Two forces did the damage in almost equal measure: a ten-year Treasury yield that keeps grinding higher after Kevin Warsh’s Jackson Hole speech, and a Brent barrel that jumped back above $90 as the US and Iran traded missile salvos again over the weekend.

Under the surface, a couple of single-name stories deserve more attention than the tape gave them.

The Rundown

• The Nasdaq and Dow closed lower, ending a positive August on a red session.

• Long yields kept climbing and September hike odds now sit near 70%.

• Renewed US-Iran strikes pushed Brent above $90 and lifted the energy complex.

• A California wildfire liability vote crushed the state’s biggest utility names.

Warsh’s echo keeps pressuring the long end

The tape has been trading with a hawkish tilt ever since Kevin Warsh used his first Jackson Hole speech as Fed chair to remind the market that the inflation fight is not finished.

Monday extended that read. The ten-year Treasury yield added another four basis points to close at 4.76%, which is the kind of level that starts to matter for equity multiples. According to the CME FedWatch Tool, roughly 70% of traders now expect a rate hike at the September 16 meeting. That is a full repricing from where the curve was two weeks ago.

The Dow Jones (DJI | ▼0.70%) took the brunt of it, losing 0.7% on the session. The Nasdaq (COMPX | ▼0.12%) held up better but still closed in the red. The euro/dollar drifted up to 1.1616, gold gave back about 1% on the day but printed a 10% monthly gain, and bitcoin closed August roughly 25% higher.

The message under all of that is consistent: fiscal discipline is not coming back any time soon, the AI capex arms race is competing with Treasury issuance for dollars, and investors are hedging by whatever means they have.

Iran-US strikes put oil back on the front page

The weekend brought the first serious military exchange between the US and Iran in over a month.

American strikes hit targets on Larak Island near the Strait of Hormuz, where Iran was reportedly staging missile launchers. Tehran retaliated with strikes on US assets in Jordan, with damage to Air Force infrastructure and reported casualties.

President Donald Trump escalated the rhetoric further, threatening on Truth Social that Iran’s main crude terminal on Kharg Island will be “shot to pieces.”

The oil market did the math immediately.

Brent for November delivery climbed 1.3% to $90.49, while WTI for October added 2.8% to $85.76. Both benchmarks are back to levels that will feed straight into the next round of headline inflation prints.

The energy complex was the one place equity money wanted to be.

CHEVRON CORP (CVX | ▲2.12%), EXXONMOBIL HOLDINGS CORP (XOM | ▲2.71%), HALLIBURTON CO (HAL | ▲1.85%) and CONOCOPHILLIPS (COP | ▲1.64%) all traded up two to three percent on the day.

Amazon draws an FTC lawsuit over ad practices

The single biggest single-name story of the day was in tech.

The Federal Trade Commission is preparing to sue AMAZON.COM INC (AMZN | ▼2.50%) over its sponsored-advertising business, with the agency alleging that the company misled advertisers about how prices and auctions actually worked.

According to the Wall Street Journal reporting that Amazon itself confirmed, the FTC estimates that the practice generated tens of billions of dollars in revenue over the last seven years.

The stock lost 2.5% on the news. That is a measured reaction rather than a panic one, and it fits with how these cases usually play out.

The regulatory clock is long, the appeals are longer, and the true financial impact is often smaller than the initial headline implies. What this does do is add a second front to the antitrust file Amazon is already fighting, and it puts a permanent question mark on the growth rate of one of the highest-margin businesses inside the company. That is worth pricing in.

Nvidia keeps writing checks into its ecosystem

NVIDIA CORP (NVDA | ▲1.48%) went the other way.

The company committed $3.5 billion to Taiwanese peer MediaTek in a convertible note structure that can eventually flip into equity. CEO Jensen Huang framed the tie-up as an “engineering alignment,” which is the polite way of saying that Nvidia wants MediaTek’s chip designs built around Nvidia’s platform standards.

This is the same playbook I have watched Nvidia run all year: use the balance sheet to make sure that every serious player in the AI chip supply chain is either a partner, a customer, or both.

Whether $3.5 billion into a single Taiwanese chip designer moves the needle on the ecosystem thesis is debatable. What is not debatable is that Nvidia is running out of ways to spend the cash it prints, and taking equity-linked positions in strategic partners is a rational use of that pile.

