI clusters, higher data movement, and the transition toward faster interconnect cycles.

Optics remain one of the cleanest read-throughs from AI cluster growth. Larger clusters require more data movement, higher bandwidth, lower latency, and better power efficiency per bit. Lumentum’s strong move, up more than 6%, supports the view that investors continue accumulating optical and photonics beneficiaries.

The market is increasingly recognizing that AI scaling is not only about compute density. It is also about moving data efficiently between accelerators, racks, clusters, and data centers.

Investment implication: Optics and photonics remain positive. This layer benefits from 800G and 1.6T transitions, AI cluster scale, and the need to reduce power per bit as data movement explodes.

6. Power Infrastructure

Health: Constrained
Direction: Worsening
CRI Read: Power remains the most important physical bottleneck in the AI infrastructure buildout.

Power constraints are intensifying, and the issue is no longer only total megawatts. It is also power quality, ramp behavior, system resilience, and grid-level buffering. GE Vernova introduced a medium-voltage UPS product designed to provide stable, continuous, high-quality power for data centers, AI factories, and other energy-intensive facilities. GE Vernova also highlighted its broader data-center offering across turbines, substations, transformers, switchgear, energy management, and power-conversion technologies.

Investor’s Business Daily reported that GE Vernova’s MV-UPS is designed to help protect turbines and grid networks from drastic AI-related load fluctuations.

This is an important development because it confirms that AI is creating grid-level infrastructure problems, not just data-center-level problems. UPS systems, transformers, switchgear, turbines, power electronics, and grid stabilization are becoming core AI infrastructure categories.

GE Vernova traded lower despite the product news, suggesting the power trade may be structurally strong but increasingly valuation-sensitive.

Investment implication: Power remains the highest-conviction physical bottleneck theme. The most attractive opportunities may shift toward companies solving power reliability, grid buffering, medium-voltage distribution, power electronics, and interconnection complexity.

7. Cooling / Thermal

Health: Strong demand
Direction: Positive
CRI Read: Liquid and hybrid cooling are moving from specialized deployments toward standard AI system architecture.

Cisco’s expanded Secure AI Factory offering includes high-density liquid- and air-cooled systems through Supermicro, reinforcing that cooling is now embedded into rack-scale AI architecture rather than treated as a facility afterthought.

Network World also highlighted Cisco’s addition of Supermicro liquid-cooled servers to its AI infrastructure portfolio, noting that liquid-cooling options are an important component of the partnership.

Vertiv’s stock remained positive, but the move was modest compared with some other AI infrastructure names. That does not weaken the cooling thesis. It suggests that the market is becoming more selective on valuation and positioning.

Investment implication: Cooling remains structurally positive. The risk is not demand. The risk is valuation and whether investor expectations already discount several years of strong AI infrastructure growth.

8. Data Centers / Real Estate

Health: Demand strong, execution constrained
Direction: Mixed
CRI Read: Data-center demand remains strong, but power-ready capacity is becoming the scarce asset.

The data-center buildout is still expanding, but the market is increasingly separating announced capacity from usable capacity. The most valuable capacity is not just land, shells, or square footage. It is capacity with power access, grid interconnection, cooling, permitting, equipment availability, and a realistic path to energization.

Cisco’s rack-scale AI infrastructure push reinforces that the data-center market is entering an execution phase. It is no longer enough to announce capacity. Customers need validated systems, networking, cooling, security, and deployment reliability.

Investment implication: Data-center demand remains strong, but investors should favor power-ready capacity over headline capacity. Capacity without credible power strategy deserves a discount.

9. Advanced Packaging / Substrates

Health: Tight but investable
Direction: Stable
CRI Read: Advanced packaging remains critical to next-generation accelerators, HBM integration, and system-level AI performance.

Advanced packaging remains a foundational constraint even when it is not the loudest weekly headline. The AI system depends on high-bandwidth memory integration, dense interconnects, substrate availability, and the ability to scale increasingly complex packages.

This layer is stable rather than worsening this week, but it remains essential. Any disruption in packaging capacity, substrate supply, or HBM integration would ripple directly into accelerator availability and system delivery timelines.

Investment implication: Advanced packaging remains a durable AI infrastructure theme. It is not the most visible weekly bottleneck, but it remains one of the most important structural layers.

10. Financing / Capex Returns

Health: Watch item
Direction: Deteriorating
CRI Read: Debt-funded AI infrastructure and vendor-financing scrutiny are rising.

The financing layer is becoming more important. As AI infrastructure spending scales, investors are starting to ask whether customer revenue, utilization, and productivity gains can justify the capital intensity.

Alibaba is a useful example. Its AI infrastructure spending weighed heavily on profitability, while its cloud and AI-related services continued to grow. AP reported a 75% profit drop as AI investment increased, while AI-related services revenue rose 45%.

MarketWatch also reported that Alibaba’s share sale proceeds were designated for AI chips, infrastructure, and AI models, reinforcing the scale of funding required to remain competitive.

This does not mean the AI trade is broken. It means the next phase will be judged more heavily on returns. Investors will increasingly ask whether AI workloads are generating enough revenue to support the capex, whether vendors are financing demand that would otherwise be slower, and whether debt markets remain willing to fund the buildout at acceptable terms.

Investment implication: Financing is moving from a background issue to a core AI infrastructure bottleneck. The market will increasingly favor companies with visible backlog, real utilization, strong pricing power, and credible customer ROI.

Leadership Signals

Networking and rack-scale systems are gaining confirmation

Cisco’s expanded Nvidia and Supermicro architecture confirms that the AI infrastructure trade is broadening into secure, validated, rack-scale deployment platforms. This supports the second-layer AI thesis.

Optics and photonics continue to attract capital

Lumentum’s strong move suggests investors still want exposure to bandwidth and optical interconnect beneficiaries.

Memory scarcity remains investable

Micron’s positive price action and continued HBM/DRAM inflation signals confirm memory as a core bottleneck layer.

Power reliability is becoming a new category

GE Vernova’s MV-UPS launch shows that grid stabilization, turbine protection, and power-quality management are becoming investable AI infrastructure themes.

Deterioration Signals

Memory inflation is no longer theoretical

Reported AI server price increases tied to DRAM and HBM costs show that memory scarcity is now directly affecting system costs. This is positive for memory suppliers but negative for downstream capex efficiency.

Power execution risk is rising

AI workload volatility creates challenges for grids, turbines, UPS systems, and off-grid power solutions. The power bottleneck is broadening from supply availability to power quality and reliability.

Financing scrutiny is increasing

Large AI infrastructure investments are putting pressure on free cash flow, equity issuance, debt issuance, and future return expectations. The market will increasingly question whether AI capex can generate sufficient revenue.

Valuation risk is building in popular infrastructure names

Power, cooling, optics, and networking remain structurally attractive, but some names are now heavily sponsored. Strong themes can still experience sharp pullbacks if positioning becomes crowded.

CRI Investment Theme Implications

AI second-layer infrastructure remains the key opportunity

The strongest confirmation this week came from networking, optics, cooling, memory, power reliability, and rack-scale integration. This supports the CRI view that the next phase of the AI trade should broaden beyond GPUs alone.

HBM and memory scarcity remain high conviction

Memory is becoming both a profit pool and a system-cost problem. That makes it one of the most important layers to track over the next several quarters.

Power infrastructure remains the most important physical bottleneck

The AI power theme is expanding from generation and grid access into switchgear, transformers, UPS, grid stabilization, power conversion, turbines, and power-quality management.

Custom silicon gains strategic importance

As AI infrastructure becomes more expensive, hyperscalers have more incentive to develop internal chips and optimize system economics.

Financing quality matters more now

The AI buildout remains investable, but the market will increasingly reward companies that can prove utilization, backlog, pricing power, and customer ROI.

Positioning Read

The AI infrastructure trade is not ending. It is becoming more layered.

The best risk/reward is likely shifting away from only owning the most obvious AI winners and toward companies solving the next set of constraints:

HBM and memory supply
Memory remains one of the most important constraints and one of the clearest pricing-power areas.

High-speed networking
AI clusters need more switching, Ethernet fabric, system validation, and secure deployment architecture.

Optical interconnects
Bandwidth demand continues to rise as clusters scale and data movement becomes more expensive.

Liquid cooling and thermal management
Higher rack density makes cooling a required system feature rather than a facilities upgrade.

Power electronics and grid stabilization
AI is creating power-quality and load-fluctuation problems that require new infrastructure solutions.

Power-ready data-center capacity
The market should reward capacity that is actually energized, permitted, cooled, and deployable.

Custom silicon economics
Rising system costs make internal silicon programs more strategically important.

The biggest near-term watch is Nvidia earnings and guidance. A strong report can re-accelerate the AI complex, but the quality of the reaction will matter. If only Nvidia rallies, leadership is narrowing. If networking, optics, memory, power, cooling, and data-center infrastructure participate, the second-layer AI infrastructure thesis remains strong.

Watch List for the Next 1–4 Weeks

Nvidia earnings and next-platform commentary

Watch data-center demand, Rubin timing, gross margin, HBM availability, customer concentration, and financing commentary.

HBM pricing and allocation

Further memory inflation would support memory suppliers but pressure AI server economics and hyperscaler capex efficiency.

Networking and optics order commentary

Look for backlog strength, hyperscaler design wins, 800G and 1.6T ramps, Ethernet fabric demand, and rack-scale validation.

Power bottleneck updates

Track turbine lead times, transformer constraints, switchgear availability, grid interconnection delays, UPS adoption, and off-grid reliability.

Cooling adoption

Watch whether liquid cooling moves further from specialized deployments into standard AI rack-scale systems.

AI financing conditions

Any widening in AI-related credit spreads or more aggressive equity/debt issuance would raise the financing-risk signal.

Capex ROI commentary

The next phase of the AI trade requires stronger evidence that infrastructure spending is generating measurable revenue, productivity, or utilization gains.

Bottom Line

This week’s AI Bottleneck Intelligence remains bullish but more selective.

The strongest signal is that capital continues broadening into the second layer of AI infrastructure: memory, networking, optics, cooling, rack-scale systems, and power infrastructure.

The biggest warning signal is that the buildout is becoming more expensive, more power-constrained, and more dependent on financing assumptions.

For CRI, the conclusion is sharpened:

The AI infrastructure supercycle remains intact, but the bottlenecks are moving from chips alone to the physical, electrical, thermal, networking, memory, and financing layers required to make AI scale work.

Executive Read

This week’s Microcap Discovery screen delivered the strongest expansion in confirmed leadership we’ve seen in several weeks.

The engine screened 322 eligible companies, producing:

• 84 Promotion Watchlist names

• 69 Accumulation Watch candidates

• 169 Early Monitor names

• 0 liquidity rejects

But the headline is not simply that Promotion increased.

It is where that expansion occurred.

Healthcare/Biotech has moved decisively to the front of the microcap leadership cycle:

• 98 companies screened

• 57 Promotion leaders

• 16 Accumulation candidates

• 67.4 average MicroCap score

• 17.9 average Capital Flow score

• Positive sector-relative strength over both 4 and 13 weeks

Meanwhile, AI Software/Data remains a developing pipeline with 8 Promotion names and 14 Accumulation candidates.

And perhaps most interestingly, AI Infrastructure/Semiconductors currently has zero Promotion names.

That does not invalidate the long-term AI infrastructure thesis.

It tells us something more useful:

At the microcap level, capital leadership has rotated.

The Leadership Map Changed

Last week’s report showed three distinct layers:

AI Infrastructure/Semis — concentrated quality

Healthcare/Biotech — broad leadership

AI Software/Data — rapidly developing breadth

One week later, the hierarchy is considerably clearer.

Current Leadership

Healthcare/Biotech

Capital flow, relative strength and breadth have aligned.

Developing Pipeline

AI Software/Data, Materials/Commodities and selected Financials

These groups contain meaningful Accumulation populations beneath existing Promotion leaders.

Cooling / Rebuilding

AI Infrastructure/Semiconductors and AI Optical/Networking

These remain strategically important themes, but this week’s microcap model does not show confirmed Promotion leadership.

That distinction is exactly why we run the screen every week.

The structural investment story and the current capital-flow story are not always the same thing.

Active Leaders: Where Capital Is Already Moving

This week’s Promotion list is unusually large, so simply displaying the highest scores would miss an important distinction.

Some leaders have improving businesses beneath the technical move.

Others remain catalyst-driven or highly speculative.

Here are the names I believe best illustrate the current leadership environment.

PSNL — Personalis

MicroCap Score: 94

• 1W RS vs SPY: +32.3%

• 4W RS vs SPY: +52.4%

• 13W RS vs SPY: +114.0%

• Revenue Growth YoY: +30.0%

• Capital Flow: Very Strong

• Cash: $212.7M

• Debt: $38.0M

Personalis sits at the intersection of precision oncology, genomic testing and data-intensive healthcare.

Unlike a binary pre-revenue biotechnology story, PSNL has a commercial revenue base. Revenue is growing approximately 30%, while the company maintains a substantial net-cash position.

The weakness remains cash generation, with both operating and free cash flow still negative.

Long-Term Growth Multiple Path

PSNL needs to translate expanding oncology adoption into operating leverage.

If revenue growth persists while losses narrow, the company could transition from a speculative diagnostics story toward a scalable precision-medicine platform.

This is one of the more interesting combinations of model strength and identifiable commercial growth in the current Promotion group.

OABI — OmniAb

MicroCap Score: 94

• 1W RS vs SPY: +29.5%

• 4W RS vs SPY: +109.7%

• 13W RS vs SPY: +60.4%

• Revenue Growth YoY: +244.2%

• Gross Margin: 98.3%

• Capital Flow: Very Strong

OABI has become one of the most striking signals in the screen.

Rather than developing a single drug, OmniAb provides antibody-discovery technologies used by pharmaceutical and biotechnology partners.

That platform model potentially creates economics very different from conventional clinical biotech.

Free cash flow is near breakeven despite continued operating investment, while the reported gross-margin profile is exceptionally high.

Long-Term Growth Multiple Path

The key is partner productivity.

More successful partner programs can create additional milestone and royalty economics without requiring OmniAb to fund every downstream development program itself.

If that model scales, OABI has the potential to become more than a catalyst trade.

VREX — Varex Imaging

MicroCap Score: 93

• 4W RS vs SPY: +65.9%

• 13W RS vs SPY: +80.9%

• Revenue Growth YoY: +3.7%

• Gross Margin: 34.3%

• Operating Cash Flow: +$11.3M

• Capital Flow: Very Strong

VREX is useful because it demonstrates that Healthcare leadership is not limited to drug development.

The company supplies imaging components and technology into medical and industrial applications.

Growth is currently modest and debt is significant, so this is not a classic high-growth compounder.

Long-Term Growth Multiple Path

A sustained re-rating requires better revenue growth, margin improvement and stronger free-cash-flow conversion.

For now, the technical leadership is substantially stronger than the underlying growth profile.

CRON — Cronos Group

MicroCap Score: 90

• 4W RS vs SPY: +14.2%

• 13W RS vs SPY: +17.8%

• Revenue Growth YoY: +58.4%

• Operating Cash Flow: +$60.1M

• Free Cash Flow: +$6.1M

• Cash: $797.0M

• Debt: $1.3M

• Capital Flow: Very Strong

CRON remains one of the more fundamentally unusual companies in the Promotion group.

The company combines rapid reported revenue growth with positive operating cash flow and an exceptionally strong net-cash position.

The cannabis industry carries its own regulatory and competitive risks, but the balance sheet substantially changes the risk profile.

Long-Term Growth Multiple Path

CRON needs to prove that recent growth can persist and ultimately generate attractive returns on its enormous liquidity base.

If operating performance continues improving, the balance sheet gives management considerable strategic flexibility.

IRWD — Ironwood Pharmaceuticals

MicroCap Score: 89

• Revenue Growth YoY: +32.6%

• Gross Margin: 77.2%

• Operating Cash Flow: +$185.6M

• Free Cash Flow: +$121.8M

• Capital Flow: Very Strong

IRWD stands apart from much of the biotech-heavy Promotion list because it is already producing meaningful cash.

The balance sheet carries considerable debt, but the underlying business provides actual cash-flow support beneath the model signal.

Long-Term Growth Multiple Path

The opportunity is less about explosive speculative upside and more about durable commercial execution, cash generation and balance-sheet improvement.

This is the type of company where fundamental quality can potentially sustain technical leadership longer than a purely catalyst-driven move.

WEAV — Weave Communications

MicroCap Score: 89

• 1W RS vs SPY: +34.9%

• 13W RS vs SPY: +24.7%

• Revenue Growth YoY: +15.5%

• Gross Margin: 72.4%

• Operating Cash Flow: +$16.9M

• Free Cash Flow: +$39.4M

• Capital Flow: Very Strong

WEAV is classified within Healthcare because of its vertical exposure, but economically it looks much more like a software platform.

That makes it particularly interesting.

The combination of double-digit growth, 70%+ gross margins and positive free cash flow provides a fundamentally cleaner setup than many of the highest-scoring names.

Long-Term Growth Multiple Path

The key is maintaining growth while increasing operating leverage.

If Weave can expand its healthcare communications platform without sacrificing margins, this could become a more traditional small-cap software compounding story.

The Catalyst Leaders

Several other top-scoring companies deserve attention—but they belong in a different risk category.

ENTX — Score 92
13W RS vs SPY +157.5%. Strong capital flow, but development-stage economics.

ACHV — Score 92
Very strong structure and substantial liquidity, but commercialization and regulatory outcomes remain central.

FBRX — Score 91
13W RS vs SPY +253.3%. Extraordinary momentum, but still a catalyst-sensitive biotechnology story.

EDIT — Score 89
Revenue growth is strong off a small/volatile base, while operating cash burn remains substantial.

KURA — Score 89
Strong liquidity and pipeline optionality, but significant ongoing cash consumption.

AQST — Score 89
Revenue +38.1% with a 67% gross margin, but cash flow remains negative.

These can generate enormous moves.

But a 90 MicroCap score does not make a pre-commercial biotech economically equivalent to a profitable operating company.

That remains one of the most important principles of this research.

Emerging Leaders: Where Capital May Be Positioning Next

The Accumulation list contains 69 companies.

Again, I don’t want to simply take the highest ten scores.

The objective is to find earlier-stage candidates where improving capital flow intersects with a business capable of creating longer-duration value.

CRMD — CorMedix

MicroCap Score: 68

• Revenue Growth YoY: +156.5%

• Gross Margin: 90.2%

• Operating Cash Flow: +$253.9M

• Free Cash Flow: +$178.1M

• Cash: $256.7M

• Capital Flow: Very Strong

CRMD immediately stands out.

Unlike most biotechnology Accumulation candidates, this is not simply a promise of future economics.

The current financial profile shows rapid revenue growth, very high gross margins and substantial positive cash generation.

The technical structure remains mixed, which is why it has not graduated into Promotion.

What Needs to Happen

If the operating trajectory persists and price structure confirms, CRMD could become one of the more fundamentally supported Promotion candidates in a future screen.

This is exactly the kind of Accumulation-to-Promotion transition worth monitoring.

APPS — Digital Turbine

MicroCap Score: 68

• 4W RS vs SPY: +25.0%

• 13W RS vs SPY: +135.9%

• Revenue Growth YoY: +26.8%

• Operating Cash Flow: +$50.9M

• Free Cash Flow: +$28.9M

• Capital Flow: Strong

APPS is particularly interesting because it sits inside the AI Software/Data pipeline while already producing cash.

The 13-week relative-strength move is substantial, but the stock has not yet achieved the model characteristics required for Promotion.

What Needs to Happen

Continued revenue improvement and sustained cash generation would provide fundamental confirmation.

If capital flow strengthens further, APPS could become an important graduation candidate.

ANGX — Angel Studios

MicroCap Score: 67

• 13W RS vs SPY: +48.8%

• Revenue Growth YoY: +27.5%

• Gross Margin: 57.9%

• Free Cash Flow: +$8.1M

• Capital Flow: Strong

ANGX is an unusual addition to the software/data group and provides useful diversification away from AI infrastructure.

The company is growing revenue and has produced positive free cash flow, although operating cash flow remains negative.

What Needs to Happen

Consistent cash conversion and sustained audience/platform growth would improve the quality of the thesis considerably.

ALOY — REalloys

MicroCap Score: 69

• 4W RS vs SPY: +74.3%

• Revenue Growth YoY: +82.7%

• Capital Flow: Very Strong

• Theme: Materials / Commodities

ALOY represents something we’ve been watching for several weeks: the possibility of a deeper materials rotation.

Revenue growth is strong and capital flow is very strong, although the company remains cash-flow negative.

What Needs to Happen

Growth needs to convert into durable cash economics.

More importantly, we need to see additional Materials names graduate alongside it before calling this a true sector leadership shift.

ITRG — Integra Resources

MicroCap Score: 69

• 4W RS vs SPY: +28.3%

• 13W RS vs SPY: +15.3%

• Revenue Growth YoY: +15.9%

• Operating Cash Flow: +$76.8M

• Capital Flow: Very Strong

• Structure: Strong Trend + Accumulation

ITRG provides another indication that Materials/Commodities deserves attention.