California utilities take a $40 billion haircut

The most brutal single-day chart of the session belonged to two names outside the tech complex.

P G & E CORP (PCG | ▼20.06%) and EDISON INTERNATIONAL (EIX | ▼23.07%) each collapsed by roughly 20% after California lawmakers blocked a bill that would have limited utility liability for wildfires.

Under the current regime, both companies remain liable for damage caused by their infrastructure even in cases where negligence is not proven, and they can be pushed into faster payouts on claims.

PG&E has been here before. The utility spent years working its way out of the bankruptcy that followed the Camp Fire liability claims, and the whole point of the industry lobbying effort in Sacramento was to make sure that wound never re-opened.

Musk’s tweet ambushes Howmet

Elon Musk continues to be the single most efficient market-moving Twitter account on the planet, and HOWMET AEROSPACE INC (HWM | ▼7.51%) learned that again on Monday.

Musk posted that SPACE EXPLORATION TECHN-CL A (SPCX | ▲1.55%) will begin producing its own blades and vanes for gas turbines, arguing that vertical integration will “compress gas-turbine commissioning times to at most 18 months.”

Howmet, which supplies exactly those components, dropped 7.5%.

Whether SpaceX can actually deliver on that timeline is a separate question, and the answer is usually “eventually, and later than promised.” But when the largest downstream vertical integrator in aerospace and defense signals he is going to make in-house what you were counting on selling him, the multiple compresses immediately. The market rarely waits for the execution to confirm the risk.

Management changes at Herbalife and Pinterest

Two governance headlines hit harder than they usually would in a nervous tape.

HERBALIFE LTD (HLF | ▼11.58%) lost close to 12% after announcing that CEO Stephan Gratziani will step down on October 31, with CFO John DeSimone stepping in on an interim basis. Full-year guidance of 2.5% to 5.5% revenue growth was reaffirmed, but that range brackets the FactSet consensus of 4.4%, and a CEO departure in a franchise that has spent years fighting for narrative stability is not what shareholders wanted to see.

PINTEREST INC- CLASS A (PINS | ▼6.38%) dropped 6.4% on a similar theme, with CFO Julia Brau Donnelly announced as leaving at the end of October.

VP of finance and operations Vikram Naidu takes the role in an interim capacity. In both cases the market is repricing execution risk faster than the underlying business is actually deteriorating, but that is what happens in tape environments where investors have no patience for uncertainty.

Tesla and GameStop swim against the current

TESLA INC (TSLA | ▲5.51%) added 5.5% ahead of expected updates to its self-driving stack. Musk has been telegraphing an incoming pothole-avoidance feature and continued progress toward true autonomy, and the stock trades on that narrative in a way no other automaker’s does.

GAMESTOP CORP-CLASS A (GME | ▲2.85%) picked up nearly 3% on preliminary second-quarter numbers.

Net sales are expected to be down year-over-year, but both operating income and net income should improve on the comparable period. The net result includes roughly $238 million of gains linked to eBay-related interests, partially offset by a roughly $75 million loss on digital assets and related receivables.

It is the eBay windfall doing the heavy lifting on the bottom line, not the core retail business, which tells you exactly what kind of company GameStop has become.

What to watch this week

The macro calendar builds into Friday.

Purchasing managers’ indices land through the week, the Fed’s Beige Book prints Wednesday evening, and the August payrolls report closes the week on Friday.

Given the September 16 meeting and the pricing that has already moved to 70% for a hike, the payrolls print is the single most important data release of the month. A hot number cements the hike. A soft one puts the pause trade back on the table.

US Market Breadth

Latest article: August 31 after market close (link)

Current Market breadth: September 01 after market close - Neutral, Negative Bias (no change)

Conclusion

Higher long yields, rising oil, a Fed chair who has planted his hawkish flag, and a couple of single-name blowups on the regulatory and governance side.

None of it is a break in the trend. But August was a positive month that ended on a red session, and the setup into September is meaningfully less friendly than it looked two weeks ago.

I would rather be patient here than aggressive.

ChartMill Market Desk - Kristoff

With regard to the stocks discussed in the article above; the author owns individual shares in Nvidia.

This daily update is prepared by ChartMill for informational purposes only and does not constitute investment advice. Always do your own due diligence before making investment decisions.

For a stock that already represented 17.08% of my US Longterm Portfolio before the release - and, together with Microsoft, accounts for almost 35% IT exposure - it is fair to say this was an important update.

Let me get straight to the point: I’m holding both tranches.

• First tranche at $175 (April 1), second at $225 (June 2), average purchase price $189.29. Unrealized return at the premarket price: +17.4% in dollar terms.

• Tranche 1 is comfortably in profit (+27%), while tranche 2 is almost at break-even (-1.25%).

Why I’m keeping both purchases after this quarterly update is explained below.

The numbers themselves

Revenue: $96.2 billion. +106% year-over-year. Consensus was $92.2 billion, so that is a $4 billion beat. Fourth consecutive quarter of accelerating growth. For a company of this scale, that is unusual.

Gross margin: 75.0%. Stable versus Q1 (74.9%), and 2.5 percentage points higher than a year ago. Data Center revenue was $89 billion (+117% YoY), of which hyperscalers contributed $48.7 billion and the new ACIE category (AI clouds, industry, enterprise) $40.3 billion. The latter is growing +138% YoY. The customer base is measurably broadening.

Non-GAAP EPS came in at $2.22 versus consensus of $2.09. GAAP EPS was $2.46. $26 billion was returned to shareholders in a single quarter, including $20 billion in buybacks. Another $99 billion of buyback authorization remains available.

The guidance is where it gets interesting

For Q3, Nvidia is guiding for $108 billion. That implies another sequential acceleration of +12%, and that is still excluding any Data Center revenue from China (which is set at zero in the outlook, out of caution).

What is new: CFO Colette Kress provided explicit full-year guidance for FY28 (calendar year 2027) for the first time, pointing to approximately 70% revenue growth. An important detail: that 70% is supply-constrained.

According to CEO Jensen Huang, “unconstrained” demand is closer to 100%. Nvidia is limiting its own forecast because supply cannot keep up, not because demand is weakening. Companies of this size rarely provide that level of visibility one year ahead, especially at these growth rates. To me, that is a very clear sign of significant confidence in the future.

Where the headwinds are, and why I’m keeping both positions in the portfolio

Gross margin will decline over the next few quarters. Q3 is expected at 74.0%, Q4 at 71–72% (the trough), followed by a recovery toward 72–73% in FY28 once implemented price increases begin to take effect.

The reason: extremely high memory prices (HBM). Nvidia itself acknowledged that the cost peak is “higher than expected and will rise further next year.”

To me, that sounds like a headwind, but not a break in the thesis. The pricing power underpinning the thesis remains clearly measurable. The right metric is revenue per gigawatt of data center capacity:

• Hopper generation: $18 billion per gigawatt

• Grace Blackwell: $25 billion

• Vera Rubin (in production since August): $40 billion

Each generation generates more revenue per physical data center site. Customers are willing to pay for that because the new hardware delivers 30x more throughput per megawatt and 35x lower cost per token. For a hyperscaler evaluating its AI factory based on return on invested capital, that is an easy conversation.

HBM costs are rising, and Nvidia is passing them through without losing customers. That is exactly what pricing power looks like in practice.

The risk that is new

Nvidia is now structurally investing in its own customers.

Approximately $50 billion has gone into frontier AI labs (OpenAI, Anthropic and others). On top of that, there is a framework with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to channel $500 billion in external capital toward AI infrastructure.

Analysts literally referred to this as “circular financing” on the call: Nvidia invests in companies that subsequently buy Nvidia chips.

Jensen defends it as a “once in a generation opportunity.” I interpret it as a shift from a pure fabless chip designer toward an infrastructure financier. For a long-term investor, that is a new variable that needs to be reassessed every quarter.

Today, it remains manageable (~$50 billion versus free cash flow that can comfortably exceed $100 billion). If that pool rises toward $150 billion without a clear ROIC justification, the conversation changes.

Why I’m not adding now

Position size!

NVDA currently represents approximately 17.1% of my US Longterm Portfolio. Together with Microsoft, IT accounts for 34.4% of the portfolio. Both benefit from the same AI capex cycle.

Averaging down on the second tranche would only push the concentration in NVDA toward 19–20% and IT exposure above 36%.

The portfolio is built around quality-longterm rules: no stop-losses, and selling only when the thesis breaks or a position becomes too dominant.