The operating cash-flow profile is much stronger than many speculative mining names, although free cash flow remains negative as capital is deployed.

What Needs to Happen

Additional production growth and improving free cash flow would strengthen the long-term case.

UMAC — Unusual Machines

MicroCap Score: 69

• 4W RS vs SPY: +38.1%

• 13W RS vs SPY: +60.5%

• Revenue Growth YoY: +687.3%

• Cash: $316.4M

• Capital Flow: Strong

UMAC was last week’s highest-scoring Promotion leader.

It has now fallen back into Accumulation.

That is exactly why weekly lifecycle tracking matters.

The long-term drone and domestic supply-chain opportunity has not disappeared—but the model is telling us that current leadership confirmation weakened.

What Needs to Happen

We want to see renewed capital-flow confirmation rather than assume last week’s signal remains valid indefinitely.

PRCH — Porch Group

MicroCap Score: 68

• 4W RS vs SPY: +40.7%

• 13W RS vs SPY: +75.8%

• Revenue Growth YoY: +11.7%

• Gross Margin: 72.1%

• Operating Cash Flow: +$96.3M

PRCH continues showing improving relative strength alongside a real revenue and cash-generating business.

Balance-sheet complexity remains the principal fundamental concern.

What Needs to Happen

Better free-cash-flow conversion and financial simplification could provide the next leg of fundamental validation.

Theme Leadership: The Biggest Change This Week

Healthcare/Biotech — Dominant Leadership

Average MicroCap Score: 67.4

• 57 Promotion

• 16 Accumulation

• Average Capital Flow Score: 17.9

• Positive 4W and 13W sector-relative strength

This is no longer merely emerging breadth.

Healthcare/Biotech is now unequivocally the dominant microcap leadership group.

The challenge is separating scalable operating businesses from clinical speculation.

That is where fundamental validation becomes essential.

Materials/Commodities — Quietly Improving

Average Score: 51.7

• 4 Promotion

• 9 Accumulation

Materials now has more than twice as many Accumulation candidates as Promotion leaders.

This is the type of ratio that can precede a broader rotation.

ALoy, ITRG and several other names suggest capital is probing the group.

Not leadership yet—but increasingly worth watching.

Financials — Another Developing Pipeline

Average Score: 52.7

• 2 Promotion

• 6 Accumulation

The 13-week sector-relative strength remains positive even though shorter-term relative strength has softened.

That combination could represent consolidation beneath an improving longer-term trend.

AI Software/Data — Pipeline Remains Intact

Average Score: 46.1

• 8 Promotion

• 14 Accumulation

The theme has cooled from last week’s stronger expansion, but the pipeline remains substantial.

APPS is one of the more fundamentally interesting names within that Accumulation group.

The important question is whether these 14 candidates begin graduating into Promotion over the next several weeks.

Energy/Power — Selective Improvement

Average Score: 47.3

• 5 Promotion

• 3 Accumulation

Energy has moved somewhat further through the lifecycle than Industrials, but participation remains selective rather than broad.

AI Infrastructure/Semiconductors — Leadership Reset

Average Score: 45.6

• 0 Promotion

• 2 Accumulation

This is probably the most important negative signal in this week’s screen.

It does not mean AI infrastructure is structurally impaired.

It means that within this specific microcap universe, the capital-flow leadership that was present last week has reset.

That is information—not a reason to force the old thesis onto new data.

AI Optical/Networking — No Current Confirmation

Only four companies remain in the theme, with:

• 0 Promotion

• 0 Accumulation

Again, this does not change the long-duration importance of optical networking.

It simply means:

the current microcap model is not confirming leadership there today.

What Changed This Week?

This is the section I want to make permanent in Microcap Discovery.

Promotion exploded higher

Last week: 54

This week: 84

That is a 56% increase in confirmed Promotion names in one week.

Healthcare became dominant

Healthcare Promotion increased from 27 to 57.

The group now accounts for roughly two-thirds of the entire Promotion Watchlist.

UMAC moved backward

Last week’s #1 Promotion name dropped into Accumulation.

That is not failure—it is lifecycle information.

AI Infrastructure cooled sharply

Last week it held the highest average theme score.

This week it has zero Promotion leaders.

Materials is building beneath the surface

Nine Accumulation candidates versus four Promotion names creates an increasingly interesting pipeline.

CRI Market Read

The broadening in Promotion is constructive, but there is an important nuance.

A 56% weekly increase in Promotion sounds broadly risk-on.

Yet most of that expansion is concentrated in one sector.

That means:

breadth within Healthcare has exploded, but breadth across the entire microcap market remains more selective.

This is not indiscriminate microcap speculation.

It is sector-concentrated risk appetite.

That distinction will matter if Healthcare begins losing relative strength.

The healthiest next development would be for:

• Materials

• AI Software/Data

• Financials

• Energy

to begin graduating more of their Accumulation pipelines.

That would turn today’s sector-specific expansion into genuine market breadth.

Fundamental Quality Matters More Than Ever

This week’s 84-name Promotion list makes the fundamental layer increasingly important.

A model score tells us:

where capital is moving.

It does not automatically tell us:

why the company deserves to compound for five or ten years.

For that, we still need:

• sustainable revenue growth

• gross-margin quality

• operating leverage

• free cash flow

• balance-sheet strength

• manageable dilution

• a growing addressable market

• durable competitive positioning

That is why names such as PSNL, OABI, IRWD, WEAV and CRMD deserve a different level of attention than companies whose entire thesis rests on a single clinical event.

What I’m Watching Next

Healthcare breadth: Can 57 Promotion names persist, or is the group becoming crowded?

CRMD: Does fundamentally strong Accumulation graduate into Promotion?

Materials: Do the nine Accumulation candidates begin moving up the lifecycle?

AI Software/Data: Can the 14-name pipeline rebuild Promotion leadership?

AI Infrastructure: Does capital return after this week’s reset?

UMAC: Does the former #1 leader regain Promotion status—or continue fading?

Those transitions will tell us considerably more than simply comparing scores.

Bottom Line

This week’s Microcap Discovery screen delivered a major leadership shift.

Healthcare/Biotech has broken out as the dominant microcap theme.

But beneath that headline, several potentially more important developments are forming:

Materials is accumulating.

AI Software/Data retains a deep pipeline.

Financials are quietly improving.

AI Infrastructure has entered a leadership reset.

And individual companies are moving both forward and backward through the lifecycle.

That is precisely what this framework is designed to capture.

The objective isn’t to predict which stock will jump tomorrow.

It is to identify the smaller companies where:

capital flow → structural confirmation → fundamental growth → operating leverage

can eventually converge.

Because the most interesting microcap opportunities are rarely obvious when the compounding story begins.

Microcap investing involves substantial volatility, liquidity, dilution and company-specific risk. This report is intended as a research and discovery framework, not individualized investment advice.

The broad sector rotation model tells us where money is moving.

The subsector model tells us what investors are actually buying and selling inside those sectors.

And this week, that distinction matters.

Energy may be the #1 sector, but the strength is concentrated in exploration and production while midstream weakened.

Health Care is surging, but the winners are pharmaceuticals and life-science tools—not every medical company.

Industrials are cooling, yet agriculture equipment and railroads are rallying.

Technology is weak, but not every part of the AI infrastructure chain is breaking equally.

The market is becoming increasingly selective.

The winning strategy right now is not simply choosing the right sector. It is choosing the right part of the right sector.

AI & Semiconductor Infrastructure

This was a difficult week across the AI complex.

There were no positive weekly returns among the primary AI names in the model.

But there was a meaningful difference between relative winners and outright losers.

Holding Up Best

TSM: −1.7%
4-week: +3.9%

Microsoft: −2.3%
4-week: +26.8%

ASML: −4.4%
4-week: +0.5%
12-week: +9.5%

Nvidia: −4.6%
4-week: +3.8%

These names were weak, but they held up better than the semiconductor complex overall.

Microsoft is particularly interesting.

Despite this week’s decline, it remains up nearly 27% over four weeks, suggesting the market is distinguishing between hyperscale AI demand and semiconductor beta.

TSM also continues to show relative resilience.

Under the Most Pressure

AMD: −8.0%
4-week: −9.3%
12-week: −8.3%

Broadcom: −6.2%
4-week: −3.5%
12-week: −17.4%

SOXX: −5.5%
4-week: −1.3%
12-week: −8.6%

Arista Networks: −5.1%

But ANET deserves an important qualification:

12-week return: +18.3%

That means Arista experienced a sharp weekly correction while maintaining one of the strongest intermediate-term trends in the group.

CRI Read-Through

The AI trade is not behaving like one unified theme anymore.

The market is separating:

Hyperscale platforms

from

Semiconductor beta

from

Networking

from

equipment and manufacturing infrastructure.

The broad semiconductor group is clearly under pressure.

But I would not interpret the current move as evidence that AI capital spending is collapsing.

Instead, this increasingly looks like valuation compression and rotation inside the AI ecosystem.

The names worth watching most closely on stabilization are those where intermediate-term strength survived the selloff.

ANET is a good example.

Communication Services

Communication Services produced almost the opposite setup.

Traditional telecom held up well while one of the largest digital-platform companies broke sharply.

Winners

Verizon: +2.0%
4-week: +6.6%

Netflix: +1.8%
4-week: +13.6%

AT&T: +1.6%
4-week: +4.8%

Disney: +0.9%
4-week: +13.6%

There are really two different trades occurring here.

Telecom is attracting defensive capital.

Meanwhile, Netflix and Disney are showing improving entertainment/media momentum.

Biggest Loser

Meta: −6.8%

4-week: −7.6%

12-week: −13.0%

Meta is clearly the weak point inside Communication Services.

Alphabet was comparatively stable at approximately −0.3% for the week, while Meta suffered a much larger decline.

CRI Read-Through

This is another example of why the sector ETF alone can be misleading.

Communication Services as a whole is weak in our sector model.

But underneath it:

Telecom is acting defensively.

Selected entertainment names are strengthening.

Meta is deteriorating.

That is not one trade.

It is three different capital flows occurring inside the same sector.

Consumer Complex

The Consumer model produced one of the widest dispersions of any group this week.

Biggest Winner

Tesla: +6.0%

4-week: +15.9%

But there is an important longer-term qualification:

12-week: −16.7%

Tesla is therefore showing a strong rebound rather than confirmed intermediate-term leadership.

Staples Strength

Coca-Cola: +3.9%

4-week: +10.8%

12-week: +16.0%

PepsiCo: +1.9%

4-week: +5.0%

Coca-Cola is one of the clearest defensive winners in the consumer complex.

Its performance across one, four and twelve weeks is strong.

That suggests this is more than a one-week defensive reaction.

Retail Damage

The real weakness appeared in retail.

Walmart: −10.0%

4-week: −5.3%

12-week: −10.4%

TJX: −7.6%

4-week: −8.6%

12-week: −8.9%

Those are significant declines.

Amazon was considerably more stable:

Amazon: −1.5%

4-week: +11.4%

That divergence is notable.

CRI Read-Through

The consumer is not sending one unified recession signal.

Instead, investors appear to be favoring:

Defensive branded staples

and selectively:

large digital platforms

while punishing portions of traditional retail.

Walmart’s decline is particularly important because the company is normally considered one of the most defensive retailers in the market.

I would watch whether this weakness remains company-specific or spreads into other mass-market retailers.

Energy

Energy’s #1 sector ranking is being driven by a very specific group.

Exploration and production is winning.

Major Winners

EOG Resources: +7.3%

4-week: +4.5%

12-week: +15.6%

ConocoPhillips: +7.1%

4-week: +12.9%

12-week: +19.1%

Devon Energy: +7.1%

4-week: +9.0%

12-week: +11.1%

Occidental Petroleum: +5.0%

4-week: +7.0%

12-week: +8.7%

This is broad E&P strength.

It is one of the cleanest subsector signals in the entire report.

Midstream Weakness

Meanwhile:

Williams: −6.3%

Kinder Morgan: −5.6%

AMLP: −0.4%

That is a major divergence.

Energy is leading—but pipelines and midstream are not leading with it.

CRI Read-Through

The market is currently favoring direct commodity-price leverage.

Exploration and production companies benefit more immediately when oil and natural-gas economics improve.

Midstream tends to behave more like a yield-oriented infrastructure asset.

That means the current Energy trade is not merely investors looking for dividends.

It increasingly looks like a commodity and upstream cash-flow trade.

This reinforces our preference for selected producers and energy businesses with direct operating leverage.

Financials

Financials remain one of the strongest sectors in the rotation model, but this week was weak underneath the surface.

Clear Winner

Charles Schwab: +1.1%

4-week: +10.5%

12-week: +28.9%

Schwab is one of the strongest individual trends in the entire subsector dataset.

Even as most financial stocks fell, SCHW remained positive.

Relatively Stable

Goldman Sachs: essentially flat

GS held up substantially better than most large banks.

Major Bank Weakness

Wells Fargo: −5.6%

Citigroup: −5.5%

Bank of America: −4.3%

JPMorgan: −3.1%

This is a noteworthy reversal.

However, the intermediate-term picture remains much stronger:

BAC is still up approximately 20% over 12 weeks.

JPM is up approximately 18%.

XLF itself remains up approximately 11.8%.

CRI Read-Through

This currently looks more like profit-taking inside a strong Financial sector than a structural financial-system warning.

Schwab’s relative strength is particularly notable.

I would become considerably more concerned if the banking weakness were accompanied by deterioration in our broader Credit model.

That has not happened.

Credit remains one of the strongest components in the MTP framework.

Health Care

Health Care produced some of the strongest individual stock performance in the entire market.

And the winners tell an interesting story.

Pharmaceutical Leadership

Merck: +12.3%

4-week: +16.4%

12-week: +29.4%

Merck was the strongest major Health Care name this week.

Life-Science Tools Surge

Danaher: +8.1%

4-week: +14.3%

12-week: +20.1%

Thermo Fisher: +7.0%

4-week: +10.7%

12-week: +27.9%

This is particularly interesting.

Life-science tools are not simply defensive Health Care exposure.

They are tied to pharmaceutical research, diagnostics and biotechnology spending.

The simultaneous strength in DHR and TMO looks much more like subsector accumulation than random stock movement.

Additional Strength

Eli Lilly: +6.4%

Regeneron: +3.9%

Regeneron is now up approximately 35.8% over 12 weeks.

Weak Areas

Intuitive Surgical: −4.0%

UnitedHealth: −2.9%

Stryker: −2.9%

CRI Read-Through

Health Care leadership is broadening, but it is not centered on traditional defensive managed care.

The strongest money is moving toward:

Pharmaceuticals

Biotechnology-related therapeutics

and especially:

life-science tools

That makes Health Care’s rise considerably more interesting than a simple market-defense trade.

Industrials

Industrials were weak at the sector level, but the internal picture was remarkably divided.

Agriculture Equipment Wins

Deere: +6.3%

4-week: +3.1%

12-week: +19.7%

DE is showing genuine intermediate-term leadership.

Railroads Rebound

Union Pacific: +4.9%

12-week: +17.3%

Norfolk Southern: +4.9%

12-week: +15.5%

The simultaneous strength in both major railroads is significant.

This looks like a subsector move rather than an isolated stock event.

Major Losers

Honeywell: −7.7%

4-week: −10.9%

12-week: −13.2%

Eaton: −7.2%

RTX: −5.9%

GE: −5.4%

Eaton’s decline is especially worth watching because it sits directly inside the electrification and data-center power theme.

However, ETN remains positive over four and twelve weeks, so this is not yet a structural breakdown.

CRI Read-Through

Industrials are undergoing a major internal rotation.

Money moved toward:

Agriculture machinery

and

Railroads

while selling hit:

electrification equipment

aerospace

and

diversified industrials.

The weakness in Eaton deserves monitoring because it intersects directly with one of our strongest long-duration investment themes.

One bad week doesn’t invalidate that thesis.

But it tells us the market is taking profits in some of the most crowded infrastructure winners.

Materials

Important: Materials data in the submitted workbook currently runs through August 14, one week behind the other subsector models.

I would therefore treat these readings as directional rather than this week’s synchronized signal.

Winners Through August 14

Dow: +5.7%

Newmont: +4.3%

4-week Newmont return: +31.3%

Air Products: +1.6%

Newmont’s strength is easily the most important signal.

Gold-mining exposure was being aggressively accumulated.

Weakness

Freeport-McMoRan: −4.2% for that week

But:

4-week return: +14.3%

That suggests a pullback inside an otherwise strong copper move.

Sherwin-Williams: −3.3%

Ecolab: −3.1%

Linde: −1.7%

CRI Read-Through

Even with the stale observation, Materials continue to show the same split we have been monitoring:

Precious metals strong

Copper structurally constructive but volatile

Chemical/industrial-material names mixed

I would like to see the next August 21 observation before making a stronger call here.

Real Estate

Real Estate remains mixed and generally lacks decisive leadership.

Best Performer

Ventas: +1.7%

12-week: +10.9%

Ventas is the strongest weekly name and continues to show good intermediate-term performance.

That points toward relative strength in health-care real estate.

Stable Areas

Prologis: +0.5%

American Tower: +0.1%

Weak Areas

Equity Residential: −3.5%

Equinix: −2.9%

Boston Properties: −1.1%

Public Storage: −1.1%

CRI Read-Through

Health-care REITs are currently holding up better than several other Real Estate subsectors.

Data-center REIT exposure through Equinix was weak this week despite the longer-term AI infrastructure story.

Again, that illustrates a recurring theme:

A powerful secular demand story does not guarantee short-term equity leadership when valuations and rates matter.

Utilities

Utilities are currently the cleanest weak group in the report.

There were no meaningful winners among the major utility names.

Least Weak

Consolidated Edison: −1.5%

Broader Weakness

NextEra: −2.9%

Dominion: −3.2%

Constellation Energy: −3.3%

Xcel Energy: −3.6%

AEP: −3.7%

Sempra: −4.1%

PSEG: −4.5%

This is broad weakness rather than a single-company problem.

CRI Read-Through

This confirms what we saw in the sector model.

The electricity-demand thesis remains strong.

But that is not the same thing as saying regulated Utility equities must outperform.

Higher yields, capital requirements and financing costs remain major headwinds.

Even Constellation—one of the most direct beneficiaries of data-center and nuclear-power demand—fell more than 3% this week.

The market currently appears to prefer power infrastructure suppliers and selected generation exposure over broad regulated Utilities.

This Week’s Strongest Internal Themes

Stepping away from individual sectors, several clear subsector trends emerge.

1. Exploration & Production

EOG
COP
DVN
OXY

This is currently one of the strongest coordinated moves in the market.

2. Pharmaceuticals & Life-Science Tools

MRK
DHR
TMO
LLY
REGN

Health Care leadership is becoming broader and higher quality.

3. Agriculture Equipment & Railroads

DE
UNP
NSC

This is a notable physical-economy rotation inside Industrials.

4. Defensive Consumer Brands

KO
PEP

Coca-Cola in particular has developed strong intermediate momentum.

5. Telecom

VZ
T

Traditional telecom is attracting defensive capital even while the broader Communication sector remains weak.

Where Capital Is Leaving

Several groups stand out on the opposite side.

Broad Semiconductors

AMD
AVGO
SOXX

The AI infrastructure thesis remains intact, but semiconductor beta is clearly being de-risked.

Traditional Retail

WMT
TJX

This is one of the more surprising areas of weakness.

Large Banks

WFC
C
BAC
JPM

Not yet systemic—but worth watching.

Electrification / High-Multiple Industrials

ETN
GE
HON

The structural themes remain intact, but investors are taking profits.

Regulated Utilities

The weakness is broad rather than isolated.

CRI Subsector Leadership Board

For the coming week, these are the groups I would watch most closely.

Strongest Current Momentum

Exploration & Production

Pharmaceuticals

Life-Science Tools

Agriculture Equipment

Railroads

Selected defensive Consumer Staples

Strong Intermediate Trends Experiencing Pullbacks

AI networking

Selected semiconductor infrastructure

Large banks

Electrification equipment

Copper

These could become attractive if the underlying trend stabilizes rather than continuing to deteriorate.

Weakest Areas

Broad semiconductor beta

Traditional retail

Regulated Utilities

Selected diversified Industrials

Meta / portions of digital Communication

Bottom Line

The subsector data tells a much richer story than the broad sector rankings.

Energy is #1—but E&P is winning while midstream is losing.

Health Care is strong—but pharmaceuticals and life-science tools are driving the advance.

Industrials are weak—but Deere and the railroads are rallying while electrification and aerospace names are being sold.

Communication Services is weak—but telecom and selected entertainment companies are holding up while Meta falls sharply.

Technology is under pressure—but the AI complex is showing very different degrees of weakness rather than one uniform collapse.

And Utilities remain broadly weak despite one of the strongest long-term electricity-demand environments we have seen.

That is the central message this week:

Sector selection is no longer enough. Dispersion inside sectors is becoming one of the dominant features of this market.