How a name like Nvidia ends up in this portfolio in the first place - which screener I use, why the thresholds sit where they do, and why I manage risk through position sizing rather than stops - I walked through in detail in this post: ‘How I Build The Shortlist for My US Long-Term Portfolio’.

It’s the foundation every one of these quarterly updates, this one included, is built on.

So, a third tranche in Nvidia would only make sense if:

• (i) the share price falls below my average cost basis of $189.29,

• (ii) Q3 confirms that gross margin remains at least 73.5%, and

• (iii) Data Center growth remains at least +40% YoY.

The +6% premarket reaction does not bring the stock into that area. Averaging down is therefore not a current scenario.

What would change my mind

I maintain the thesis as long as these four checkpoints remain intact:

1. Gross margin remains in the 70%+ range (confirmed; 75.0%)

2. Data Center and networking grow faster than the broader AI capex cycle (confirmed; DC +117%, networking +18% QoQ)

3. No competitive erosion of pricing power (confirmed; revenue per gigawatt scales from $18B to $40B)

4. Vera Rubin scales without production issues or demand shortfalls (confirmed; in production, 20% of Q3 Data Center revenue)

Developments that would make me reconsider:

• Gross margin below 68% without a path to recovery

• Sequential revenue decline in Data Center

• Material write-downs (>$10 billion) on the Frontier Labs investments

• Measurable decline in hyperscaler orders due to in-house custom silicon

Bottom line

Q2 confirms the thesis for continuing to hold the shares.

FY28 guidance of 70% growth (supply-constrained) provides the kind of visibility that supports a long-term investment horizon. The margin pressure has been openly anticipated, not hidden. That, to me, is a clear sign of management transparency.

The position remains unchanged. No additional buying unless specific conditions are met. Selling is not on the table.

Next checkpoint: Q3 gross margin (target 74.0% ± 50 bps). If that comes in as expected and Vera Rubin reaches the guided 20% mix, this remains the most compelling AI infrastructure position in the portfolio.

Kristoff - ChartMill

The honest answer: It all starts with the screener. Not just any screener: ChartMill is the software that a few fellow investors and myself founded back in 2012.

It was originally built purely as a tool for our own analyses. It wasn’t until later that we made it available to a wider audience. I still use it today, and this article is living proof of that.

Why does a screener work for me?

Not because a screener hands me the “winners” - no screener does that - but because it lets me take a universe of thousands of US-listed companies and cut it down, in a handful of seconds, to a list small enough to actually look at.

What follows is the exact screen I use for that first pass.

No mystery, no magic formula. Just a set of filters that together try to do one thing: surface fundamentally strong companies that are also performing right now.

You can open and replicate the full screen here: [link]

To make the theory concrete right away: at the bottom of this article you’ll find a handful of charts of open positions in my US Long-Term Portfolio.

No backtest, no paper trading, no theoretical model portfolio, just actual positions with my own money in them. Every one of them came out of this exact screen at some point.

Those charts read very differently once you understand why they made it through the filters in the first place. Hence the order: framework first, examples after.

The Hygiene Filters (Briefly)

Before I get to the substantive filters, a few basic choices that barely need explaining:

- Exchange: US-listed only. This portfolio is US-only by design.
- Price: Minimum $10. Anything below that tends to have a different character (higher volatility, thinner liquidity, more penny-stock behavior). Not my universe.
- Volume: Enough average daily volume to get in and out without slippage.
- Market cap: Small, mid and large. I deliberately exclude micro-caps.

These aren’t real selection criteria. They’re hygiene filters, they don’t tell me which companies are interesting, they just make sure what’s left is actually investable for my style.

The real work comes next.

The Substantive Filters

ROIC ≥ 12% (excluding cash)

Of all the fundamental metrics, Return on Invested Capital is the first one I look at.

Put simply: how much return does a company generate on every dollar actually deployed in the business?

Why 12% as a minimum?

Because growth only creates value when return on invested capital exceeds the cost of capital. For most companies, WACC sits somewhere between 8% and 10%.

A ROIC of 12% is comfortably above that. Enough of a margin to know that every additional dollar the company reinvests is actually adding value rather than destroying it.

On top of that, a consistently high ROIC is one of the clearest markers of a durable competitive advantage (MOAT).

And that’s exactly what the quality investor is hunting for: businesses with something competitors can’t easily replicate, and that can therefore keep delivering above-average returns over long periods.

I use the “excluding cash” variant because a large cash balance artificially drags the ratio down for companies holding a lot of liquidity. I want to know how productive the actually deployed capital is.

ChartMill Fundamental Analysis Score ≥ 6/10

The FA score is a composite rating covering earnings growth, margins, capital efficiency and a handful of other fundamental parameters. I set the minimum at 6 out of 9. Above average, without setting the bar at perfection.

Why not push it to 8 or 9?

Because filters that are too strict collapse the list down to a handful of names everyone already talks about. I want breathing room. A score of 6 lets in interesting companies that aren’t (yet) in the spotlight.

Health Score ≥ 5/10

Health looks at balance sheet structure: liquidity, solvency, debt position relative to cash flow. I’m deliberately not too strict here.

A company can absolutely use debt actively to grow, but there’s a floor. Below 5/10 you’re in territory where the balance sheet is structurally weak, and that’s something I want to avoid in a long-term portfolio.

Debt / FCF ≤ 5

On top of the Health score, I use a separate hard filter: total debt can be at most five times free cash flow. Concretely, that means if a company stopped investing in growth tomorrow and used all of its FCF to pay down debt, it should be debt-free in five years or less.

That’s a deliberately conservative criterion. It filters out plenty of high-growth names that lean heavily on borrowed capital, and that’s fine by me. I’m not looking for leverage. I’m looking for earning power.

Valuation Score ≥ 4/10

Valuation is never my starting point, but it is a guardrail. A score of 4 out of 10 means: not a structurally overvalued business. Anything below that gets risky for anyone with a longer holding period.

You’re too often paying for growth that’s already priced in for the next several years.

Important: I intentionally don’t set the bar high here. Quality companies are rarely cheap, and waiting until they are usually means waiting for something to go seriously wrong with the business. Reasonably priced is enough.

Growth Score ≥ 4/10

Growth is the counterweight to valuation. Without growth, every valuation story eventually runs out of runway. A minimum of 4/10 isn’t asking for hyper or even strong growth, just a clear, sustained trend in revenue and/or earnings.

Profitability Score ≥ 7/10

Profitability is a composite score that pulls together several profitability metrics at once: net and operating margins, ROIC, ROE, ROA, and how consistent they are over time.

Why do I set the bar at 7 here instead of 4 like most of the other scores?

Because profitability is the heart of what I’m looking for. Filtering on ROIC alone isn’t enough.

A company can have solid ROIC and weak margins at the same time, or great margins and mediocre ROA. I want the business firing on multiple profitability dimensions at once.

Why do these differences between companies exist in the first place?

It comes down to how fast capital cycles back.

A supermarket earns just a few percent per sale, but turns its inventory over dozens of times a year. Little margin per transaction, many transactions on little invested capital, and the result is a solid ROIC.

A telecom or utility company sits at the other extreme: decent margins, but so much capital tied up in networks and infrastructure that ROIC stays muted.

The businesses that score on both fronts, high margins and little invested capital, are rare.

Think Visa, Mastercard, credit rating agencies, or software with network effects.

Those are the “compounders” quality investors are hunting for, and that’s exactly why Profitability as a composite score tells you more than ROIC alone.

A profitability score of 7 out of 10 means, in practice, that the business consistently earns more than its peers and has done so over multiple years. Not one lucky quarter.

That’s exactly the kind of profile you want to see in a long-term portfolio.

Why Not Set the Bar at 10/10 Anywhere?

You’ll notice I never push a ChartMill score to 9 or 10, and most minimums sit somewhere between 4 and 7. That’s deliberate.

A screen that’s too strict returns five names everyone is already talking about. Right now my screen leaves me with roughly 250 stocks.

That sounds like a lot, but against the full universe of thousands of US-listed names it’s still a genuinely strict selection.

Those 250 names are my universe, the pond I actually fish in for portfolio candidates. Broad enough for real diversity, tight enough to work with in a structured way.

And Then What? The Real Selection Starts Here.

The output of this screen is a universe, not a portfolio. Out of those roughly 250 names, only a fraction ends up as a serious candidate.

The filters that come after the screen matter as much as the screen itself.

Price trend.

I prefer to work with stocks in Stage 2 of the Weinstein model, either coming off a temporary pullback within that trend or breaking out of a new base.