The best opportunities increasingly sit at the subsector and individual-company level.

For now, the clearest capital flows are toward:

Exploration & Production

Pharmaceuticals

Life-Science Tools

Agriculture Equipment

Railroads

while capital is moving away from:

Broad semiconductor beta

Traditional retail

Regulated Utilities

and selected high-multiple industrial winners.

Follow the capital one level deeper.

The sector model delivered a significant rotation this week.

The biggest development is the collapse in Technology’s relative-strength ranking, while Energy and Health Care strengthened sharply. Financials remain near the top, but Energy has now joined them with the highest model score.

This is consistent with the broader market signal we saw in this week’s MTP report: the market remains risk-on, but leadership underneath the indexes is changing rapidly.

The rotation model is increasingly favoring Energy, Financials, Health Care and Materials, while Technology, Communication Services and Utilities have moved to the bottom of the ranking.

Weekly Sector Ranking

1. Energy — XLE

Sector Score: 77.4
Model Action: ACCUMULATE
Rotation: Leader
Structural Phase: Expansion

1-Week Return: +2.79%
4-Week Return: +6.74%
12-Week Return: +13.87%

12-Week Relative Performance vs. S&P 500: +12.39%

Constituents Above 20-Week Average: 100%

Previous Rank: 5
Current Rank: 1

Energy produced one of the strongest signals in the entire model.

Relative strength jumped substantially, Trend reached 100, and breadth remains exceptionally strong. Every measured constituent is above its 20-week trend.

The most important part of the Energy signal is that this is no longer simply a short-term commodity trade.

Energy is now classified as both:

Leader

and

Expansion

That is the combination we want to see when identifying sustainable sector leadership.

CRI Take

Energy moves into the highest-conviction group this week.

The sector also fits directly into several structural themes we continue to follow: increasing electricity demand, AI-related power requirements, constrained hydrocarbon investment, geopolitical energy security and rising infrastructure spending.

Energy has moved from a secondary rotation opportunity into a core leadership sector.

2. Financials — XLF

Sector Score: 77.4
Model Action: ACCUMULATE
Rotation: Leader
Structural Phase: Expansion

1-Week Return: −1.17%
4-Week Return: +2.08%
12-Week Return: +11.83%

12-Week Relative Performance vs. S&P 500: +10.35%

Constituents Above 20-Week Average: 100%

Previous Rank: 1
Current Rank: 2

Financials lost the top ranking to Energy, but there is very little difference between the two model scores.

Both score 77.4.

More importantly, Financials still have 100% breadth above the 20-week trend.

The one-week decline therefore looks more like consolidation than structural deterioration.

Relative strength remains excellent, and the sector continues to outperform the S&P 500 significantly over the intermediate term.

CRI Take

Financials remain constructive.

Higher long-term rates can benefit portions of the financial sector, particularly when credit conditions remain healthy.

That qualification is important.

Our MTP Credit Score is currently extremely strong, which makes the present environment much more favorable for Financials than an environment where yields are rising because credit stress is accelerating.

Financials remain a primary leadership sector, but I would continue focusing on companies showing actual earnings and revenue momentum rather than treating the entire sector equally.

3. Health Care — XLV

Sector Score: 71.9
Model Action: ACCUMULATE
Rotation: Leader
Structural Phase: Expansion

1-Week Return: +4.33%

4-Week Return: +7.41%

12-Week Return: +17.34%

12-Week Relative Performance vs. S&P 500: +15.86%

Constituents Above 20-Week Average: 66.7%

Previous Rank: 4
Current Rank: 3

Health Care may be the most interesting sector in this week’s report.

It delivered the strongest weekly return among the major sectors at +4.33% and now has the strongest 12-week performance at +17.34%.

Relative performance versus the S&P 500 is an exceptional +15.86 percentage points.

Relative-strength scoring has reached 100.

Trend also strengthened materially.

This is increasingly looking like genuine institutional accumulation rather than simple defensive rotation.

CRI Take

Health Care moves into the top opportunity tier.

The sector is especially interesting because it can participate in two different environments.

If the market stabilizes, Health Care has momentum.

If broader market volatility increases, its defensive characteristics can attract additional capital.

The major limitation is breadth. At roughly 67% above the 20-week trend, participation is good but not nearly as comprehensive as Energy or Financials.

This means stock selection remains important.

The Leadership Tier

The model now identifies three clear sectors for ACCUMULATE status:

Energy — 77.4

Financials — 77.4

Health Care — 71.9

All three are:

Leaders

Expansion phase

Strong intermediate-term outperformers

This is one of the clearest leadership clusters the model has produced recently.

But notice what is missing:

Technology.

That is an important change.

4. Industrials — XLI

Sector Score: 59.5
Model Action: HOLD
Rotation: Pullback Leader
Structural Phase: Late Cycle

1-Week Return: −3.36%

4-Week Return: −1.32%

12-Week Return: +4.37%

12-Week Relative Performance: +2.89%

Breadth Above 20-Week Average: 71.4%

Previous Rank: 2
Current Rank: 4

Industrials fell from second to fourth.

The model still considers the sector a Pullback Leader, which is considerably more constructive than a Laggard designation.

Breadth remains relatively healthy at 71%.

The problem is momentum.

One-week performance was −3.36%, and four-week performance has now turned negative.

CRI Take

I would not abandon Industrials.

The secular investment cycle involving automation, reshoring, grid infrastructure, defense, power equipment and data-center construction remains powerful.

But the model is telling us that the sector has entered a cooling period.

This is a HOLD rather than an ACCUMULATE signal.

Individual companies tied directly to the strongest capital-spending themes can still outperform substantially even if the sector ETF pauses.

5. Materials — XLB

Sector Score: 57.3
Model Action: HOLD
Rotation: Leader
Structural Phase: Expansion

1-Week Return: +1.90%

4-Week Return: +4.45%

12-Week Return: +5.06%

12-Week Relative Performance: +3.58%

Breadth Above 20-Week Average: 50%

Previous Rank: 7
Current Rank: 5

Materials quietly improved again.

The sector jumped two ranking positions and remains classified as:

Leader

and

Expansion.

Relative strength improved, and short-term momentum is positive.

The weakness is breadth.

Only about half of the underlying group is above the 20-week trend.

CRI Take

Materials remain a selective opportunity rather than a broad-sector bet.

Copper and other strategic-material exposures remain particularly attractive within the longer-term CRI framework.

The sector would become considerably more interesting if breadth begins moving toward 70–80%.

For now, I prefer specific commodity and infrastructure beneficiaries over blanket Materials exposure.

6. Consumer Staples — XLP

Sector Score: 49.5
Model Action: HOLD
Rotation: Leader
Structural Phase: Late Cycle

1-Week Return: −0.12%

4-Week Return: +2.21%

12-Week Return: +4.43%

12-Week Relative Performance: +2.95%

Breadth Above 20-Week Average: 33.3%

Consumer Staples continue to rise through the ranking.

The sector has improved from near the bottom of the model to sixth.

That normally indicates increasing defensive positioning.

However, breadth remains weak.

Only one-third of the measured constituents are above their 20-week trend.

CRI Take

Staples are showing selective defensive accumulation, but this is not yet a broad leadership signal.

The move is still worth noting because it supports the message from the MTP Leadership score:

Investors are becoming more defensive.

7. Real Estate — XLRE

Sector Score: 49.2
Model Action: HOLD
Rotation: Pullback Leader
Structural Phase: Late Cycle

1-Week Return: −0.42%

4-Week Return: −1.89%

12-Week Return: +3.38%

12-Week Relative Performance: +1.89%

Breadth Above 20-Week Average: 60%

Real Estate remains in the middle of the model.

The sector has reasonable breadth but weak short-term momentum.

With long-term Treasury yields remaining elevated, Real Estate continues to face a significant competing-yield problem.

CRI Take

This remains a HOLD.

There may be individual opportunities, particularly where cash flow growth is strong, but the macro environment does not currently support aggressive broad REIT exposure.

8. Information Technology — XLK

Sector Score: 47.2
Model Action: HOLD
Rotation: Laggard
Structural Phase: Late Cycle

1-Week Return: −3.53%

4-Week Return: +4.22%

12-Week Return: −3.92%

12-Week Relative Performance: −5.40%

Breadth Above 20-Week Average: 87.5%

Previous Rank: 3
Current Rank: 8

This is the most important negative change in this week’s sector report.

Technology fell from third to eighth.

The sector’s overall score dropped from roughly 69.7 to 47.2 in a single week.

Relative-strength scoring collapsed to 8.3.

Yet Trend remains very strong at 91.7, and 87.5% of constituents remain above their 20-week moving averages.

That creates an unusual setup.

Technology has not structurally broken.

Instead, capital is rotating out of Technology faster than the underlying long-term price trends have deteriorated.

CRI Take

This distinction is extremely important for the AI thesis.

The model is not saying the AI infrastructure cycle is over.

It is saying Technology is currently losing the capital-allocation competition to other sectors.

That means this is not the environment to indiscriminately chase every AI-related stock.

The focus should remain on companies where earnings, revenue and AI infrastructure demand are actually accelerating.

Within Technology, I would continue emphasizing the bottlenecks:

Semiconductor equipment

Memory

Optical networking

Networking infrastructure

Data-center power

Cooling

The broad XLK signal is weakening, but the strongest AI capital-spending niches can still diverge sharply from the sector ETF.

9. Consumer Discretionary — XLY

Sector Score: 35.8
Model Action: WATCH
Rotation: Laggard
Structural Phase: Expansion

1-Week Return: −0.15%

4-Week Return: +7.87%

12-Week Return: −2.16%

Relative 12-Week Performance: −3.65%

This is an interesting contradiction.

Consumer Discretionary remains a Laggard, but its structural phase is still classified as Expansion.

The sector also has excellent four-week momentum at +7.87%.

CRI Take

This is a WATCH rather than something to dismiss.

If relative strength begins improving, Consumer Discretionary could become an emerging rotation candidate.

But the intermediate-term relative-performance numbers are not strong enough yet.

10. Communication Services — XLC

Sector Score: 27.1
Model Action: WATCH
Rotation: Laggard
Structural Phase: Distribution

1-Week Return: −1.37%

4-Week Return: +4.80%

12-Week Return: −3.46%

12-Week Relative Performance: −4.94%

Communication Services remains near the bottom.

The combination of Laggard + Distribution is particularly weak.

Short-term rebounds should be treated cautiously until relative strength improves.

CRI Take

WATCH only.

This sector does not currently offer enough strength to compete with Energy, Financials or Health Care for new capital.

11. Utilities — XLU

Sector Score: 25.2
Model Action: WATCH
Rotation: Laggard
Structural Phase: Distribution

1-Week Return: −3.48%

4-Week Return: −7.60%

12-Week Return: −3.10%

Breadth Above 20-Week Average: 0%

Utilities remain the weakest sector in the model.

The most concerning statistic is breadth:

0% of measured constituents are above their 20-week trend.

That is extremely weak.

CRI Take

This is an important reminder that power demand and Utilities are not the same investment thesis.

The AI power thesis remains exceptionally strong.

But regulated utility equities can be hurt by rising long-term interest rates, financing requirements and bond-market competition even while electricity demand rises.

The better opportunity may continue to exist upstream in:

Generation equipment

Grid infrastructure

Transformers

Gas turbines

Electrical equipment

Data-center power systems

rather than traditional regulated Utilities.

The Rotation Map

ACCUMULATE

Energy

Financials

Health Care

These are currently the strongest combinations of relative strength, trend, breadth and structural phase.

HOLD

Industrials

Materials

Consumer Staples

Real Estate

Information Technology

There are attractive opportunities inside these sectors, but the model does not currently support aggressive broad-sector accumulation.

WATCH

Consumer Discretionary

Communication Services

Utilities

These sectors need additional confirmation before capital should be increased.

Biggest Weekly Changes

Energy: 5 → 1

This is the strongest positive rotation.

Energy moved decisively into leadership with excellent trend, breadth and relative strength.

Health Care: 4 → 3

The ranking change appears modest, but the underlying performance is extremely strong.

Health Care has become a genuine leadership sector.

Technology: 3 → 8

This is the biggest deterioration.

The long-term trend remains intact, but relative strength has collapsed.

That tells us capital is moving elsewhere.

Industrials: 2 → 4

Still constructive, but momentum has cooled enough to move the sector out of the highest-conviction tier.

Materials: 7 → 5

Gradually improving.

Breadth is the primary missing confirmation.

What This Says About the Market

The sector model is giving us an important confirmation of this week’s MTP report.

Capital is rotating toward:

Energy

Financials

Health Care

and selectively toward:

Materials

Staples

Meanwhile:

Technology

Communication Services

and Utilities

are losing relative leadership.

This is not the sector configuration I would expect from a simple broad risk-on technology rally.

It looks increasingly like a late-cycle rotation combined with a physical-economy capital-spending cycle.

That distinction matters.

The market is rewarding sectors tied to:

Energy

Cash flow

Financial intermediation

Health Care

Commodities

Industrial capacity

while becoming more selective with high-duration growth.

CRI Sector Ranking

For the coming week, I would group the sectors this way:

Highest Conviction

1. Energy

2. Health Care

3. Financials

All three receive ACCUMULATE signals.

Constructive / Selective

4. Industrials

5. Materials

Both remain structurally attractive but need better short-term confirmation.

Neutral

6. Consumer Staples

7. Real Estate

Useful as diversification and defensive exposure, but neither is displaying strong enough broad momentum for aggressive accumulation.

Selective Technology

8. Information Technology

The broad sector has weakened sharply.

But this remains the sector where subsector selection matters most.

The AI infrastructure cycle remains intact even though XLK has dropped in the rotation ranking.

Weak / Watch

9. Consumer Discretionary

10. Communication Services

11. Utilities

These currently offer the weakest model setups.

Bottom Line

The sector model underwent a meaningful change this week.

Energy moved into first place.

Financials remain exceptionally strong.

Health Care is accelerating.

Materials are improving.

At the same time:

Technology fell from third to eighth.

That does not mean the AI trade is finished.

It means capital is broadening into areas of the market that increasingly benefit from physical scarcity, higher nominal growth, energy demand and real-economy capital investment.

The combination of our MTP and sector models now paints a fairly consistent picture:

The market remains risk-on, but leadership is rotating away from broad Technology and toward Energy, Financials and Health Care.

The key question for the next several weeks is whether Technology stabilizes and rejoins the leadership group, creating a healthier broad advance, or whether its relative weakness spreads into the broader market.

For now, the strongest message from the sector model is straightforward:

Follow the capital—not the narrative.

Right now, the capital is increasingly moving toward Energy, Financials and Health Care.

The Market Turning Point model remains in a formal RISK-ON regime, but this week’s data produced one of the more significant internal divergences we have seen recently.

The headline model has not broken.

The shock state remains NORMAL.

The panic score remains 0.

Credit conditions are actually strengthening.

But underneath those stabilizing signals, Trend and Leadership have collapsed into Extreme Fear, the equity market declined approximately 2% during the week, and volatility increased.

This creates a very different setup from a traditional broad risk-off event.

The market is showing substantial internal stress without confirmation from credit or the MTP shock engine.

That makes the next several weeks particularly important.

MTP Dashboard

MTP Regime: RISK-ON
Regime Change: None

Stress Warning Score: 22.1
Previous week: 8.8
Direction: Rising

Recovery Warning Score: 65.8
Previous week: 73.8
Direction: Falling, but still elevated

Shock State: NORMAL
Panic Score: 0

Recommended Model Action: HOLD

Model Allocation:
Market exposure: 80%
Defensive exposure: 20%

The important development is not an outright regime change.

It is the movement of the underlying stress and recovery engines toward each other.

The Stress Warning Score increased from 8.8 to 22.1, while the Recovery Warning Score declined from 73.8 to 65.8.

Neither move is individually threatening yet. Together, however, they indicate that the extremely supportive conditions seen earlier in August have weakened.

The Weekly Market Signal Deteriorated

The five-day market inputs shifted noticeably:

Equity market: −2.02%

VIX: +9.43%

Long-duration bonds: −0.17%

U.S. Dollar Index: −1.06%

Normally, a roughly 2% equity decline accompanied by rising volatility deserves attention.

But two important things did not happen.

There was no bond-market shock.

And there was no dollar surge.

That matters because the MTP shock engine looks for confirmation across markets rather than reacting to equity weakness alone.

The result:

No major shock flags were triggered.

This continues to look more like an equity-market correction and internal risk reassessment than the beginning of a systemic liquidation.

At least for now.

Fear & Greed Has Collapsed Back Into Fear

The deterioration is even clearer in the Fear & Greed model.

Equal-Weighted Composite

Current: 30.0 — FEAR

One-week change: −22.7 points

Four-week change: −11.4 points

Regime-Weighted Composite

Current: 32.0 — FEAR

One-week change: −19.6 points

Four-week change: −1.7 points

Only a few weeks ago, the equal-weight composite had recovered to nearly 60.

Now it has fallen back to 30.

The speed of that reversal is more important than the absolute number.

This is not a market that has gradually become cautious.

Risk appetite changed rapidly.

The Internal Divergence Is the Biggest Story

The four Fear & Greed components are now sending dramatically different signals.

Trend

Score: 16.5

Classification: EXTREME FEAR

One-week change: −31.2

Four-week change: −10.7

Trend suffered the largest deterioration of any component.

That tells us the price structure beneath the broader market has weakened materially.

Volatility

Score: 64.9

Classification: GREED

One-week change: −2.2

Four-week change: +7.9

Despite recent equity weakness, volatility conditions remain surprisingly constructive.

This is important.

The market is nervous, but it is not behaving like a panic.

Leadership

Score: 8.5

Classification: EXTREME FEAR

One-week change: −10.5

Four-week change: −7.7

This is now the weakest component in the entire model.

A Leadership score of 8.5 indicates an extremely defensive internal market.

That is a major warning.

Markets can continue higher for some time with narrow leadership, but sustainable bull markets generally require participation to eventually broaden.

Credit

Score: 75.5

Classification: GREED

One-week change: +3.3

Four-week change: +12.0

And this is where the story becomes unusual.

While Trend and Leadership collapsed, Credit strengthened.

Credit is now the strongest component of the entire Fear & Greed framework.

The model summary captures the divergence clearly: credit conditions remain supportive, volatility is calm, while Trend is deteriorating and Leadership remains defensive.

Why Credit Matters So Much

Credit has historically been one of the most important confirmation signals during serious market declines.

Equity investors can become nervous very quickly.

Momentum can reverse.

Market leadership can narrow.

Volatility can increase.

But systemic market events usually require something more.

Eventually, credit begins to deteriorate.

That is not happening yet.

Instead:

Credit Score: 75.5

while

Trend Score: 16.5

and

Leadership Score: 8.5

That is an extraordinary spread.

It tells us financial conditions beneath the market remain relatively healthy even though equity investors have become significantly more defensive.

This is one reason I would not interpret the current data as confirmation of a major bear-market transition.

The MTP Analog Engine Is Also Refusing to Turn Bearish

The historical analog model remains neutral over both forecast horizons.

Two-Week Outlook

Probability Higher: 44%

Probability Lower: 24%

Confidence Spread: 20 percentage points

Signal: NEUTRAL

Four-Week Outlook

Probability Higher: 44%

Probability Lower: 28%

Confidence Spread: 16 percentage points

Signal: NEUTRAL

Neither model reaches the required 55% directional probability threshold.

More importantly, neither model is showing strong downside probability.

The two-week analog engine assigns only a 24% probability of a meaningful downward outcome.

The four-week model places that probability at 28%.

This doesn’t mean markets cannot fall.

It means the current combination of MTP indicators does not historically resemble a high-confidence bearish setup.

The Most Interesting Historical Analogs

The model selected the same five historical periods for both the two-week and four-week engines:

September 27, 2013

April 8, 2016

September 17, 2010

August 24, 2012

August 4, 2017

That clustering is worth watching.

The algorithm is finding prior environments in which underlying stress existed but had not yet transitioned into a confirmed systemic event.

The takeaway is not that 2026 must follow any particular historical episode.

Instead, the analog engine reinforces the same message being generated elsewhere:

The current weakness has not yet crossed the threshold into a confirmed bearish regime.

What Changed Since Last Week

This is the section I believe matters most.

One Week Ago

Stress Warning Score: 8.8

Recovery Warning Score: 73.8

Market return: modestly positive

Volatility falling

Fear & Greed considerably stronger

This Week

Stress Warning Score: 22.1

Recovery Warning Score: 65.8

Market return: −2.02%

Volatility: +9.43%

Fear & Greed Equal Composite: 30

Trend: 16.5

Leadership: 8.5

Credit: 75.5

The deterioration is real.

But it has occurred almost entirely inside the equity-risk complex.

That distinction is critical.

The Market Is Developing a Barbell

The current readings suggest two very different forces are operating simultaneously.

Defensive / Risk-Off Side

Trend has collapsed.

Leadership is extremely weak.

Investor sentiment has returned to Fear.

The equity market declined sharply during the week.

Volatility has risen.

Risk-On / Stability Side

Credit is strengthening.

Volatility remains structurally supportive according to the model.

The MTP regime remains RISK-ON.

There is no shock signal.

There is no panic signal.

Historical analogs remain neutral rather than bearish.

That is effectively a barbell market.

The internal equity market is behaving defensively while broader financial conditions remain supportive.

What Would Make This Bullish Again

The cleanest bullish resolution would be for Trend and Leadership to begin catching up with Credit.

Watch for:

Trend: Recover above 20 first, then begin moving toward 40.

Leadership: Recover out of single digits and reclaim 20.

Fear & Greed: Move back above 40–50.

Credit: Remain above 60.

Volatility: Remain constructive.

MTP Stress Score: Fall back below 20.

If that happens, the current episode would increasingly look like a corrective reset inside an intact risk-on regime.

And because sentiment has already deteriorated significantly, a recovery in Trend could potentially produce a powerful rebound.

What Would Turn This Into a Larger Warning

The bearish case becomes much more important if the deterioration spreads beyond equities.

The signals I am watching are:

Credit falls below 60.

Then below 50.

Stress Warning Score rises above 30–40.

Recovery Warning Score collapses toward 40.

MTP regime changes from RISK-ON to STRESS.

Shock flag activates.

Panic score rises above zero.

Leadership remains below 10 while Trend continues deteriorating.

Long-duration Treasuries begin falling aggressively during equity declines.

The dollar begins strengthening sharply at the same time.

That would tell us the current equity correction is spreading into the broader financial system.

We do not have that confirmation today.

CRI Positioning Read-Through

The MTP continues to support selectivity rather than broad defensive liquidation.

The model itself remains at:

80% market exposure

20% defensive exposure

That appears appropriate given the current signals.

I would be cautious about aggressively chasing broad equity exposure while Leadership sits at 8.5.

But I would also be cautious about interpreting Extreme Fear readings as an automatic reason to abandon fundamentally strong structural themes while credit remains exceptionally healthy.

The areas I continue to favor are those where capital spending is supported by structural necessity:

AI infrastructure

Semiconductor equipment

Memory

Optical networking and photonics

Power generation

Grid infrastructure

Cooling

Copper

Energy

Defense

Industrial automation

These areas can certainly correct with the broader market.

But their longer-term capital cycles remain fundamentally different from purely liquidity-driven trades.

The Signal I Care About Most Next Week

There are many numbers in this week’s update.

But the next decision point may ultimately be very simple.

Does Leadership Follow Credit Higher?

or

Does Credit Follow Leadership Lower?

Right now we have:

Leadership: 8.5

versus

Credit: 75.5

That gap cannot continue indefinitely.

One side will eventually move toward the other.

If Leadership begins recovering while Credit remains strong, the market could be setting up for another risk-on advance.

If Credit begins deteriorating toward Leadership, the character of this market changes dramatically.

That is the signal I will be watching most closely.

Bottom Line

The Weekly MTP remains RISK-ON.

There has been no regime change.

The shock state remains NORMAL.

The panic score remains 0.

The historical analog model remains NEUTRAL over both two and four weeks.

But the internal market has deteriorated substantially.

Fear & Greed has fallen back to 30.

Trend has collapsed to 16.5.

Leadership has fallen to just 8.5.

Yet Credit has strengthened to 75.5.

That produces a highly unusual but potentially important setup:

Extreme equity-market fear without confirmation from credit or systemic stress indicators.

For now, I would describe the regime as:

RISK-ON, BUT FRAGILE.

The current weakness looks more like a correction and internal rotation than a confirmed bear-market transition.

But unlike several earlier pullbacks, the deterioration in Trend and Leadership is significant enough that it should not be dismissed.

The next phase will be determined by whether market participation recovers toward healthy credit conditions—or credit begins deteriorating toward the weakness already visible inside equities.

That divergence is now the central MTP signal.

Executive Read

This week’s Microcap Discovery screen produced 315 eligible companies, with:

• 54 Promotion candidates

• 72 Accumulation candidates

• 189 Early Monitor names

• Zero liquidity rejects

The headline is not simply that microcaps are strengthening.

It is that multiple forms of leadership are now developing simultaneously.

AI Infrastructure/Semiconductors has reclaimed the highest average theme score at 64.0, supported by exceptionally strong capital-flow readings.

Healthcare/Biotech remains the deepest source of breadth, producing 27 Promotion candidates and 27 Accumulation candidates.

But the development I find particularly important this week is AI Software/Data. It now has 12 Promotion candidates and another 11 in Accumulation.

That gives us three different signals:

AI Infrastructure = highest-quality thematic leadership

Healthcare/Biotech = deepest current breadth

AI Software/Data = increasingly important expansion pipeline

This is becoming less of a single-theme microcap market and more of a selective leadership expansion.

The Microcap Leadership Map

The model separates companies into two groups that matter most for this report.

Active / Promotion Leaders

These are companies where price structure, relative strength and capital flow have already aligned.

Emerging / Accumulation Leaders

These are earlier-stage candidates where enough evidence exists to warrant attention, but confirmation is still developing.

The distinction matters.

The highest-return opportunity is not necessarily the company with the highest current score. It may be the company whose fundamental opportunity and capital-flow structure are beginning to align before the market fully recognizes it.

Active Leaders: Where Capital Is Already Moving

1. UMAC — Unusual Machines

MicroCap Score: 94

• 1W RS vs SPY: +30.0%

• 4W RS vs SPY: +96.7%

• 13W RS vs SPY: +104.9%

• Revenue Growth YoY: +687.3%

• Capital Flow: Very Strong

• Structure: Compression + Capital Flow

UMAC is the highest-scoring company in this week’s screen.

The combination of extraordinary revenue growth, accelerating relative strength and very strong capital flow makes it impossible to ignore.

The attraction is exposure to the rapidly developing drone ecosystem and the potential localization of drone-component supply chains.

But this is also a useful example of why Microcap Discovery cannot stop at the score.

The company remains cash-flow negative despite its rapid growth.

Long-Term Growth Multiple Path: UMAC needs to convert explosive top-line growth into scalable economics while proving that demand is structural rather than temporary. If it can do that, the addressable opportunity is substantial. If not, the current valuation can become highly sensitive to expectations.

2. BLFS — BioLife Solutions

MicroCap Score: 90

• 4W RS vs SPY: +15.9%

• 13W RS vs SPY: +63.6%

• Revenue Growth YoY: +21.5%

• Gross Margin: 63.6%

• Operating Cash Flow: +$16.9M

• Free Cash Flow: +$4.7M

• Capital Flow: Strong

BLFS remains one of the most fundamentally interesting companies in the healthcare leadership group.

Unlike many biotech names in the screen, BioLife sells the infrastructure used across the cell and gene therapy ecosystem.

That means its opportunity is less dependent on the success of one drug.

Revenue growth, high gross margins and positive cash generation provide fundamental support beneath the technical signal.

Long-Term Growth Multiple Path: Continued expansion of cell and gene therapy development could increase demand for preservation, storage and bioproduction infrastructure. Sustained 15–25% growth combined with margin and cash-flow expansion would provide a much stronger basis for long-term compounding than catalyst-dependent biotechnology.

3. XNCR — Xencor

MicroCap Score: 88

• 1W RS vs SPY: +11.8%

• 4W RS vs SPY: +29.2%

• 13W RS vs SPY: +102.7%

• Revenue Growth YoY: +17.5%

• Cash: $456M

• Capital Flow: Strong

Xencor’s relative strength has become significant.

The company also has substantial liquidity, which matters enormously for a development-stage biotechnology company.

However, this remains fundamentally different from BLFS.

XNCR is still consuming significant cash, making pipeline execution and clinical outcomes central to the thesis.

Long-Term Growth Multiple Path: The platform must translate into successful programs, partnerships and eventually durable commercial economics. Until then, the model signal should be treated as strong capital-flow confirmation around a catalyst-sensitive business.

4. SABR — Sabre

MicroCap Score: 88

• 1W RS vs SPY: +7.3%

• 4W RS vs SPY: +24.5%

• 13W RS vs SPY: +34.5%

• Revenue Growth YoY: +3.6%

• Gross Margin: 56.4%

• Free Cash Flow: +$100.8M

• Capital Flow: Very Strong

SABR is one of the more interesting non-biotech signals this week.

Revenue growth is modest, but the business is generating free cash flow and capital flow has become very strong.

The complication is substantial debt.

This is therefore primarily a cash-flow and deleveraging re-rating thesis, rather than a pure high-growth story.

Long-Term Growth Multiple Path: Better margins, sustained free cash flow and meaningful debt reduction could create equity operating leverage. Failure to deleverage would limit that opportunity.

5. INDI — indie Semiconductor

MicroCap Score: 88

• 1W RS vs SPY: +16.2%

• 4W RS vs SPY: +18.5%

• Revenue Growth YoY: +24.0%

• Capital Flow: Very Strong

• Theme: AI Infrastructure / Semiconductors

INDI is important because it expands the semiconductor story beyond data centers.

The company’s exposure to automotive sensing, connectivity and vehicle intelligence places it within the broader transition toward increasingly compute-intensive vehicles.

Revenue growth supports the thematic thesis, although profitability remains the key hurdle.

Long-Term Growth Multiple Path: Sustained 20%+ revenue growth, design-win conversion and a transition toward positive cash generation would substantially improve the quality of the story.

Other Promotion Leaders Worth Watching

QMCO — Quantum Corporation | Score 84
4W RS vs SPY +141.2%; 13W +207.4%; revenue +25.7%. Data-storage infrastructure with very strong momentum and improving cash generation.

EXFY — Expensify | Score 83
13W RS vs SPY +96.5%. Positive operating and free cash flow, but revenue remains down 5.3%. A re-rating requires renewed growth.

CRON — Cronos Group | Score 80
Revenue +58.4%, positive operating cash flow and approximately $797M of cash against minimal debt. One of the stronger balance-sheet profiles in the screen.

SONO — Sonos | Score 80
Revenue +8.8%, 46.6% gross margin and more than $111M of trailing free cash flow. Less explosive technically, but fundamentally more mature than many Promotion names.

AEVA — Aeva Technologies | Score 79
Capital Flow Very Strong. A potentially important autonomy/sensing platform, but still deeply cash-flow negative and therefore dependent on commercialization execution.

What the Promotion List Is Really Telling Us

Healthcare still dominates numerically, but this week’s Promotion layer is becoming more diverse.

The model now contains meaningful representation from:

• AI infrastructure

• AI software and data

• semiconductors

• autonomy

• data storage

• consumer technology

• industrials

• healthcare infrastructure

• clinical biotechnology

That distinction matters.

Breadth is improving without eliminating selectivity.

That is generally healthier than a market where every low-quality microcap begins moving simultaneously.

Emerging Leaders: Where the Next Rotation May Be Forming

The 72-name Accumulation list is too large to treat every company equally.

The objective here is therefore not simply to show the ten highest scores.

It is to identify companies where capital-flow evidence and a plausible long-term business opportunity intersect.

SPT — Sprout Social

MicroCap Score: 68

• 4W RS vs SPY: +10.3%

• 13W RS vs SPY: +52.9%

• Revenue Growth: +10.8%

• Gross Margin: 77.4%

• Operating Cash Flow: +$53.9M

• Capital Flow: Very Strong

SPT may be one of the more interesting Accumulation names.

It combines a high-margin software model with positive cash generation and improving relative strength.

What Needs to Happen: Revenue growth needs to reaccelerate while operating leverage improves. If that occurs, the combination of recurring software economics and AI-enabled social/data workflows could support a meaningful re-rating.

HLMN — Hillman Solutions

MicroCap Score: 69

• 13W RS vs SPY: +15.4%

• Revenue Growth: +9.8%

• Gross Margin: 48.1%

• Operating Cash Flow: +$125.6M

• Free Cash Flow: +$60.1M

HLMN is not an exciting AI story—and that is precisely why it is useful.

It represents an established industrial business showing improving structure while producing real cash.

What Needs to Happen: Continued organic growth, margin expansion and debt reduction could turn a relatively ordinary industrial company into a steady small-cap compounder.

OMDA — Omada Health

MicroCap Score: 67

• Revenue Growth: +43.1%

• Gross Margin: 68.2%

• Operating Cash Flow: +$23.7M

• Free Cash Flow: +$11.9M

• Net cash balance sheet

OMDA stands out fundamentally within healthcare.

Strong revenue growth, attractive gross margins, positive cash generation and no reported debt give it a very different profile from speculative biotech.

The technical structure remains mixed, which is precisely why it belongs in Accumulation rather than Promotion.

What Needs to Happen: Continued growth with durable profitability could allow the market to value OMDA as a scalable digital-health platform rather than simply another healthcare small cap.

WYFI — WhiteFiber

MicroCap Score: 67

• 1W RS vs SPY: +19.5%

• Revenue Growth: +55.7%

• Gross Margin: 85.9%

• Capital Flow: Very Strong

• Theme: AI Optical / Networking

WYFI deserves attention because the fundamental growth and margin numbers are unusually strong.

The complication is cash flow: reported free cash flow remains deeply negative.

What Needs to Happen: Revenue growth must translate into cash generation. If it does, the combination of AI networking exposure, high margins and rapid growth could become particularly interesting.

PRCH — Porch Group

MicroCap Score: 68

• 4W RS vs SPY: +22.6%

• 13W RS vs SPY: +71.3%

• Revenue Growth: +11.7%

• Gross Margin: 72.1%

• Operating Cash Flow: +$96.3M

PRCH combines improving relative strength with real operating cash generation.

Its financial structure remains more complicated, so balance-sheet and free-cash-flow quality deserve continued scrutiny.

What Needs to Happen: Improved cash conversion and financial simplification would strengthen the re-rating case.

NEOV — NeoVolta

MicroCap Score: 69

• 1W RS vs SPY: +27.6%

• 4W RS vs SPY: +67.6%

• 13W RS vs SPY: +36.5%

• Capital Flow: Moderate

The price signal is considerably stronger than the current operating fundamentals.

That makes NEOV a classic early-stage watch rather than a fundamentally confirmed leader.

What Needs to Happen: Commercial energy-storage growth must accelerate materially and ultimately generate positive cash flow.

AGEN — Agenus

MicroCap Score: 66

• 4W RS vs SPY: +41.7%

• 13W RS vs SPY: +99.2%

• Revenue Growth: +34.4%

• Capital Flow: Strong

AGEN has powerful price momentum, but remains highly dependent on biotechnology development outcomes.

It belongs on the radar, but in a fundamentally different risk bucket from operating businesses such as BLFS, SPT or HLMN.

PSNL — Personalis

MicroCap Score: 66

• 13W RS vs SPY: +115.5%

• Revenue Growth: +30.0%

• Cash: $212.7M

PSNL remains interesting because diagnostics and precision oncology provide a real commercialization path beyond binary drug-development outcomes.

The current structure has weakened over the shorter 4-week period, making this an appropriate Accumulation rather than Promotion candidate.

Theme Leadership

1. AI Infrastructure / Semiconductors — Highest Quality

Average Score: 64.0

• 3 Promotion

• 1 Accumulation

• Average RS Score: 16.0

• Average Capital Flow Score: 17.1

Only seven companies fall into this group, but the quality of the signals is exceptionally high.

This is concentrated leadership, not broad participation.

2. Healthcare / Biotech — Deepest Breadth

Average Score: 55.3

• 98 companies

• 27 Promotion

• 27 Accumulation

• 13W sector-relative strength remains positive

Healthcare remains the largest leadership reservoir.

But readers should not interpret all 54 Promotion/Accumulation healthcare names equally.

The strongest opportunities are those where capital flow is supported by either commercial revenue, strong balance sheets, platform economics or clearly identifiable catalysts.

3. AI Optical / Networking — Small but Strong

Average Score: 53.8

There are only four names in this category.

Yet capital-flow and relative-strength readings remain strong.

That makes optical/networking another example of narrow but high-quality participation.

4. AI Software / Data — Expansion Accelerating

Average Score: 49.8

• 12 Promotion

• 11 Accumulation

This is arguably the most important pipeline development this week.

AI Software/Data now has 23 companies across the two primary discovery layers.

Unlike semiconductor infrastructure, this is becoming a breadth story.

If this persists, software/data could become one of the next major sources of microcap leadership.

5. Energy and Industrials — Still Forming

Energy/Power has:

• 1 Promotion

• 8 Accumulation

Industrials/Automation has:

• 1 Promotion

• 6 Accumulation

Neither has confirmed broad leadership.

But both have considerably more companies in Accumulation than Promotion.

That is exactly the relationship we want to monitor for an early rotation.

The Fundamental Divide

This week’s report reinforces an important rule for Microcap Discovery:

Technical leadership identifies where capital is moving. Fundamental validation helps determine whether that movement can become durable.

There are three broad types of opportunity in the screen:

Operating Compounder Candidates

Examples include BLFS, SPT, OMDA, HLMN and SONO.

These businesses have some combination of revenue, margins and cash generation that can support longer-duration valuation expansion.

Early Commercial Growth Candidates

Examples include UMAC, INDI, WYFI and AEVA.

These can produce much larger growth rates, but execution and cash generation still need to catch up with expectations.

Catalyst-Driven Opportunities

A large portion of biotechnology falls here.

These can generate enormous returns, but outcomes may depend on trials, regulatory decisions, partnerships or financing.

The Microcap score tells us where capital is flowing. It does not erase the difference between these business models.

CRI Interpretation

The August 14 screen is constructive.

We have:

• exceptionally strong AI infrastructure signals

• persistent healthcare breadth

• rapidly expanding AI software participation

• early energy and industrial pipelines

• no liquidity rejects in this week’s eligible universe

What we do not have is indiscriminate speculative participation.

That is an important distinction.

Leadership is broadening, but capital is still discriminating between companies.

That is a considerably healthier backdrop for discovery than broad microcap euphoria.

What I’m Watching Next

Three developments matter most over the coming weeks:

1. Does AI Software/Data continue graduating from Accumulation into Promotion?

Twelve Promotion names already make this a meaningful development.

2. Does Healthcare leadership become fundamentally broader—or remain primarily catalyst driven?

The answer determines how durable the current healthcare signal becomes.

3. Do Energy and Industrials begin graduating their accumulation pipelines?

Together those groups now contain 14 Accumulation names but only two Promotion candidates.

If that ratio begins changing, it could mark the beginning of the next sector rotation.

Bottom Line

This week’s Microcap Discovery screen is stronger than a simple list of high-scoring stocks suggests.

The underlying leadership map shows:

AI Infrastructure — concentrated quality

Healthcare/Biotech — broad established leadership

AI Software/Data — rapidly developing breadth

Energy/Industrials — early pipeline

And within the individual companies, an equally important divide is emerging between technically strong speculation and businesses with the financial characteristics capable of supporting sustained growth.

That is ultimately what Microcap Discovery is trying to find.

Not simply the next stock to move 20%.

But the small companies where capital flow, structural growth and improving business economics can converge early enough to create the possibility of much larger long-term value creation.

Microcaps involve substantial volatility, liquidity and company-specific risk. This research framework is designed for discovery and monitoring, not as a substitute for individual due diligence.

Tags

microcap stocks, small cap stocks, emerging growth stocks, AI infrastructure, AI software, semiconductor stocks, biotech stocks, healthcare stocks, optical networking, industrial automation, energy infrastructure, capital flows, sector rotation, emerging leaders, CRI, Capital Regime Intelligence

What Is Working Beneath the Sector Headlines

The sector rotation model tells us where capital is flowing.

The subsector model tells us whether that move is broad, narrow, investable, or misleading.

That distinction matters this week.

At the surface, the market looks constructive. Financials, Industrials, Technology, Health Care, and Energy are all showing leadership or improving rotation signals.

But underneath the surface, the picture is more specific.

The strongest internal leadership is concentrated in banks and capital markets, industrial infrastructure, electrical equipment, machinery, AI networking, semiconductors, optical infrastructure, refiners, integrated energy, E&P, midstream, select health care, copper, and a few specialty real estate names.

The weakest areas remain broad consumer discretionary, broad communication services, most utilities outside select power-linked names, weaker materials pockets outside copper and select metals, and broad real estate outside specialty winners.

The CRI read:

This remains a selective risk-on market. Capital is still favoring the physical economy beneath the market: financials, industrial infrastructure, AI supply chains, energy, copper, power, and select quality. But broad consumer, broad utilities, and broad communication services are not confirming leadership.

Executive Subsector Read

What is working best underneath the surface

Financials are broad.
The strength is not isolated to one bank. The model is showing broad participation across banks, capital markets, brokers, asset managers, insurers, and credit-sensitive financials.

Industrials remain infrastructure-led.
The strongest read-through is coming from machinery, electrical equipment, automation, aerospace/defense, and industrial infrastructure. This continues to support the AI power, grid, reshoring, defense, and capital-equipment themes.

Energy improved with real breadth.
The move is not just one oil major. Integrated energy, E&P, refiners, and midstream are all participating. Services remain the softer pocket, but the internal Energy signal is much better than it was.