Stocks breaking out of Stage 1 (basing) with real conviction are also welcome. You catch the trend close to its starting point that way.

What I avoid at all costs, no matter how strong the fundamentals look, are stocks in an obvious Stage 4 downtrend. That’s exactly what I use technical analysis for: to stay out of falling knives.

Support and resistance.

Where is the price sitting relative to the levels that matter?

A stock running straight into heavy resistance calls for a different approach than one coming back to test solid support.

This doesn’t decide whether I buy, but it does decide when and at what price I actually step in.

Sector weighting.

If my portfolio already consists largely of tech stocks and the screening identifies three more tech stocks that present a good entry point, that’s unfortunate.

The only thing I can do then is wait for other opportunities to arise.

I know not everyone will agree with this rule because it severely limits profit potential, but I believe that such concentration risk only belongs in a much more actively managed portfolio.

Position sizing, not stops.

Again, this is deliberately a long-term portfolio, not an active trading strategy. I don’t use stop losses here when entering a new position. Initial risk per position is managed by sizing that position against the rest of the portfolio.

Once bought, I want to hold as long as possible.

Re-evaluation happens at two fixed moments: after earnings, or when a position grows too large relative to the total portfolio.

Some current open LONG positions

Netflix

H&R Block

Nvidia

Bottom Line

A screener isn’t an oracle. It’s a filter, and filters only work if you know what you want to pull out of them and what still needs to happen after.

This particular screen consistently gives me a working list of fundamentally healthy US-listed companies that score above average across multiple profitability angles.

What comes next - the technical picture, timing, sector weighting - matters just as much.

If you want to experiment yourself: Check my personal screen through this link.

Adjusting it to your own style is a matter of moving a few sliders. And that’s the whole point.

Trade Safe!

Kristoff, Co-founder ChartMill

Wednesday belonged to Moderna, whose biggest one-day rally since the pandemic era was a validation of a decade of mRNA research.

Around that story, the tape did something quieter but arguably more important: long-dated Treasury yields finally exhaled after a fresh intervention from the Treasury Department, and the Fed minutes reminded everyone that the hawkish camp inside the FOMC is bigger than the vote count suggests.

The Rundown

• A landmark late-stage cancer vaccine readout reshaped the biotech narrative in a single afternoon.

• Treasury’s decision to double its long-bond buyback program pulled yields lower and gave equities room to breathe.

• Fed minutes revealed a broader hawkish contingent than the July dissent tally implied.

• US federal debt quietly crossed a symbolic threshold that most investors barely noticed.

Moderna’s monster session

MODERNA INC (MRNA | ▲176.97%) added roughly $45 billion in market cap in a single day after its personalised mRNA vaccine, developed with MERCK & CO. INC. (MRK | ▲12.60%), cleared its Phase 3 trial in advanced melanoma.

This is the first therapeutic cancer vaccine to succeed in a late-stage stud, meaningful not only for the two companies filing for approval but for the whole thesis that mRNA platforms could extend well beyond infectious disease.

For context: the shares closed near $174, which is a spectacular move on paper but still nowhere close to the roughly $400 peak from late 2021, when Moderna’s Covid vaccine put the platform on the map. What today’s readout does is give the story a second act and a much broader one, because a melanoma win opens a credible path to trialling the same approach in other tumour types.

I’d flag the risk too: a $45B repricing in a day discounts a lot of what still needs to be executed, from regulatory approval to manufacturing scale to real-world uptake. But as a scientific milestone, this one is hard to overstate.

Treasury steps in, yields step back

The other move that mattered was in fixed income. Treasury Secretary Scott Bessent announced that the buyback programme for long-dated government paper will be at least doubled, from $2 billion to a minimum of $4 billion per tranche.

The 30-year yield dropped nine basis points to 5.20%, the 10-year eased five basis points to 4.65%, and equities were happy to accept the assist.

I’ll be blunt about my read here: this is symptom management.

The underlying problem - a persistently large federal deficit and a rapidly compounding debt stock - is not solved by the Treasury being a more aggressive buyer of its own paper. It buys time, and time is not nothing when markets are jittery, but it doesn’t change the trajectory. Which brings us to the number nobody wants to discuss.