AI infrastructure remains healthy but selective.
NVIDIA, Broadcom, AMD, Arista, ASML, TSMC, Microsoft, and SOXX remain important leadership markers. The short-term caution is that the semiconductor complex is digesting after a major run, so the next move requires confirmation.

Health Care is stable, but leadership is not broad enough to call it aggressive.
Eli Lilly remains the standout. The broader group is constructive, but Health Care is acting more like a quality stabilizer than a full risk-on leader.

What is not working well

Consumer is narrow.
Amazon, Walmart, and Costco-type leadership is not the same as broad consumer strength. The broader discretionary and staples groups remain weak or inconsistent.

Communication Services is still narrow.
Google remains the clearest leader, but the sector is not showing broad internal strength.

Utilities are not confirming broad leadership.
Constellation Energy continues to matter for the AI power theme, but broad Utilities remain weak. This is an important distinction.

Materials are selective.
Freeport-McMoRan and select metals exposure remain important, but broad Materials are not yet confirming durable sector-wide leadership.

Real Estate is stock-specific.
Equinix and select specialty names matter, but broad REIT exposure is still not leadership.

Subsector Health Cards

AI / Semiconductor Infrastructure

Current health: Strong but selective
Direction: Constructive, but consolidating

The AI infrastructure model remains structurally healthy, but the move is not equally strong across every layer.

The leadership basket continues to center on NVIDIA, Broadcom, AMD, Arista, ASML, TSMC, Microsoft, and SOXX.

The strongest internal pockets remain:

Compute and accelerators
NVIDIA and AMD remain central to the AI compute layer. NVIDIA continues to anchor the dominant GPU infrastructure trade, while AMD remains important as the market broadens toward alternate accelerators and rack-scale systems.

Custom silicon and advanced foundry exposure
Broadcom and TSMC remain key read-through names. Broadcom captures the custom accelerator and AI networking theme, while TSMC remains the manufacturing backbone for advanced AI silicon.

Networking
Arista remains one of the cleanest signs that AI networking is no longer a secondary theme. The market continues to reward companies tied to high-speed AI fabrics, Ethernet networking, and the scale-out architecture required to keep accelerator clusters productive.

Semiconductor equipment and advanced manufacturing
ASML remains an important long-cycle marker for leading-edge semiconductor capacity. This does not always move week to week with AI hype, but it remains one of the foundation layers beneath advanced compute.

Mega-cap cloud infrastructure
Microsoft remains a key AI infrastructure demand signal because Azure growth, AI capacity, and cloud capex all connect directly to the buildout.

The softer area is not the long-term AI thesis. The softer area is short-term momentum. SOXX remains structurally above long-term trend, but it is digesting after a strong move. That argues for selectivity rather than abandoning the theme.

CRI read-through

AI remains healthy, but the best opportunity is not simply “own anything AI.” The better read is to focus on the infrastructure stack: compute, custom silicon, networking, foundry, optical, power, cooling, and packaging.

What is working

NVIDIA, Broadcom, AMD, Arista, TSMC, ASML, Microsoft, SOXX.

What is weaker or needs confirmation

Broad semiconductor momentum after the recent run. The group needs SOXX and networking leadership to continue confirming.

Financials

Current health: Very Strong
Direction: Broad leadership confirmed

Financials remain one of the cleanest internal leadership groups in the model.

The important point is breadth. This is not one bank dragging the group higher. Strength is showing across several financial pockets.

Banks
JPMorgan, Bank of America, Wells Fargo, and Citi remain important markers for credit, lending, balance-sheet confidence, and economic participation.

Capital markets and investment banking
Goldman Sachs and Morgan Stanley matter because they reflect capital-market activity, deal flow, trading, underwriting, and risk appetite.

Asset management and brokerage
BlackRock and Schwab help confirm market participation beyond banks. When these names are firm, it usually supports a healthier capital-market backdrop.

Insurance and diversified financials
Berkshire and other insurance-linked financials add a quality and balance-sheet component to the group.

Payments and credit-sensitive consumer finance
American Express is important because it ties financial leadership to higher-end consumer credit and spending resilience.

CRI read-through

Financials are still confirming risk-on participation. This is not what a broad defensive rotation usually looks like.

If banks, brokers, asset managers, and insurers are all participating, the market is still comfortable owning credit-sensitive and capital-market-sensitive exposure.

What is working

JPMorgan, Goldman Sachs, Morgan Stanley, Bank of America, Wells Fargo, BlackRock, Schwab, American Express, Citi, Berkshire.

What would weaken the view

A break in bank breadth, weakness in capital-market names, widening credit stress, or a rotation out of insurers and brokers.

Industrials

Current health: Very Strong
Direction: Leadership persistent

Industrials remain one of the strongest subsector groups because the strength is tied to the real economy and infrastructure buildout.

The strongest internal areas remain:

Electrical equipment and power infrastructure
Eaton and other electrical-equipment names remain central to the AI power, grid, data-center electrification, and infrastructure-upgrade theme. This is one of the most important bridges between AI and industrials.

Machinery and capital equipment
Caterpillar and Deere remain important cyclical markers. CAT ties to construction, mining, infrastructure, and energy investment. DE is more tied to agriculture and equipment cycles, but still helps track machinery participation.

Motion, hydraulics, and industrial systems
Parker-Hannifin is a strong industrial-quality signal because it cuts across automation, aerospace, manufacturing, hydraulics, and engineered systems.

Aerospace and defense
GE Aerospace, RTX, and Lockheed Martin keep defense and aerospace in the leadership conversation. These areas can support Industrials even when some economically sensitive areas pause.

Transportation and rails
Union Pacific and Norfolk Southern are useful freight-cycle markers. The rails matter because they help distinguish true industrial-cycle strength from only defense or electrical-equipment leadership.

CRI read-through

Industrials continue to confirm the “physical economy beneath the AI trade.” The strength is not just traditional cyclicals. It is electrical equipment, automation, power infrastructure, aerospace, defense, machinery, and transportation.

This supports the CRI themes of AI power demand, grid expansion, reshoring, automation, defense spending, and capital-equipment replacement.

What is working

CAT, DE, ETN, PH, GE, RTX, UNP, NSC, LMT.

What needs monitoring

Transportation and freight-sensitive areas. If rails weaken while electrical equipment remains strong, the market is narrowing toward infrastructure rather than broad industrial expansion.

Energy

Current health: Very Strong
Direction: Accelerating

Energy delivered the most important improvement in the subsector model.

This is not a one-stock move.

The internal strength is broad.

Integrated energy

The integrated majors are participating. This matters because they usually reflect the market’s view on the overall oil and gas cash-flow cycle, capital discipline, production, refining exposure, LNG, and shareholder returns.

Exploration and production

E&P strength confirms that upstream commodity sensitivity is back on the screen. This is important for crude, natural gas, drilling economics, and real-asset exposure.

Refiners

Refiners showing strength is especially important because they are more tied to crack spreads, product demand, diesel, gasoline, and margin conditions. Refining strength can indicate a different part of the energy cycle than simply crude moving higher.

Midstream

Midstream strength adds a cash-flow and infrastructure component. This fits the CRI preference for real assets, pipelines, LNG, natural gas, and infrastructure-like energy exposure.

Services

Energy services remain the softer pocket. This matters because a fully durable Energy advance would ideally include stronger confirmation from services, equipment, and activity-sensitive names.

CRI read-through

Energy is back on the rotation screen, and the internal breadth makes the move worth respecting.

The best interpretation is not simply “oil is up.” The better interpretation is that capital is rotating back into real assets, energy cash flow, refining, midstream infrastructure, and potentially the broader AI power discussion.

Energy also matters because the AI buildout is increasingly power-constrained. Data centers require electricity, backup power, natural gas, turbines, grid equipment, and reliable fuel systems. Energy strength adds another confirmation that the market is beginning to price the physical inputs behind AI.

What is working

Integrated energy, E&P, refiners, midstream.

What is lagging

Energy services remain the area that needs better confirmation.

What would improve the view

Follow-through for two to three more weeks, with services participation improving.

Health Care

Current health: Strong / constructive
Direction: Stable, but leadership is selective

Health Care remains internally healthy, but the leadership is not as aggressive as Financials, Industrials, Energy, or AI infrastructure.

The most important distinction is that Health Care is acting more like a quality stabilizer than a pure risk-on leadership sector.

Large-cap pharma and obesity/diabetes leadership

Eli Lilly remains the clearest leadership name inside the group. LLY continues to act as the strongest quality-growth marker in Health Care.

Medical devices and medtech

Medtech remains important to watch because it can provide steadier growth exposure, but the group is not clearly driving the broader market leadership signal.

Managed care and health services

This area remains more mixed. When managed care is weak, it can hold back the broader sector even if pharma leadership is strong.

Biotech

Biotech can help confirm risk appetite inside Health Care, but the broad sector signal is still more quality-oriented than speculative.

CRI read-through

Health Care remains useful in the portfolio mix, but the subsector evidence does not yet suggest broad aggressive leadership. It is more defensive quality plus selective growth.

The market still wants exposure to high-quality health care winners, but it is not treating the entire sector as a leadership engine.

What is working

LLY and select quality health care leadership.

What needs confirmation

Broader participation from medtech, managed care, services, and biotech.

Materials

Current health: Mixed / selective
Direction: Improving in pockets, not broad leadership

Materials remain a selective trade, not a broad sector call.

The strongest internal message is still coming from metals and infrastructure-linked materials rather than the entire group.

Copper and electrification

Freeport-McMoRan remains the most important Materials name in the CRI framework. FCX connects directly to copper, electrification, grid expansion, AI power demand, transmission, EV infrastructure, and industrial buildout.

Gold and precious metals

Newmont remains important as a gold and precious-metals marker. This gives the Materials basket a real-asset and monetary-hedge component.

Industrial gases and chemicals

Linde and Air Products matter as quality industrial gas names, but the broader chemicals complex is not showing the same clear leadership as copper or infrastructure-linked materials.

Aggregates and construction materials

Vulcan Materials and Martin Marietta remain important infrastructure read-through names. If they strengthen, that would support construction, public infrastructure, data-center construction, and real-asset investment.

Coatings and specialty materials

Sherwin-Williams and Ecolab provide quality exposure, but these are not the same as commodity-cycle leadership.

CRI read-through

Materials are not broad enough to upgrade aggressively.

The right read is:

Own the theme selectively. Copper matters. Gold matters. Aggregates matter. But broad Materials still need confirmation.

What is working

FCX, NEM, LIN, VMC, MLM, and select infrastructure-linked materials.

What is weaker

Broad chemicals and less differentiated Materials exposure.

What would improve the view

Broader participation beyond FCX and select real-asset names.

Real Estate

Current health: Mixed / selective
Direction: Internally improving, but broad sector still lagging

Real Estate remains one of the clearest examples of why subsector work matters.

Broad Real Estate is not leadership.

But select real estate pockets matter.

Data centers and digital infrastructure

Equinix remains the most important name in this framework because data-center real estate is tied to AI capacity, connectivity, power availability, cloud infrastructure, and enterprise demand.

This is very different from broad REIT exposure.

Retail and mall exposure

Simon Property Group remains a useful read-through for high-quality retail real estate. Strength here can indicate that premium physical retail assets are holding up better than weaker consumer-sensitive real estate.

Broad REITs

The broader REIT complex remains less attractive because of rate sensitivity, financing costs, refinancing pressure, and weak relative strength versus SPY.

CRI read-through

Real Estate is not a broad sector allocation yet.

The investable distinction is:

Data-center and specialty real estate can be attractive even when broad REITs are not.

That means the model supports selectivity, not blanket Real Estate exposure.

What is working

EQIX, SPG, and select specialty real estate.

What is weaker

Broad REIT exposure, rate-sensitive real estate, and lower-quality property types.

Consumer

Current health: Weak / narrow
Direction: Deteriorating

Consumer remains one of the weaker subsector areas.

The most important message is narrowness.

Platform retail

Amazon remains the most important leadership name in Consumer Discretionary. AMZN strength can support the sector optically, but it does not automatically mean the full consumer complex is healthy.

Discount and warehouse retail

Walmart and Costco remain stronger consumer names. That suggests consumers may still be favoring value, scale, membership models, and defensive retail strength.

Broad discretionary

The broader Consumer Discretionary group remains weak. This includes areas tied to apparel, restaurants, home improvement, autos, and rate-sensitive spending.

Staples

Staples are not showing enough leadership to confirm a defensive consumer rotation. Walmart and Costco can be strong while the broader Staples sector remains unconvincing.

CRI read-through

Consumer is not confirming broad expansion.

The consumer market is split between scaled winners and weaker broad discretionary participation. That argues against a broad consumer-led market call.

What is working

AMZN, WMT, COST.

What is weaker

Broad discretionary, weaker retail, apparel, restaurants, home improvement, and broad staples participation.

What would improve the view

XLY improving versus SPY and breadth expanding beyond Amazon and a few large defensive retailers.

Communication Services

Current health: Weak / narrow
Direction: Stabilizing, but not leading

Communication Services remains weak internally.

The key issue is narrowness.

Search / platform leadership

Google remains the strongest internal name in the group. GOOG is still the clearest stock-specific leadership signal.

Broader platform exposure

The rest of the sector is not producing a strong enough breadth signal to call Communication Services a leadership group.

Media, telecom, and weaker communication pockets

Traditional media, telecom, and weaker advertising-linked areas remain less convincing.

CRI read-through

Communication Services is not confirming broad leadership.

This matters because mega-cap platform stocks can make the sector look more important than its internal breadth supports. The model says there may be individual winners, but the group as a whole is still not leading.

What is working

GOOG.

What is weaker

Broad Communication Services, telecom, media, and weaker platform participation.

What would improve the view

Breadth above the 20-week trend threshold and leadership expanding beyond Google.

Utilities

Current health: Weak overall
Direction: Mixed, with one major exception

Utilities remain weak as a broad group.

But there is one important exception.

Nuclear / AI power exposure

Constellation Energy remains the standout. CEG continues to fit the AI power theme because large data centers increasingly need reliable, high-capacity, low-carbon baseload power.

This is the key distinction:

CEG strength supports the AI power thesis. It does not mean broad Utilities are healthy.

Regulated utilities

Most regulated utilities remain weaker because the group is still dealing with rate sensitivity, capital intensity, regulatory lag, and lack of broad relative strength.

Renewable and defensive yield exposure

The defensive yield parts of Utilities are not confirming a major risk-off rotation. If investors were fully hiding in defensives, Utilities breadth should be stronger.

CRI read-through

Do not confuse the AI power trade with the Utilities sector.

The better expression of the AI power theme may still be targeted exposure to nuclear, turbines, electrical equipment, transformers, switchgear, grid infrastructure, and data-center power systems rather than broad XLU.

What is working

CEG and targeted AI power exposure.

What is weaker

Broad Utilities, regulated utilities, and defensive yield exposure.

What would improve the view

XLU relative strength improving and breadth expanding beyond CEG.

Cross-Subsector Theme Map

1. Financial breadth confirms risk-on participation

The strongest Financials signal is breadth across banks, brokers, capital markets, insurers, asset managers, and credit-sensitive names.

That supports a market still willing to own economic sensitivity.

This is not a pure defensive market.

2. Industrials confirm the physical infrastructure cycle

The strength inside Industrials is exactly where CRI wants to see it:

Electrical equipment
Machinery
Automation
Aerospace
Defense
Infrastructure
Rails and transport markers

That keeps the “real economy beneath the AI trade” thesis alive.

3. Energy is no longer just a watchlist group

Energy has moved into active rotation.

The important confirmation is internal breadth across integrated energy, E&P, refiners, and midstream.

Services are still the weak link, but the group is much healthier than it was.

4. AI infrastructure remains healthy, but leadership is more selective

The strongest AI names remain tied to compute, custom silicon, networking, foundry, and cloud infrastructure.

NVIDIA, Broadcom, AMD, Arista, TSMC, ASML, Microsoft, and SOXX remain the key markers.

The message is not “AI is weak.”

The message is “AI needs confirmation after a major move.”

5. Consumer is not confirming broad strength

Consumer leadership is narrow.

Amazon, Walmart, and Costco can be strong while broad discretionary, apparel, restaurants, home improvement, and weaker retail remain under pressure.

That is not broad consumer leadership.

6. Utilities are not the same as the power trade

This is one of the most important conclusions of the week.

Broad Utilities remain weak.

Constellation Energy remains important.

The AI power trade should be tracked through targeted power, nuclear, grid, electrical equipment, turbines, transformers, and data-center energy beneficiaries—not simply through XLU.

7. Materials remain a targeted real-asset theme

Materials are not broad leadership.

The stronger interpretation is selective:

Copper
Gold
Aggregates
Industrial gases
Infrastructure-linked materials

FCX remains the most important CRI Materials marker because copper is tied directly to AI power, grid, electrification, and industrial infrastructure.

CRI Bottom Line

This week’s subsector model confirms a selective risk-on market.

The strongest leadership is underneath the surface, not evenly spread across every sector.

Best internal confirmation

Financials: banks, brokers, asset managers, insurers, and capital markets.
Industrials: electrical equipment, machinery, aerospace/defense, infrastructure, rails.
Energy: integrated energy, E&P, refiners, and midstream.
AI infrastructure: compute, custom silicon, networking, foundry, and cloud infrastructure.
Materials: copper, gold, and infrastructure-linked materials.
Health Care: Eli Lilly and select quality health care.

Weakest internal confirmation

Consumer: narrow leadership around AMZN, WMT, and COST.
Communication Services: narrow leadership around GOOG.
Utilities: broad weakness outside CEG.
Real Estate: select EQIX/SPG strength, but broad REITs still lag.
Materials: broad chemicals and less differentiated materials remain mixed.

The core CRI conclusion:

The market is still rewarding the physical economy beneath the index: financials, industrial infrastructure, energy, AI supply chains, copper, power, and select quality.

But this is not broad market participation.

It is not broad consumer leadership.

It is not broad defensive rotation.

The tactical message for next week:

Stay with internal leadership. Respect the Energy breadth. Keep AI selective. Favor infrastructure-linked Industrials and Financials. Do not confuse broad Utilities with the AI power trade. Do not assume Consumer weakness has been resolved until the model confirms it.

This week’s sector model shows a market that remains constructive, but not evenly distributed.

The leadership stack is still concentrated in Financials, Industrials, Technology, and Health Care, with Energy making the largest weekly improvement and moving from an improving/watch position into a leadership quadrant.

The important message is not simply that the market is rising.

The message is that capital is still favoring sectors tied to credit, industrial activity, AI infrastructure, health care stability, and now renewed energy momentum.

At the same time, the weakest areas remain clear: Communication Services, Utilities, Consumer Discretionary, and Consumer Staples continue to lag the model, suggesting that defensive rotation and consumer-led leadership are not yet the dominant market regime.

The weekly CRI read:

The market remains in a risk-on but selective rotation. Leadership is still cyclical and infrastructure-oriented, while traditional defensive sectors are not confirming a broad risk-off turn.

Executive Rotation Read

Current leadership stack

1. Financials — Rank 1 — Overweight
2. Industrials — Rank 2 — Accumulate
3. Information Technology — Rank 3 — Accumulate
4. Health Care — Rank 4 — Accumulate
5. Energy — Rank 5 — Hold, but sharply improving

The top of the model remains consistent with a market that is still rewarding economically sensitive and capital-market-linked leadership.

Financials hold the top rank with the strongest overall score. Industrials remain near the top, supporting the broader infrastructure and reshoring theme. Technology remains strong, especially on trend. Health Care remains a constructive stabilizer. Energy delivered the biggest weekly improvement and moved into the Leader quadrant.

That combination does not look like a defensive breakdown.

It looks like a market still willing to own cyclical leadership, AI-adjacent infrastructure, capital-market participation, and selective real-asset exposure.

Market Structure Snapshot

SPY gained roughly 0.4% for the week and is up about 4.4% over four weeks in the model data.

The strongest one-week sector move came from Energy, up approximately 7.7%. Technology and Energy were the only major sectors with clear positive four-week relative strength versus SPY.

The 12-week leadership picture is more balanced. Financials, Health Care, Industrials, and Technology all remain positive on relative 12-week trends.

That tells us the market is not only chasing one short-term group. There is still a broader rotation structure underneath the index.

Sector Health Cards

Financials

Current rank: 1
Model action: Overweight
Rotation quadrant: Leader
Structural phase: Expansion
Sector score: 80.8

Financials remain the strongest sector in the model.

The sector has the best overall score, the highest relative-strength score, and very strong breadth. Financials also improved slightly from last week and moved from Accumulate to Overweight.

The breadth signal is especially important. The model shows 100% of constituents above the 20-week moving average, with strong 40-week and 52-week participation as well. That is not narrow leadership.

The one caution is that Financials slightly lagged SPY over four weeks, but the 12-week relative trend remains strong.

CRI read-through

Financials remain the cleanest sector leadership signal. This supports the idea that credit, capital markets, banks, insurers, and rate-sensitive financial activity are still participating in the expansion phase.

What would change the view

A break in breadth, deterioration in credit spreads, falling yield-curve confidence, or a move from Leader into Pullback Leader would reduce conviction.

Industrials

Current rank: 2
Model action: Accumulate
Rotation quadrant: Pullback Leader
Structural phase: Expansion
Sector score: 73.7

Industrials remain one of the healthiest sectors in the model.

The sector is still ranked second, with strong trend, strong relative strength, and excellent medium-term breadth. All constituents are above the 40-week and 52-week moving averages in the model, while 85.7% remain above the 20-week.

The sector’s quadrant classification is Pullback Leader, which is generally constructive. It suggests Industrials are still part of the leadership structure but are digesting gains or consolidating relative momentum.

CRI read-through

Industrials remain a key confirmation for the AI infrastructure, power, grid, automation, reshoring, construction, and equipment-cycle themes. This sector still fits the “real economy beneath the AI trade” framework.

What would change the view

A continued loss of relative strength versus SPY, deterioration in breadth, or movement into a lagging quadrant would weaken the infrastructure-cycle confirmation.

Information Technology

Current rank: 3
Model action: Accumulate
Rotation quadrant: Leader
Structural phase: Late Cycle
Sector score: 69.7

Technology remains a top-three sector and is now back in the Leader quadrant.

The trend score is the strongest in the model at 100, and Technology is the best-performing sector over four weeks, up roughly 8.2%. It also outperformed SPY by approximately 3.8 percentage points over four weeks.

That confirms continued capital flow into the technology leadership complex.

The caution is volatility. Technology has one of the weakest volatility scores in the model, which means leadership is strong but not low-risk. This is still a high-expectation sector.

CRI read-through

Technology remains constructive and continues to support the AI infrastructure and semiconductor leadership framework. However, the late-cycle classification and weak volatility score mean this is not a low-risk accumulation zone. It is leadership, but leadership that must continue delivering.

What would change the view

A break in trend, loss of four-week relative strength, or deterioration in breadth would matter quickly because expectations remain high.

Health Care

Current rank: 4
Model action: Accumulate
Rotation quadrant: Pullback Leader
Structural phase: Expansion
Sector score: 65.2

Health Care remains constructive.

The sector is ranked fourth and continues to show strong 12-week relative strength versus SPY. It also retains a very strong relative-strength score and acceptable breadth, with more than 83% of constituents above the 20-week and 40-week moving averages.

The score did weaken from last week, but the sector remains in the upper leadership group.

CRI read-through

Health Care is acting as a quality stabilizer inside a still-risk-on market. It is not leading the way Energy did this week or Technology did over four weeks, but it remains strong enough to be included in the Accumulate group.

What would change the view

Further score deterioration, loss of breadth, or movement from Pullback Leader into Laggard would indicate defensive leadership is failing rather than stabilizing.

Energy

Current rank: 5
Model action: Hold
Rotation quadrant: Leader
Structural phase: Late Cycle
Sector score: 64.9

Energy was the most important mover this week.

The sector jumped from rank 10 to rank 5 and improved its score by nearly 25 points, the largest weekly improvement in the model. Energy gained approximately 7.7% for the week and outperformed SPY by roughly 12.1 percentage points over six weeks.

Breadth is extremely strong. The model shows 100% of constituents above the 20-week, 40-week, and 52-week moving averages.

The reason Energy remains a Hold rather than Accumulate or Overweight is likely tied to volatility and late-cycle classification. The sector’s volatility score is weak even though price and breadth improved sharply.

CRI read-through

Energy is back on the rotation screen. This matters for the broader inflation, real-asset, power, utility-fuel, data-center-energy, and commodity-cycle discussion. The move is strong enough to respect, but the late-cycle and volatility flags argue against chasing aggressively after a sharp weekly move.

What would change the view

If Energy holds leadership for another week or two while volatility improves, it could move from tactical improvement to durable leadership. If the move reverses quickly, this may have been a sharp catch-up rally rather than a regime shift.

Materials

Current rank: 6
Model action: Hold
Rotation quadrant: Pullback Leader
Structural phase: Expansion
Sector score: 52.3

Materials remain in the middle of the model.

The sector is still in a Pullback Leader quadrant, which is constructive, but the score is much weaker than the top-five sectors. Breadth is mixed, with only 50% of constituents above the 20-week moving average, while 40-week and 52-week participation are better.

Materials are close to SPY over four and 12 weeks, but the sector is not showing the same confirmation as Industrials or Energy.

CRI read-through

Materials remain a watchable but not leading sector. Copper, chemicals, metals, construction inputs, and industrial commodities still matter to the infrastructure cycle, but the sector model is not yet showing enough strength to upgrade the group.

What would change the view

A breadth improvement above the 20-week moving average and stronger relative performance versus SPY would make Materials more attractive.

Real Estate

Current rank: 7
Model action: Hold
Rotation quadrant: Laggard
Structural phase: Late Cycle
Sector score: 47.6

Real Estate improved modestly this week, moving from Watch to Hold, but it remains in the Laggard quadrant.

The sector is still negative versus SPY over four, six, and 12 weeks. It is not showing broad leadership. However, the score improvement and respectable medium-term breadth keep it from falling into the weakest bucket.

This is an important distinction. Real Estate is not leading, but it is also not collapsing in the model.

CRI read-through

Real Estate remains selective. Data-center real estate is a different theme than broad REIT exposure. The sector ETF does not fully capture the AI data-center winners, so the model’s Real Estate weakness should not be interpreted as a rejection of the data-center infrastructure thesis.

What would change the view

Improved relative strength and a move out of the Laggard quadrant would be needed before broad Real Estate becomes more interesting.

Consumer Staples

Current rank: 8
Model action: Watch
Rotation quadrant: Laggard
Structural phase: Late Cycle
Sector score: 39.5

Consumer Staples remain weak.

The sector has poor breadth, with only one-third of constituents above the 20-week moving average. Relative strength remains negative over four, six, and 12 weeks.

Staples did gain for the week, but the model is not confirming durable leadership.

CRI read-through

Staples are not acting like a major defensive rotation leader. That reduces the probability that the market has fully moved into a broad risk-off phase.

What would change the view

Improving breadth and sustained relative strength would signal a more defensive market rotation.

Consumer Discretionary

Current rank: 9
Model action: Watch
Rotation quadrant: Laggard
Structural phase: Expansion
Sector score: 33.3

Consumer Discretionary remains a weak spot.

The sector lost approximately 1.4% for the week, making it one of the weakest short-term performers in the model. It is negative versus SPY over four, six, and 12 weeks. The score also deteriorated significantly from last week.

This is important because a healthy expansion often benefits discretionary spending. The fact that the sector remains weak suggests the market is not fully embracing a consumer-led growth phase.

CRI read-through

The consumer side of the market remains suspect. This argues for staying more focused on infrastructure, financials, industrials, AI, energy, and selected health care rather than broad consumer cyclicals.

What would change the view

Consumer Discretionary needs stronger breadth and positive relative strength before it becomes an accumulation candidate.

Utilities

Current rank: 10
Model action: Watch
Rotation quadrant: Laggard
Structural phase: Distribution
Sector score: 25.2

Utilities remain weak despite a positive one-week return.

The sector gained about 1.6% for the week, but it remains deeply negative versus SPY over four, six, and 12 weeks. Breadth is also poor, with only 20% of constituents above the 20-week moving average.

This is an important signal. The market is not yet rewarding broad defensive yield sectors, even though power infrastructure remains a major CRI theme.

That distinction matters.

Broad Utilities exposure is not the same as owning the specific power-infrastructure beneficiaries tied to data centers, grid expansion, transformers, switchgear, turbines, electrical equipment, and energy management.

CRI read-through

Utilities are still a laggard sector. The AI power thesis should remain targeted toward infrastructure and equipment beneficiaries, not broad utility exposure unless the sector model improves.

What would change the view

A move out of Distribution and improving relative strength would be required before Utilities become more attractive.

Communication Services

Current rank: 11
Model action: Avoid
Rotation quadrant: Laggard
Structural phase: Accumulation
Sector score: 22.1

Communication Services remains the weakest sector in the model.

The sector is ranked last, has the weakest relative-strength and trend scores, and remains negative versus SPY over four, six, and 12 weeks. It gained for the week, but the broader structure remains poor.

The Accumulation structural phase is interesting, but the model has not yet confirmed that accumulation is translating into leadership.

CRI read-through

Communication Services remains an avoid in the sector model. This is important because many mega-cap platform names sit here, but the sector as a whole is not confirming leadership.

What would change the view

A move from Laggard into Improving, better trend, and stronger breadth would be needed before the sector returns to the investable leadership stack.

Weekly Movers

Biggest positive change

Energy was the clear winner.

It improved by nearly 25 model points, jumped from rank 10 to rank 5, and moved into the Leader quadrant.

This is the sector to watch next week.

The key question is whether Energy becomes durable leadership or simply a sharp tactical catch-up move.

Biggest negative change

Consumer Discretionary saw the largest score deterioration among the major sectors, falling more than 8 model points.

That keeps the consumer-spending side of the market weak and argues against broad discretionary exposure.

Most important upgrade

Financials moved from Accumulate to Overweight while holding the top rank.

That is the strongest allocation signal in this week’s model.

Most important caution

Technology remains strong, but volatility risk is elevated.

The trend score is excellent, and four-week relative strength is positive. But the weak volatility score means this remains a high-expectation leadership group.

CRI Allocation View

Overweight

Financials

Financials are the cleanest leadership sector in the model. The sector has the strongest score, strong breadth, a Leader quadrant designation, and an Expansion structural phase.

Accumulate

Industrials
Information Technology
Health Care

Industrials remain a key infrastructure-cycle confirmation. Technology continues to lead on trend and relative performance. Health Care provides a constructive quality/stability layer.

Hold

Energy
Materials
Real Estate

Energy is the most improved sector but needs confirmation after a sharp move. Materials remain middle-of-pack but constructive. Real Estate improved but remains a laggard, so broad exposure should remain selective.

Watch

Consumer Staples
Consumer Discretionary
Utilities

These sectors remain weak relative to the leadership stack. Utilities and Staples are not confirming a broad defensive rotation. Consumer Discretionary remains the weakest consumer signal.

Avoid

Communication Services

The model keeps Communication Services at the bottom of the ranking structure.

Sector Rotation Theme Map

Strongest current themes

Financial leadership
Banks, insurers, capital markets, and credit-sensitive groups remain favored by the model.

Industrial infrastructure
Industrials continue to support the AI buildout, grid expansion, automation, defense, reshoring, and capital-equipment cycle.

Technology trend leadership
Technology remains a top sector, with the strongest trend score and best four-week performance.

Energy resurgence
Energy re-entered the leadership conversation with the largest weekly improvement.

Health Care stabilization
Health Care remains a constructive upper-tier sector, acting as a quality stabilizer.

Weakest current themes

Broad consumer exposure
Consumer Discretionary remains weak and deteriorating.

Traditional defensives
Staples and Utilities remain laggards despite short-term bounces.

Mega-cap platform sector exposure
Communication Services remains the weakest sector in the model.

Broad Real Estate
Real Estate is improving but still not leading. Data-center real estate remains a separate, more targeted theme than broad REIT exposure.

CRI Bottom Line

This week’s sector rotation model remains constructive, but selective.

The strongest allocation signal is Financials moving to Overweight while holding the top rank.

The strongest rotation signal is Energy, which jumped from rank 10 to rank 5 and moved into the Leader quadrant.

The strongest structural confirmation remains Industrials plus Technology, which continues to support the infrastructure and AI-adjacent leadership thesis.

The weakest signals remain Communication Services, Utilities, Consumer Discretionary, and Consumer Staples.

The key conclusion:

This is still not a broad defensive market. Leadership remains concentrated in Financials, Industrials, Technology, Health Care, and now an improving Energy sector.

The market continues to favor:

Financial leadership
Infrastructure and industrial exposure
Technology trend leadership
Selective health care
Energy catch-up potential

The market continues to avoid:

Weak consumer cyclicals
Traditional defensives without relative strength
Broad communication services weakness
Unconfirmed real estate exposure

The CRI read for the week:

Stay with leadership, respect the Energy improvement, avoid broad defensive conclusions, and keep watching whether Technology leadership broadens or becomes more volatile.

This week delivered one of the strongest cross-stack confirmations of the AI infrastructure cycle we have seen this summer.

The evidence did not come from one layer.

It came from several.

CoreWeave pushed its backlog above $104 billion. Nebius reported AI-cloud revenue growth above 500% and expects more than $9 billion of customer prepayments this year. Cisco disclosed $9.3 billion of fiscal-2026 AI infrastructure orders. Supermicro entered fiscal 2027 with record backlog after receiving more than $60 billion of new orders in its June quarter. Lumentum’s quarterly revenue more than doubled. Coherent reported strong earnings and a stronger outlook. TSMC’s July sales rose 45% year over year. Foxconn said AI servers and cloud/networking products exceeded half of quarterly revenue for the first time.

That is not a narrow GPU story.

That is a full-stack infrastructure confirmation.

The weekly CRI conclusion is simple:

The AI infrastructure cycle remains healthy, but the market’s next sorting mechanism is becoming funding quality. Backlog matters. Funded backlog matters even more.

Compute / GPUs

Current health: Very Strong
Direction: Accelerating

The clearest compute confirmation this week came from the companies actually deploying and renting AI compute capacity.

CoreWeave reported second-quarter revenue of $2.58 billion, up more than 100% year over year. More importantly, revenue backlog reached $104.2 billion, up from $99.4 billion in the prior quarter. The company also indicated that it added more than $25 billion of new customer commitments early in the third quarter.

Nebius provided another major confirmation. AI-cloud revenue grew more than 500% year over year, and the company expects to receive more than $9 billion of customer prepayments this year. That is important because prepayments move the signal from “customers are interested” to “customers are helping fund future capacity.”

Supermicro added the hardware-level confirmation. The company received more than $60 billion of new orders in its June quarter, pushing backlog to record levels heading into fiscal 2027.

Why this layer is accelerating

Compute demand remains well ahead of available infrastructure capacity. The strongest evidence is no longer only GPU enthusiasm. It is backlog, customer commitments, prepayments, and system-level orders.

What CRI is watching

NVIDIA’s August earnings, Rubin deployment timing, CoreWeave capacity activation, Nebius power deployment, Supermicro backlog conversion, and whether compute utilization remains strong enough to support the financing burden.

Custom Silicon / XPUs

Current health: Very Strong
Direction: Improving

There was less direct custom-accelerator earnings data this week, but the surrounding evidence continues to support the custom-silicon layer.

The AI infrastructure market is becoming more heterogeneous. The next phase is not simply NVIDIA GPUs versus everything else. It is NVIDIA GPUs, AMD accelerators, hyperscaler ASICs, custom XPUs, CPUs, DPUs, switching silicon, memory, optics, and software architecture all working together inside increasingly complex AI systems.

TSMC’s July revenue rose approximately 45% year over year, reinforcing that advanced AI and high-performance computing demand remains strong at the foundry level.

Foxconn also gave an important systems signal. Its cloud and networking products segment, which includes AI servers, represented 51% of second-quarter revenue, exceeding half of company revenue for the first time. Foxconn also said it is preparing NVIDIA Vera Rubin server racks for mass-production preparation in the third quarter, with shipments expected to begin in the fourth quarter.

Why this layer is improving

AI infrastructure is moving from single-chip purchasing toward workload-optimized systems. That supports custom accelerators, advanced networking silicon, high-speed interconnect, and specialized architectures built around cost, power, performance, and availability.

What CRI is watching

Broadcom’s next AI custom-accelerator update, Marvell design ramps, hyperscaler internal silicon programs, TSMC advanced-node availability, and whether custom silicon converts into durable production volume.

Memory / HBM

Current health: Very Strong
Direction: Improving

Memory remains one of the cleanest structural bottlenecks in the AI stack.

The key memory evidence remains strong: customers are increasingly treating HBM and advanced memory as strategic capacity rather than normal commodity inventory. Multiyear agreements, supply reservations, and customer commitments suggest buyers still expect constrained availability.

The new signal this week came from the surrounding stack. Compute demand is rising. AI servers are increasing as a percentage of Foxconn revenue. TSMC’s AI-driven revenue remains strong. CoreWeave and Nebius are signing large capacity commitments. All of that pulls demand into HBM, server DRAM, storage, and memory packaging.

There is also an important architectural nuance. If future accelerator platforms reduce HBM content per GPU but enable more GPUs to be produced, total memory demand may remain very strong. The memory cycle should be judged by aggregate AI system demand, not only memory content per device.

Why this layer remains very strong

AI capacity requires memory bandwidth, memory capacity, storage, and packaging. The demand signal is being confirmed by cloud contracts, server orders, and next-generation rack systems.

What CRI is watching

HBM4 qualification, Micron pricing, SK hynix capacity expansion, Samsung qualification, CoWoS availability, customer deposits, and 2027 supply contracts.

Networking

Current health: Very Strong
Direction: Accelerating

Networking received one of the strongest confirmations of the week.

Cisco reported $9.3 billion of AI infrastructure orders during fiscal 2026, ahead of earlier targets. The company also pointed to strong demand tied to the AI infrastructure buildout and hyperscaler data center construction.

This matters because networking is no longer a secondary support layer. As clusters scale, network efficiency determines whether expensive accelerators remain productive or sit idle waiting for data.

The signal follows Arista’s recent strength and its rollout of higher-speed AI fabric platforms. The market is now seeing AI networking demand arrive in multibillion-dollar increments.

Why this layer is accelerating

Large AI clusters require higher-bandwidth switching, lower latency, stronger fabrics, and more efficient east-west traffic movement. The bigger the cluster, the more the network becomes part of the compute system itself.

What CRI is watching

1.6T ramps, 102.4T switching adoption, Ethernet scale-up architectures, Cisco hyperscaler revenue conversion, Arista order growth, and networking power density.

Optics / Photonics

Current health: Very Strong
Direction: Accelerating

Optics may be the strongest layer of the week.

Lumentum reported quarterly revenue of approximately $1.01 billion, up more than 100% year over year, and guided the next quarter above prior expectations. Management pointed to advanced 1.6T transceivers and high-power lasers as important demand drivers.

Coherent also delivered strong results and a strong outlook. The company reported earnings above expectations and guided fiscal-first-quarter revenue to approximately $2.3 billion at the midpoint. Its data-center and communications exposure remains one of the key ways to track optical demand tied to AI infrastructure.

The migration from copper toward optical connectivity is now showing up in revenue, product ramps, capacity planning, and market leadership.

Why this layer is accelerating

AI clusters need more bandwidth, more fiber, more optical engines, more transceivers, more testing, and eventually more near-packaged and co-packaged optical architectures. This is the nervous system of AI factories.

What CRI is watching

1.6T volumes, CPO and NPO production ramps, Coherent manufacturing expansion, Lumentum margins, optical-engine supply, customer concentration, and whether elevated valuations begin outrunning earnings.

Power Infrastructure

Current health: Very Strong but Severely Bottlenecked
Direction: Accelerating

Power remains the hardest physical constraint in the AI infrastructure stack.

The compute providers are signing contracts faster than utilities can add generation, transmission, substations, and interconnection capacity. CoreWeave raised its 2026 capex outlook as it expands infrastructure to satisfy customer demand. Nebius is also focused on scaling available power capacity as it signs larger AI-cloud agreements.

The simple issue is this:

AI companies can order GPUs faster than the physical grid can absorb them.

That keeps power infrastructure near the center of the CRI bottleneck framework.

Why this layer is accelerating

Data center demand is now being measured in gigawatts. That pulls capital into turbines, transformers, switchgear, electrical distribution, substations, onsite generation, backup power, grid services, and energy management.

What CRI is watching

Gas-turbine orders, transformer lead times, utility interconnections, hyperscaler PPAs, onsite generation, behind-the-meter projects, and deployment delays caused by unavailable power.

Cooling / Thermal Management

Current health: Very Strong
Direction: Improving

Cooling did not receive a major standalone earnings update this week, but the surrounding evidence continues to reinforce the layer.

Supermicro’s record order intake, Foxconn’s AI server growth, Vera Rubin rack preparation, and the continued ramp of 1.6T networking all point toward higher rack density and higher thermal complexity.

The basic relationship remains unchanged:

More installed GPU capacity plus denser racks plus faster networking equals more thermal-management intensity per data center.

That supports liquid cooling, heat rejection, coolant distribution, rear-door heat exchangers, chillers, pumps, and integrated power-and-thermal systems.

Why this layer remains very strong

Cooling is becoming a deployment enabler. Without sufficient thermal management, AI capacity cannot be installed, utilized, or operated efficiently.

What CRI is watching

Vertiv backlog, Modine data-center growth, liquid-cooling penetration, coolant-distribution units, heat exchangers, commissioning capacity, and whether margins hold as projects become larger and more complex.

Data Centers / Real Estate

Current health: Very Strong
Direction: Accelerating

The neocloud earnings provided exceptionally strong confirmation.