$40 trillion, and counting

The US public debt clock quietly ticked past $40 trillion on Wednesday evening. Less than five years ago, in January 2022, it crossed $30 trillion. Even stripping out the roughly $7.8 trillion of intragovernmental holdings, the publicly held debt is now above 100% of GDP, a threshold last breached in the aftermath of World War II.

The fiscal year running through September is on track for a $1.8 trillion deficit, and annual interest costs have already crossed the $1 trillion mark, higher than defence spending and roughly triple the level from the start of the decade.

This is why the 10-year yield is trading at levels not seen since 2007: bond investors are asking for a bigger discount, and they’re doing so at exactly the moment when the largest US tech companies are also issuing record amounts of debt to fund their AI capex. Treasury’s buyback is a lever. The deficit is the machine.

Marvell wins a piece of Google’s TPU roadmap

MARVELL TECHNOLOGY INC (MRVL | ▲9.85%) rallied roughly 10% after announcing an expanded partnership with ALPHABET INC-CL A (GOOGL | ▲0.15%) to co-develop custom silicon for Google’s TPU ecosystem.

Google gets the right to take an equity stake in Marvell of up to $12.2 billion, which is a meaningful commitment even by hyperscaler standards.

Strategically, this reads as Google doubling down on reducing its Nvidia dependency and pulling more of its AI infrastructure cost curve in-house. The interesting collateral is BROADCOM INC (AVGO | ▼4.61%), which has been Google’s primary TPU partner and now has to share the field.

Every hyperscaler is under the same pressure to diversify custom-silicon relationships, and deals of this size reset the going rate for that kind of collaboration.

Estée Lauder and the lipstick effect

ESTEE LAUDER COMPANIES-CL A (EL | ▲16.30%) delivered a clean beat, with revenue up 6% to $3.6 billion. Skincare and fragrance are doing the heavy lifting, and CEO Stéphane de La Faverie - in the seat since early 2025 - appears to be executing on a tighter, younger-consumer strategy anchored around Amazon and TikTok Shop distribution.

There’s also a cyclical thread worth pulling.

Leonard Lauder coined the “lipstick effect” after 9/11 to describe how consumers, when economic confidence wobbles, cut the big-ticket purchases but keep spending on small, affordable indulgences.

Whether that dynamic is meaningfully in play right now is impossible to prove, but the strategy the current CEO is running - pushing skincare and fragrance through the channels that younger, price-conscious buyers actually use - happens to be well-designed for exactly that environment.

Fed minutes: the hawks have more company than the vote showed

The July 29 FOMC minutes carried a clear hawkish tilt. Beyond the three governors who dissented in favour of a hike, several non-voting members appear to have been sympathetic to that view.

The language - “many participants judged that a tighter stance of policy would likely be needed if inflation did not moderate” - is not subtle.

Chair Kevin Warsh pointed to the market-driven rise in yields since the June meeting as evidence that financial conditions were tightening without the Fed having to move, but some participants pushed back that conditions still weren’t restrictive enough.

Warsh also floated the idea of reducing the number of policy meetings to six per year, though that wouldn’t take effect this year. According to CME FedWatch, markets are now pricing a 67% probability that the Fed holds at the September meeting – but the minutes make it clear the bar for a hike has moved lower.

Bitcoin, oil, and the dollar

Bitcoin climbed 6% to $68,200 after a Senate vote on crypto market-structure legislation was scheduled for September. The White House also hosted crypto industry leaders on Wednesday, which reads as coordinated political signalling ahead of the vote.

WTI crude ticked up roughly 1% for a fourth straight session on unresolved Middle East risk. The dollar weakened, with euro/dollar climbing 0.9% to 1.1678 in the aftermath of the Fed minutes.

Retail: Target up, Lowe’s mixed

TARGET CORP (TGT | ▲4.28%) raised full-year revenue growth guidance to 5% from 4% after a strong second quarter, with an accompanying lift to the earnings outlook.

LOWE’S COS INC (LOW | ▲2.02%)‘s went the other direction on the top line, trimming its 2026 revenue view, but the stock still finished higher, aa reminder that positioning and expectations matter as much as the print itself.

US Market Breadth

Latest article: August 19 after market close (link)

Update: Neutral, Positive Bias - unchanged from the previous session.