CoreWeave ended Q2 with $104.2 billion of backlog and added more than $25 billion of additional commitments early in the third quarter.

Nebius reported AI-cloud revenue growth above 500% and expects more than $9 billion of customer prepayments this year. That is a very important signal because it means customers are not merely reserving capacity—they are providing capital to help ensure that capacity is built.

Customer prepayments are one of the cleanest ways to separate high-quality demand from speculative capacity.

The industry is moving from:

“We plan to build capacity.”

to:

“Customers are funding the capacity before it is installed.”

That is a much stronger demand signal.

Why this layer is accelerating

AI capacity demand is turning into backlog, prepayments, long-duration contracts, and infrastructure commitments.

What CRI is watching

Customer prepayments, signed capacity contracts, powered-site premiums, lease structures, commissioning timelines, utilization, cancellations, and the split between powered and unpowered land.

Advanced Packaging / Substrates

Current health: Very Strong
Direction: Improving

Advanced packaging remains one of the most important hidden bottlenecks.

TSMC’s July revenue growth reinforced strong AI and HPC demand. Foxconn also highlighted CoWoS availability as a key determinant of next year’s AI server rack market.

That is an unusually direct confirmation of the packaging bottleneck.

Server demand and manufacturing readiness can exist, but package availability can still determine the final output rate.

The key point is:

A finished GPU wafer is not a deployable AI system. Packaging capacity increasingly determines how quickly silicon becomes revenue-producing infrastructure.

Why this layer is improving

Rubin, HBM4, custom ASICs, 1.6T networking, chiplets, high-density substrates, and thermal interfaces all increase packaging complexity.

What CRI is watching

CoWoS capacity, hybrid bonding, substrates, HBM integration, package yields, advanced test, glass substrates, and Rubin rack production.

Financing / Capex Returns

Current health: Watch Carefully
Direction: Mixed — risk increasing alongside demand

Financing remains the most important risk layer.

CoreWeave demonstrates both sides of the AI infrastructure cycle. Demand is extraordinary, with more than $104 billion of backlog and new commitments continuing to arrive. But the company also raised its capex plan as it expands to serve that demand.

Nebius provides a potentially cleaner funding signal because it expects more than $9 billion of customer prepayments this year. That shifts part of the financing burden from the provider to the customer.

This distinction matters.

The market is increasingly separating two models.

Higher-quality financing

Customer prepayments
Take-or-pay contracts
Long-duration committed demand
Rapid capacity activation
High utilization
Clear path to revenue conversion

Higher-risk financing

Debt-funded speculative capacity
Long construction timelines
Unpowered projects
Backlog without customer funding
Aggressive residual-value assumptions
Delayed utilization

The next major AI correction is more likely to originate from financing quality than from a sudden disappearance of compute demand.

Why this layer remains mixed

Demand is strong, but capital intensity keeps rising. The market wants evidence that backlog can be converted into profitable revenue without overloading the balance sheet.

What CRI is watching

Customer prepayments, debt issuance, lease obligations, free cash flow, capex revisions, utilization, depreciation, contract duration, and return on invested capital.

CRI Market Rotation Read

This week gave a stronger market confirmation than recent AI infrastructure updates.

CoreWeave, Nebius, Supermicro, Micron, and other AI infrastructure names rallied sharply after the latest earnings evidence, while the Philadelphia Semiconductor Index also gained.

That matters because the infrastructure trade had recently experienced a significant valuation reset. The new rally looks less like blind momentum and more like fundamental re-confirmation after correction.

The strongest new market-supported layers are:

Optics / Photonics
Networking
AI Cloud / Data Centers
Server Infrastructure
Advanced Packaging

The market is recognizing that AI demand has not disappeared.

But the leadership test is becoming more specific.

The market is likely to reward:

Funded backlog
Customer prepayments
Visible utilization
Near-term revenue conversion
Clear margin expansion
Real deployment capacity

And discount:

Unfunded backlog
Unpowered land
Speculative capacity
Long construction timelines
High debt dependence
Weak free cash flow

CRI Bottom Line

This was a strong week for the AI infrastructure thesis.

The strongest confirmations came from:

CoreWeave: $104.2 billion backlog and another $25 billion-plus of commitments.
Nebius: AI-cloud revenue growth above 500% and more than $9 billion of expected customer prepayments.
Supermicro: more than $60 billion of new orders and record backlog.
Cisco: $9.3 billion of fiscal-2026 AI infrastructure orders.
Lumentum: quarterly revenue more than doubled as 1.6T optics accelerate.
Coherent: strong earnings and a stronger revenue outlook.
TSMC / Foxconn: AI production continues ramping, while CoWoS remains a physical constraint.

The strongest layers now are:

Very Strong / Accelerating

Compute / GPUs
Networking
Optics / Photonics
Data Centers / Real Estate

Very Strong / Improving

Custom Silicon / XPUs
Memory / HBM
Cooling / Thermal Management
Advanced Packaging / Substrates

Very Strong but Severely Bottlenecked

Power Infrastructure

Watch Carefully

Financing / Capex Returns

The core CRI conclusion:

The AI bottleneck trade is broadening again. Compute demand is pulling networking, optics, packaging, power and cooling higher with it.

But the next phase is not simply about who has the most backlog.

It is about who has funded backlog.

Customer prepayments, take-or-pay commitments, available power, production capacity and rapid deployment are becoming increasingly valuable differentiators.

The AI buildout remains healthy. The market’s next major sorting mechanism will be financing quality.

Tags

AI Infrastructure, NVIDIA, CoreWeave, Nebius, Supermicro, Cisco, Micron, TSMC, Foxconn, Lumentum, Coherent, Broadcom, Marvell, Vertiv, HBM, AI Networking, Optical Networking, Photonics, 1.6T, Data Centers, Power Infrastructure, Liquid Cooling, Advanced Packaging, CoWoS, AI Capex, Backlog, Customer Prepayments, Market Rotation, Capital Regime Intelligence

Short Substack Note

AI infrastructure delivered one of its strongest cross-stack confirmations of the summer.

CoreWeave now has more than $104 billion of backlog. Nebius reported AI-cloud growth above 500% and expects more than $9 billion of customer prepayments. Supermicro received over $60 billion of new orders. Cisco disclosed $9.3 billion of AI infrastructure orders. Lumentum and Coherent delivered another major confirmation of the optical bottleneck.

And this time the market responded.

The AI infrastructure cycle remains healthy, but the next distinction is becoming clear:

Backlog matters. Funded backlog matters even more.

Suggested Graphic Direction

Title:
Weekly AI Layer Health Report — August 13, 2026

Primary callout:
AI infrastructure reaccelerates: backlog, optics and networking confirm the buildout.

Layer status

Very Strong — Accelerating

Compute / GPUs
Networking
Optics / Photonics
Data Centers / Real Estate

Very Strong — Improving

Custom Silicon / XPUs
Memory / HBM
Cooling / Thermal Management
Advanced Packaging / Substrates

Very Strong but Severely Bottlenecked

Power Infrastructure

Watch Carefully — Mixed

Financing / Capex Returns

Evidence callouts for the graphic

$104.2B — CoreWeave backlog
$25B+ — new CoreWeave commitments
+514% — Nebius AI-cloud revenue
$9B+ — Nebius expected customer prepayments
$60B+ — Supermicro Q4 new orders
$9.3B — Cisco FY26 AI infrastructure orders
+109% — Lumentum quarterly revenue growth
$2.3B midpoint — Coherent fiscal-Q1 revenue outlook

CRI Bottom Line

The AI infrastructure thesis strengthened this week.

Networking and optics are now showing the same operating evidence that GPUs and HBM showed earlier in the cycle.

The next differentiator is funding quality: customer-prepaid capacity should command a premium over debt-funded speculative capacity.

Executive Read

Today may be one of the most useful AI-infrastructure earnings days of the quarter.

In a single session, earnings gave us direct confirmation from five different layers:

Optics → Coherent
Networking → Cisco
AI Cloud Capacity → Nebius
Alternative AI Compute → Cerebras
Advanced Batteries / Wearables / Drones → Enovix

The message across these companies is remarkably consistent:

AI demand is not narrowing. It is propagating farther into the physical infrastructure stack.

Even more important, several companies delivered strong numbers and strong guidance yet saw muted or negative initial stock reactions. That suggests the fundamental AI cycle remains powerful while investor expectations have become extremely demanding.

Coherent — COHR

Sector: AI Optics / Photonics / Data-Center Connectivity

Expectations vs. Actuals

Adjusted EPS: $1.74 vs. $1.62 expected — Beat by $0.12

Revenue: $2.05B vs. approximately $1.99B expected — Beat by ~$60M

Revenue increased approximately 34% year over year, while non-GAAP gross margin expanded to 40.2%.

Earnings Call Notes

The most important message from Coherent was the continued transition from electrical connectivity toward optics inside AI data centers.

Management described:

• exceptional customer demand,

• accelerating production-capacity expansion,

• increasing adoption of optics over copper,

• and several new growth platforms beginning to ramp.

This provides an immediate confirmation of the signal Lumentum delivered yesterday.

The optical infrastructure cycle is not being driven by one company.

It is increasingly industry-wide.

Guidance

For fiscal Q1 2027:

Revenue: $2.2B–$2.4B

Adjusted EPS: $1.85–$2.05

Both ranges were ahead of Wall Street expectations.

Headwinds

• Elevated investor expectations

• Rapid capacity expansion requirements

• Inventory growth

• Approximately $3.2B of debt

• Optical supply-chain constraints

Tailwinds

• 800G and 1.6T connectivity

• AI cluster scaling

• Datacenter optics

• High-power lasers

• Silicon photonics

• Copper-to-optical architecture transition

CRI Take

This is one of the strongest confirmations of the week.

Lumentum yesterday + Coherent today materially strengthen the optical bottleneck thesis.

AI compute can only scale if bandwidth scales alongside it.

The progression increasingly looks like:

GPU → Networking → Optical Transceiver → Laser → Photonic Integration

Optics is becoming its own infrastructure cycle.

Cisco — CSCO

Sector: AI Networking / Data-Center Infrastructure

Expectations vs. Actuals

Adjusted EPS: $1.22 vs. $1.17 expected — Beat by $0.05

Revenue: $17.25B vs. approximately $16.84B expected — Beat by ~$410M

Revenue increased approximately 18% year over year.

Earnings Call Notes

The most important number may have been AI infrastructure orders.

Cisco received approximately:

$4B of AI-related orders during Q4

bringing fiscal-year AI infrastructure orders to approximately:

$9.3B

Management characterized the environment as a networking supercycle driven by hyperscalers and increasingly by enterprise AI deployments.

Product revenue increased approximately 24%, with particularly strong demand across networking infrastructure.

Guidance

Fiscal Q1 2027:

Revenue: $18.0B–$18.2B

Adjusted EPS: $1.32–$1.34

Fiscal 2027:

Revenue: $72.2B–$73.4B

Adjusted EPS: $5.05–$5.11

Management expects AI infrastructure revenue to continue expanding significantly in FY27.

Headwinds

• Gross-margin pressure

• Tariffs

• Elevated expectations

• Workforce restructuring

• Enterprise spending cyclicality

Tailwinds

• Hyperscaler AI networking

• Data-center switching

• Enterprise AI infrastructure

• Silicon and optics integration

• Campus networking refresh

CRI Take

Cisco confirms another piece of the same puzzle:

Compute demand is creating a networking supercycle.

The AI cluster does not end at the GPU.

It requires increasingly sophisticated switching, routing, security and optical connectivity.

The fact that both Cisco and Coherent delivered strong results on the same day strengthens the broader infrastructure interpretation substantially.

Nebius — NBIS

Sector: AI Cloud / Neocloud / Powered Compute Infrastructure

Expectations vs. Actuals

EPS: approximately -$0.68 vs. -$0.72 expected — Better by $0.04

Revenue: $582.3M vs. approximately $570M expected — Beat by ~$12M

Revenue increased approximately 454% year over year.

Nebius AI Cloud alone generated approximately $575M, growing more than sixfold.

Earnings Call Notes

The most striking part of the call was not the quarterly beat.

Nebius disclosed four new AI-cloud contracts worth more than $1B each.

Management indicated that roughly 70% of Q2 deal value included prepayments, and Nebius expects more than:

$9B of customer prepayments during 2026.

Adjusted EBITDA reached approximately $236M, compared with expectations near $169M.

Management believes essentially all 2027 capacity could be sold at current commercial terms.

Power & Capacity

Nebius increased its contracted-power ambitions toward approximately:

5 GW

and expects to deploy more than 1 GW annually beginning in 2027.

That matters enormously to the CRI framework.

Nebius is not merely buying GPUs.

It is securing:

• power,

• land,

• interconnection,

• data-center capacity,

• cooling,

• networking,

• and financing.

Headwinds

• Very high capital expenditures

• Customer concentration

• Debt and financing requirements

• Power availability

• Construction execution

Tailwinds

• AI compute scarcity

• Multi-billion-dollar contracts

• Customer prepayments

• Strong pricing environment

• Rapid capacity expansion

CRI Take

Nebius reinforces the same signal we saw from CoreWeave yesterday:

The scarce asset is increasingly deployable AI capacity.

That means the beneficiaries underneath the neoclouds deserve increasing attention:

power → transformers → switchgear → cooling → networking → construction.

Cerebras Systems — CBRS

Sector: Alternative AI Compute / Inference Infrastructure

Expectations vs. Actuals

Using the company’s core operating presentation:

Core Revenue: approximately $209.9M vs. ~$194M expected — Beat

Adjusted EPS: approximately -$0.05 vs. -$0.17 expected — Better than expected

GAAP EPS was approximately -$2.98, heavily affected by stock-based compensation.

Core revenue increased approximately 103% year over year.

Earnings Call Notes

Cerebras continues positioning its wafer-scale architecture as an alternative to traditional GPU-based inference.

Management highlighted:

• accelerating inference demand,

• major customers including OpenAI and AWS,

• manufacturing expansion,

• and increasing production through TSMC, Flex, Sanmina and other partners.

The company is targeting approximately:

600 MW of data-center capacity

to support future deployments.

Guidance

Q3 core revenue:

Approximately $215M vs. ~$210M expected

Full-year core revenue guidance:

$880M–$890M

Management also continues targeting very aggressive longer-term growth.

Headwinds

• Heavy losses

• Stock-based compensation

• Capital requirements

• Nvidia ecosystem dominance

• Manufacturing execution

Tailwinds

• AI inference growth

• Alternative accelerator architectures

• Large-model customers

• Growing capacity

• Compute diversification

CRI Take

Cerebras adds an important wrinkle to the AI thesis.

The infrastructure cycle does not necessarily require every workload to run on the same architecture.

As inference grows, specialized architectures may gain room alongside GPUs.

That potentially expands rather than contracts the infrastructure opportunity.

Enovix — ENVX

Sector: Advanced Batteries / Smart Glasses / Drones / Defense

Expectations vs. Actuals

Adjusted EPS: -$0.13 vs. approximately -$0.15 expected — Beat by ~$0.02

Revenue: $9.0M vs. approximately $8.4M expected — Beat by ~$0.6M

Revenue increased 21% year over year.

The absolute revenue remains small, but the commercial milestones are more important than the accounting scale at this stage.

Earnings Call Notes

Three developments stood out.

Smart Eyewear

Enovix shipped approximately:

2,100 AI-1 batteries in Q2

and expects Q3 shipments to increase roughly 9x to approximately 19,000 packs.

Those shipments are part of a 50,000-pack 2026 order.

The smart-glasses connection is particularly relevant as AI moves into wearable devices where battery energy density becomes increasingly important.

Smartphones

Its lead smartphone customer has confirmed more than 1,000 cycles under testing, with the final accelerated cycle-life qualification test expected to conclude in 2026.

Drones & Defense

The drone, defense and industrial pipeline increased approximately 41% sequentially to $183M.

Drone opportunities alone exceeded:

$100M

Management believes demand for high-performance drone batteries meeting U.S. sourcing requirements may exceed supply for years.

Guidance

Q3 revenue:

$9M–$10M

Management continues emphasizing commercialization and manufacturing scale rather than near-term profitability.

Headwinds

• Early-stage commercial scale

• Manufacturing yields

• Qualification timing

• Cash burn

• Production ramp risk

Tailwinds

• AI smart glasses

• Smartphone qualification

• Defense drones

• Silicon-anode energy density

• U.S.-aligned battery supply chain

CRI Take

Enovix is not an AI-data-center company.

But it belongs in the expanded AI universe.

AI is moving from the data center into:

wearables → drones → robotics → edge devices.

Those systems increasingly require improved energy density.

The battery layer therefore becomes another second-order AI infrastructure opportunity.

CRI Cross-Company Intelligence

1. Optics Has Now Been Confirmed Twice in 24 Hours

Lumentum + Coherent

Both companies reported:

• strong revenue growth,

• expanding margins,

• strong forward guidance,

• and accelerating AI optical demand.

That materially increases confidence in the optical infrastructure thesis.

Key CRI watch names include:

LITE, COHR, FN, AAOI, CRDO, MRVL, TSEM and POET.

2. Networking Is Now Joining the Supercycle

Cisco’s approximately $9.3B of FY26 AI orders makes clear that networking is no longer a secondary beneficiary.

As AI clusters grow:

more GPUs → more east-west traffic → more switches → more optical links.

Compute and networking are scaling together.

3. Power Capacity Is Becoming Scarce Infrastructure

Nebius targeting roughly 5 GW follows yesterday’s extraordinary power-related signals from CoreWeave and Riot.

That is a major CRI confirmation.

The AI buildout is becoming constrained not simply by processors, but by:

• megawatts,

• interconnection,

• transformers,

• switchgear,

• generation,

• and cooling.

This keeps the CRI power universe extremely important:

GEV, ETN, POWL, NVT, HUBB, PWR, VST, CEG, NRG and selected natural-gas infrastructure.

4. Cooling Is Becoming More Important, Not Less

Today’s reports did not require a cooling company to report earnings to strengthen the thesis.

Nebius, Cisco, Coherent and Cerebras collectively describe increasingly dense AI infrastructure.

Higher compute density means:

more electricity + more heat + more cooling infrastructure.

That continues strengthening the read-through toward:

MOD, VRT, TT, AAON, CARR and JCI.

5. AI Is Beginning to Move Beyond the Data Center

Enovix provides an early signal for the next layer:

AI at the edge.

Smart glasses, drones, robotics and autonomous systems require improved battery performance.

The infrastructure stack may ultimately extend:

Compute → Network → Optics → Power → Cooling → Edge Devices → Batteries

CRI Earnings Dashboard — August 12

VERY STRONG

COHR — AI Optics
Beat + strong guide + capacity expansion.

NBIS — AI Compute Capacity
454% revenue growth + multi-billion-dollar contracts.

CSCO — AI Networking
$9.3B of annual AI orders + strong FY27 guide.

STRONG / EARLY

ENVX — Batteries / Edge AI
Commercialization milestones and strong drone pipeline.

HIGH-GROWTH / HIGH-RISK

CBRS — Alternative AI Compute
Rapid core revenue growth but significant capital and execution risk.

CRI Bottom Line

Today’s earnings materially strengthen the AI supercycle thesis.

But the strongest signal is not simply that AI demand remains strong.

It is that the physical requirements of AI are expanding simultaneously across multiple layers.

Coherent confirms optical bandwidth.

Cisco confirms networking.

Nebius confirms scarce powered compute capacity.

Cerebras confirms demand for additional compute architectures.

Enovix begins extending the AI ecosystem toward batteries and edge devices.

The investment map increasingly looks like:

Compute → Networking → Optics → Power → Cooling → Edge Infrastructure

That is where the earnings data are pointing.

And that is increasingly where CRI should be looking for the next layer of capital rotation.

This week’s earnings provided another strong confirmation that the AI cycle remains intact — but the market is moving past the simple question of whether AI demand exists.

The more important questions now are:

Where is the demand showing up?
Who is monetizing it?
And which physical bottlenecks are becoming increasingly valuable as the buildout scales?

The answer is becoming clearer.

AI spending is increasingly flowing through:

Cloud Compute → Networking → Optics → Power → Cooling → Advanced Infrastructure

And the earnings signals across that stack remain unusually strong.

Oracle — AI Infrastructure Demand Is Enormous

Oracle — ORCL

Adjusted EPS: $2.11 actual vs. roughly $1.96 expected — Beat

Revenue: $19.2B actual vs. roughly $19.1B expected — Beat

Oracle delivered one of the most important AI infrastructure reports of the quarter.

Cloud Infrastructure revenue increased 93% year over year to $5.8B, while total cloud revenue reached $9.9B, up 47%.

Even more striking:

Remaining Performance Obligations reached $638B, up 363% year over year and another $85B sequentially. Much of that increase came from large AI contracts.

Oracle also guided FY2027 revenue to approximately $90B and raised non-GAAP EPS guidance.

The Important Caveat

The demand is extraordinary.

So is the capital required to satisfy it.

Oracle generated $32B in operating cash flow during FY2026, but free cash flow was negative $23.7B as the company continued building AI data-center capacity.

CRI Read

This may be the most important message from this earnings week:

AI demand is not the problem. Infrastructure capacity is.

That strengthens the investment thesis beneath Oracle:

• power generation,

• electrical infrastructure,

• cooling,

• networking,

• optical connectivity,

• and data-center construction.

Adobe — AI Monetization Is Becoming Measurable

Adobe — ADBE

Adjusted EPS: $5.96 actual vs. approximately $5.82 expected — Beat

Revenue: $6.62B actual vs. approximately $6.45B expected — Beat

Adobe reported record quarterly revenue, up approximately 13% year over year, while raising full-year revenue and non-GAAP EPS targets.

The more important number may be its AI business.

AI-first ARR more than tripled year over year and exceeded $500M.

Adobe is beginning to answer one of the major questions confronting software:

Can incumbent software companies turn AI from a competitive threat into a monetizable product layer?

So far, the answer is increasingly yes.

CRI Read

The software phase is changing.

The market is moving from:

AI narrative → AI adoption → AI monetization

The companies that can embed AI into existing high-value workflows should increasingly separate from those simply adding AI features without measurable revenue conversion.

The Optics Layer — One of the Strongest Confirmation Signals in the AI Stack

The important clarification here is that the major optical companies did not all report earnings this week.

But their most recent earnings remain critical to interpreting Oracle’s massive infrastructure backlog — and the market continued trading the optical complex as a unified AI-infrastructure basket this week.

That makes the optics layer too important to omit.

Applied Optoelectronics — Hyperscaler Optical Demand Is Accelerating

Applied Optoelectronics — AAOI

In its most recent quarter:

Revenue: $151.1M
Year-over-year growth: +51%

Most importantly, data-center revenue more than doubled to $81.4M.

AAOI also began volume shipments of its 800G transceivers to a major hyperscale customer and guided toward materially higher sequential growth as expanded manufacturing capacity comes online.

Management also highlighted demand around:

• 800G,

• 1.6T,

• and additional hyperscaler deployments.

CRI Read

AAOI is confirming that optical bandwidth requirements are scaling directly with AI compute.

This is not simply a semiconductor cycle.

It is increasingly a bandwidth cycle.

Lumentum — Optical Demand Is Moving Into Another Gear

Lumentum — LITE

Lumentum’s most recent fiscal Q3 revenue was $808.4M, nearly 90% above the prior-year quarter. Adjusted EPS was $2.37 vs. $2.27 expected.

The company has been benefiting from rapid adoption of:

• high-speed optical transceivers,

• lasers,

• optical switching,

• and emerging near-packaged / co-packaged optical architectures.

CRI Read

Lumentum remains one of the cleanest ways to monitor whether AI clusters continue getting larger and more interconnected.

As compute scales, electrical connectivity becomes increasingly inefficient.

That pushes the architecture toward optics.

Coherent — Datacenter & Communications Become the Engine

Coherent — COHR

Coherent’s latest quarter produced:

Revenue: $1.81B
Growth: +21% YoY
Adjusted EPS: $1.41

Its Datacenter & Communications business reached roughly $1.36B, or about 75% of total revenue, and grew approximately 41% year over year.

The company is also expanding indium-phosphide production to support growing laser demand.

CRI Read

This is another strong confirmation that optical infrastructure is becoming a capacity constraint.

The bigger AI clusters get, the more valuable the optical layer becomes.

Fabrinet — Manufacturing Capacity Behind the Optics Boom

Fabrinet — FN

Fabrinet’s most recent quarter generated:

Revenue: $1.214B
Adjusted EPS: $3.72

Revenue increased roughly 39% from the prior-year quarter, while earnings increased substantially.

Fabrinet is especially important because it sits deeper in the supply chain.

It manufactures many of the sophisticated optical and communications systems being deployed by the companies above.

CRI Read

Fabrinet represents the manufacturing capacity layer beneath optical networking.

When optical demand accelerates, manufacturing capacity becomes another potential bottleneck.

Ciena — AI Networking Is Moving Into Optical Transport

Ciena — CIEN

Ciena reported just ahead of this week:

Revenue: $1.57B
Growth: +40% YoY
Adjusted EPS: $1.64

The company raised FY2026 revenue guidance to approximately $6.3B ± $100M.

Hyperscalers represented roughly one-third of quarterly revenue, and Ciena received its first order for its RLS Hyper Rail photonic system.

CRI Read

This is important.

AI networking is extending beyond switches and transceivers into the optical transport architecture connecting increasingly massive data-center environments.

The Optical Bottleneck Is Becoming More Important

When Oracle reports:

$638B of backlog

and nearly doubling OCI revenue, the implications extend far beyond Oracle itself.

Those data centers require:

• GPUs,

• memory,

• networking,

• optical transceivers,

• lasers,

• switches,

• power systems,

• transformers,

• and cooling.

The earnings from AAOI, LITE, COHR, FN and CIEN collectively reinforce that optics is becoming one of the most important physical bottlenecks underneath AI infrastructure.

That is exactly the type of second-order signal CRI is designed to capture.

Power & Cooling — The Next Constraint Layer

The same logic extends another level down.

As compute capacity increases:

more compute → more power → more heat

That keeps several strategic CRI names highly relevant even when they did not report this particular week:

Cooling

• Modine — MOD

• Vertiv — VRT

• Trane — TT

• Carrier — CARR

• AAON — AAON

Electrical / Grid

• Eaton — ETN

• GE Vernova — GEV

• Powell Industries — POWL

• nVent — NVT

• Hubbell — HUBB

• Quanta Services — PWR

These companies should remain part of the CRI earnings universe because hyperscaler capex ultimately flows directly into their addressable markets.

Consumer Earnings Still Matter — But They Are a Different Story

Away from AI, this week also delivered useful consumer signals.

Casey’s — CASY

EPS: $4.37 actual vs. approximately $3.31 expected
Revenue: $4.57B vs. approximately $4.33B expected

Casey’s showed continued strength in convenience, prepared foods, and recurring traffic.

CRI Read: value + convenience remains healthy.

J.M. Smucker — SJM

Adjusted EPS: $2.77 vs. $2.64 expected
Revenue: $2.27B vs. approximately $2.26B expected

Smucker continues benefiting from at-home consumption while lower coffee input costs could improve future margins.

CRI Read: consumer defensiveness remains intact.

Campbell’s — CPB

Adjusted EPS: approximately $0.50
Revenue: $2.366B

Sales declined approximately 4% and inflation continued pressuring margins.

CRI Read: defensive demand remains stable, but margin quality matters.

RH — RH

Adjusted EPS: -$1.97 vs. roughly -$2.12 expected
Revenue: $800.3M vs. approximately $792.6M expected

Premium demand remained more resilient than broad discretionary demand.

CRI Read: the higher-income consumer remains healthier.

Dave & Buster’s — PLAY

Adjusted EPS: $0.22 vs. approximately $0.60 expected
Revenue: $559.2M vs. roughly $580.6M expected

The quarter materially missed expectations.

CRI Read: experience spending is not universally strong. Premium travel and marquee events remain healthier than middle-market entertainment.

CRI Earnings Regime Dashboard

Strong

AI Compute / Cloud

Oracle

AI Monetization

Adobe

Optical Infrastructure

AAOI, LITE, COHR, FN, CIEN

Convenience / Value Consumer

Casey’s

Constructive

Premium Consumer

RH

Defensive / At-Home Consumption

Smucker

Monitor

Packaged Foods

Campbell’s

Discretionary Entertainment

Dave & Buster’s

The Bigger CRI Signal

This earnings cycle is becoming easier to map.

The first stage was:

GPU scarcity

The next stages increasingly look like:

Memory → Networking → Optics → Power → Cooling → Grid Capacity

And the economic logic is straightforward.

Every incremental unit of AI compute requires:

• more memory,

• more bandwidth,

• more electricity,

• more heat removal,

• and more physical infrastructure.

That means the AI opportunity is broadening rather than disappearing.

The investment challenge is identifying where the next capacity constraints emerge.

CRI Bottom Line

This week strengthened the AI supercycle thesis rather than weakening it.

Oracle demonstrated extraordinary demand for AI cloud capacity.

Adobe demonstrated real AI monetization.

And the recent earnings from AAOI, Lumentum, Coherent, Fabrinet and Ciena confirm that the infrastructure underneath compute is scaling rapidly.

The market may still call this an AI trade.

But underneath the headlines, it increasingly looks like:

a compute, networking, optics, power and cooling infrastructure supercycle.

The AI infrastructure buildout delivered another important signal this week.

Compute demand remains extraordinarily strong. AMD reported that Data Center revenue more than doubled from a year ago. Arista Networks continued to benefit from rapidly expanding AI networking demand. Governments are committing billions of dollars to semiconductor manufacturing ecosystems.

But underneath those headline numbers, another bottleneck is moving rapidly toward the center of the AI infrastructure story:

Connectivity.

The industry has spent several years focused on obtaining enough GPUs.

Now the challenge is increasingly connecting those GPUs together efficiently enough to function as enormous computing systems.

That means switches, DSPs, transceivers, lasers, fiber, silicon photonics and eventually more advanced architectures such as co-packaged optics and optical I/O.

And this week brought an additional complication.

The U.S. administration is reportedly preparing restrictions on new Chinese optical transceiver models used in American data centers. Chinese manufacturers currently provide significant global optical manufacturing capacity, meaning restrictions could redirect demand toward Western suppliers while simultaneously tightening available supply.

At the same time, a new optical competitor emerged from stealth with more than $900 million of funding and products already shipping into production AI data centers.

The message is becoming increasingly difficult to ignore:

The next major AI bottleneck may not be producing intelligence. It may be moving the data required to create it.

CRI AI Infrastructure Dashboard

CRI Bottleneck Pressure Index: 89 / 100 — VERY HIGH

Previous reading: 88 / 100

CRI AI Buildout Phase: 6 / 10 — INFRASTRUCTURE EXPANSION

The increase in bottleneck pressure this week is primarily attributable to optical connectivity, power infrastructure and the continued expansion of semiconductor manufacturing requirements.

Increasing Constraints

🔴 Power & Grid — 10/10

🔴 Transformers & Switchgear — 10/10

🔴 Optical Networking — 9/10

🔴 Lasers & Photonics — 8/10

🔴 Materials — 7/10

Stable but Highly Constrained

🟡 Cooling — 9/10

🟡 HBM & Memory — 8/10

Slowly Improving

🟢 Advanced Packaging — 7/10

🟢 Semiconductor Manufacturing Capacity — 7/10

The important distinction is that improving manufacturing capacity does not necessarily mean demand is weakening.

In many cases, the industry is simply investing aggressively enough to begin addressing constraints.

The Biggest Development: Optics Moves Higher on the Bottleneck Map

The most consequential development this week came from the optical networking layer.

The U.S. administration is reportedly drafting rules that would prohibit U.S. data centers from purchasing new Chinese-made optical transceiver models. Reuters reported that the potential restriction could affect major Chinese suppliers while creating opportunities for companies including Coherent and Lumentum.

This matters because optical manufacturing capacity was already becoming strategically important.

Removing or restricting a major source of global supply does not reduce demand.

It potentially redirects it.

That could create a favorable environment for Western and allied suppliers including:

Lumentum

Coherent

Applied Optoelectronics

Corning

and potentially other optical component manufacturers throughout the supply chain.

But there is an important nuance.

Western suppliers may not immediately possess enough manufacturing capacity to replace Chinese volumes. Reuters specifically noted that Coherent and Lumentum offer competing technology but currently lack the scale required to replace Chinese suppliers outright.

That means a trade restriction could actually intensify the bottleneck before new capacity comes online.

For investors, that is potentially very important.

Why Optical Networking Is Becoming So Important

A GPU does not operate independently.

Large AI systems increasingly consist of tens of thousands—or eventually hundreds of thousands—of accelerators operating together.

The larger those clusters become, the more difficult it becomes to move information between processors.

This creates two increasingly important networking problems.

Scale-Out Networking

Scale-out networking connects racks and groups of compute systems across the broader AI data center.

This is where technologies such as Ethernet switching, 800G and 1.6T optical modules become increasingly important.

Scale-Up Networking

Scale-up networking connects accelerators much more closely within a computing domain.

As AI systems become larger, the distinction between individual processors and the broader computing fabric begins to blur.

The goal increasingly becomes:

Make thousands of GPUs behave like one enormous computer.

That requires extremely high bandwidth with increasingly low latency and lower power consumption.

And that is where photons begin replacing electrons.

Lumilens: A New Competitor Worth Watching

One of the week’s most interesting developments came from a company many investors had never heard of.

Lumilens emerged from stealth after raising more than $700 million in its latest financing, bringing total funding above $900 million at a reported valuation of approximately $5.51 billion.

More importantly, this isn’t merely another pre-revenue photonics startup.

The company says its technology is already shipping into production data centers under a multibillion-dollar customer agreement.

Lumilens is targeting both scale-out and scale-up AI connectivity.

Its technology roadmap reportedly includes:

• 1.6T pluggable optics

• Near-packaged optics

• Co-packaged optics

• High-density scale-up connectivity

• Large GPU cluster interconnect

This is a major validation of the optical thesis.

Private capital doesn’t commit more than $900 million to a company because optical networking is a minor component of AI infrastructure.

Capital is moving there because connectivity is becoming a fundamental constraint.

The Optical Technology Battle Is Getting More Interesting

There isn’t one obvious winner.

Several architectures are competing simultaneously.

Pluggable Optics

Pluggable transceivers remain the dominant commercial architecture.

Their biggest advantages are straightforward:

They are mature.

They are replaceable.

They are easy to service.

And an enormous manufacturing ecosystem already supports them.

The industry is now transitioning from 800G toward 1.6T.

Pluggables are unlikely to disappear anytime soon.

Silicon Photonics

Silicon photonics integrates optical functions using semiconductor manufacturing techniques.

Potential advantages include greater integration, lower cost at scale and better power efficiency.

Companies pursuing silicon-photonics strategies include Broadcom, Marvell, Cisco and multiple private companies.

This technology is likely to become increasingly important as optical components move physically closer to compute.

Co-Packaged Optics

CPO moves optical engines directly alongside switching silicon.

This potentially reduces electrical trace distances and improves bandwidth density and energy efficiency.

But CPO introduces difficult engineering problems:

• Thermal management

• Manufacturing yield

• Optical alignment

• Reliability

• Serviceability

• Laser replacement

CPO remains one of the industry’s most promising long-term technologies, but pluggable optics are unlikely to disappear simply because CPO exists.

The architectures will probably coexist.

External Light Sources

One solution to the laser reliability and thermal problem is to move the laser outside the optical package.

External light-source architectures allow lasers to operate in a more controlled thermal environment and potentially make replacement easier.

This is an important technology area for companies including Coherent, Lumentum and POET Technologies.

For POET specifically, this remains one of the reasons we continue monitoring the company despite having previously exited our position following its run-up.

Optical I/O

Companies such as Ayar Labs are attempting to move optical connectivity even closer to processors.

Instead of connecting servers optically, optical I/O potentially connects compute packages themselves.

This could eventually become extremely important for scale-up architectures.

But manufacturing maturity, packaging, cost and qualification remain major hurdles.

AMD Confirms That Compute Demand Is Expanding Beyond NVIDIA

Another important development this week came from AMD.

The company reported second-quarter Data Center revenue of $6.7 billion, up 107% year over year, driven by EPYC CPUs and Instinct GPUs. Total quarterly revenue reached $11.5 billion.

AMD also guided toward approximately $13 billion of third-quarter revenue, above the consensus estimate reported by Reuters at the time.

The significance goes beyond AMD gaining accelerator share.

More viable AI compute architectures mean:

More GPUs.

More HBM.

More advanced packaging.

More networking.

More optical connectivity.

More cooling.

And more electricity.

This is why competition with NVIDIA does not necessarily weaken the infrastructure investment thesis.

It may actually strengthen it.

Arista Confirms the Networking Layer

Arista Networks provided another confirmation that networking is participating directly in the AI buildout.

Its latest earnings report showed continued strong demand across large data-center and AI networking environments, while the company raised its outlook following better-than-expected results.

This is important because AI infrastructure increasingly needs to be viewed as a system.

Accelerators alone cannot create AI capacity.

The architecture increasingly looks like:

Compute

↓

HBM

↓

Advanced Packaging

↓

Switching

↓

Optical Connectivity

↓

Power

↓

Cooling

Every additional accelerator increases requirements elsewhere in the stack.

Power Remains the Ultimate Physical Constraint

Despite the excitement around optics this week, nothing displaced electricity as the largest AI bottleneck.

CRI maintains:

Power & Grid: 10/10

And semiconductor manufacturing itself is becoming another major source of electricity demand.

South Korea announced plans for a new 5 trillion won semiconductor fund, equivalent to approximately $3.5 billion, targeting semiconductor materials, components, equipment and fabless companies. Another 5 trillion won of trade financing is planned for export-oriented suppliers.

But one number buried within the announcement is particularly revealing.

South Korea plans to provide approximately:

14.7 GW

of electricity to its Yongin semiconductor cluster by 2041.

For perspective, that is not data-center electricity demand.

That is electricity required to help manufacture the semiconductor ecosystem supporting the technology economy.

This creates an increasingly important second-order effect.

AI requires electricity to operate.

But AI also requires enormous amounts of electricity to manufacture the chips, memory and components needed to build AI systems.

Power demand therefore exists on both sides of the supply chain.

Manufacturing Capacity Is Improving

Semiconductor manufacturing is one area where the bottleneck picture is becoming somewhat more constructive.

Governments and manufacturers are investing heavily across:

• Leading-edge foundries

• Memory

• Semiconductor equipment

• Advanced packaging

• Materials

• Components

• Substrates

South Korea’s new initiative is particularly noteworthy because it explicitly targets the supplier ecosystem rather than simply building more fabs.

That is exactly what is required.

A semiconductor fab without:

specialty chemicals,

precision equipment,

advanced substrates,

high-purity gases,

water,

electricity,

and qualified suppliers

does not create usable semiconductor capacity.

The manufacturing ecosystem must scale together.

That remains supportive of companies such as:

Lam Research

Applied Materials

KLA

ASML

Amkor

and the broader semiconductor materials ecosystem.

Materials: The Bottleneck Investors Rarely Discuss

Materials moved slightly higher on the CRI watchlist this week.

AI infrastructure ultimately depends on physical materials.

Among them:

Copper

Optical fiber

Specialty glass

Indium phosphide

Gallium arsenide

Advanced substrates

High-purity chemicals

Industrial gases

Thermal interface materials

Coolants

Precision ceramics

As manufacturing expands simultaneously across semiconductors, power infrastructure and data centers, demand for some of these materials can increase much faster than traditional forecasts anticipated.

Qualification makes the problem harder.

A semiconductor manufacturer cannot simply replace a critical chemical or optical material with a different supplier overnight.

The same applies to lasers, substrates and power components.

That makes seemingly obscure materials strategically important.

Cooling Remains at 9/10

There was no major development this week suggesting the cooling constraint is easing.

The industry’s transition toward higher rack densities continues pushing data centers toward direct liquid cooling.

The complete thermal system increasingly requires:

• Cold plates

• Pumps

• Manifolds

• Coolant distribution units

• Heat exchangers

• Controls

• Leak detection

• Facility heat rejection

This is why Vertiv remains particularly interesting.

It increasingly provides an integrated combination of:

Power + Cooling + Controls + Services

rather than one isolated component.

The cooling bottleneck isn’t primarily a question of whether the technology exists.

It does.

The challenge is deploying it across enormous numbers of racks quickly and reliably.

Hyperscaler Capex: Still No Retreat

The broader hyperscaler spending picture remains extraordinary.

Recent industry estimates place 2026 hyperscaler infrastructure spending near $700 billion, with Microsoft, Amazon, Alphabet, Meta and Oracle representing the majority of the buildout.

The important question is beginning to change.

It is no longer:

Will hyperscalers spend?

The evidence overwhelmingly says yes.

The more important question is becoming:

Can they generate sufficient returns from everything they’re building?

That will eventually become one of the most important indicators in the AI cycle.

CRI will increasingly monitor:

AI revenue growth

Cloud growth

Infrastructure utilization

Remaining performance obligations

Customer prepayments

Capex growth

Depreciation

Free cash flow

and long-term lease commitments.

A slowdown in capex alone would not necessarily invalidate the AI thesis.

But slowing capex combined with slowing utilization, weakening backlogs and disappointing AI monetization would.

We are not seeing that combination today.

CRI Emerging Constraint Watchlist

🔴 Optical Manufacturing Capacity

Moved higher this week.

Potential restrictions on Chinese suppliers could tighten supply while Western manufacturers attempt to expand capacity.

Key names:

LITE

COHR

AAOI

GLW

🔴 Power Availability

Still the deepest physical constraint.

Key names:

ETN

GEV

POWL

VRT

plus utilities, independent power producers and electrical-equipment manufacturers.

🟠 Semiconductor Materials