Conclusion

Wednesday was a session where good news finally showed up in size. Moderna’s cancer vaccine result is the kind of biotech readout that reshapes a whole sub-sector’s narrative, Marvell picked up a genuinely strategic customer win, and Treasury bought equities a day of yield relief. But underneath, the two structural signals – hawkish Fed minutes and a $40 trillion debt clock – haven’t gone anywhere. The tape can lift on a great biotech day. The rate backdrop is a slower story, and it’s still moving in the wrong direction.

ChartMill Market Desk - Kristoff

This daily update is prepared by ChartMill for informational purposes only and does not constitute investment advice. Always do your own due diligence before making investment decisions.

👉 This article is also available at ChartMill.com.

The longer-term structure remains constructive, but enough short-term deterioration has developed to downgrade the breadth trend from positive to neutral with a positive bias.

Index Overview (SPY, QQQ, IWM)

Short Term (Daily)

Long Term (Weekly)

SS SPDR S&P 500 ETF TRUST-US (SPY | ▼0.68%) fell 0.68% to 767.45, retreating from last week’s highs and closing below its EMA9 at 769.89. It remains above the EMA21 at 762.53, so this is still a pullback within the existing short-term uptrend rather than a confirmed trend break.

The weekly structure also remains positive, with SPY comfortably above its rising long-term ChartMill Trend Indicator.

INVESCO QQQ TRUST SERIES 1 (QQQ | ▼1.69%) was considerably weaker, falling 1.7% to 717.51. It also closed below its EMA9 and is now only slightly above the EMA21 at 714.47. Of the three ETFs, QQQ has the smallest short-term cushion before the recent advance starts to look more technically damaged.

The weekly long-term trend nevertheless remains clearly positive.

ISHARES RUSSELL 2000 ETF (IWM | ▼1.26%) declined 1.3% to 300.23. Like SPY and QQQ, it finished below its EMA9 but remains marginally above the EMA21 at 298.96. After small caps had strengthened substantially through the first half of August, Tuesday interrupted that improvement without yet reversing it.

The weekly chart remains constructive and well above the rising long-term trend line.

The common message across all three charts is unusually consistent: short-term momentum weakened, but the intermediate and long-term structures remain intact. All three ETFs lost their EMA9 on the same day while holding their EMA21s.

Daily Breadth Deteriorates Further

The internal market picture was weaker than the index declines alone suggest.

Only 32.5% of stocks advanced Tuesday, while 64.5% declined. Monday had already produced a similarly weak 33.4% versus 64.0% split, so Tuesday was not an isolated negative session.

The deterioration was more visible among the larger movers. Stocks gaining at least 4% fell from 3.1% to 2.7%, while 4% decliners jumped from 4.3% to 6.4%. That widening downside imbalance is more concerning than the headline advance/decline figure because selling pressure is beginning to produce larger individual-stock losses.

New highs also lost ground. Only 2.3% of stocks registered a new high, down from 3.5% Monday, while new lows increased to 3.3%. New lows exceeding new highs is a clear change from the much healthier readings seen last week.

Pocket pivots were one of the few stable readings, edging up from 21.1% to 21.7%, but they are not strong enough to offset the broader deterioration elsewhere.

Moving-Average Breadth Loses Momentum

The moving-average data still argues against calling this market broadly negative, but the direction has changed noticeably.

The percentage of stocks above their 20-day SMA fell from 64.5% to 59.6%, while the 50-day reading declined from 59.9% to 55.4%. Participation above the 100-day average dropped to 54.0%, and the 200-day figure slipped to 51.1%.

All four measures remain above 50%, which preserves the market’s constructive intermediate structure. Yet the margin has narrowed quickly. As recently as August 13, 71.1% of stocks were above their SMA20; five days later that figure is below 60%.

The 200-day breadth reading deserves particular attention. At 51.1%, the longer-term market is still technically on the positive side of the ledger, but only barely. Further weakness would leave the major indexes near their highs while fewer than half of individual stocks remained in long-term uptrends.

Short-Term Participation Rolls Over

The largest day-to-day change appeared in weekly breadth.

Weekly advancers dropped from 57.5% to 42.7%, while weekly decliners jumped from 41.3% to 56.2%. This confirms that the weakness is no longer limited to Tuesday’s session; it is beginning to show up across the one-week window as well.

The medium-term numbers remain more resilient. Monthly breadth is still positive at 57.4% advancing versus 42.1% declining, while three-month breadth stands at 53.1% versus 46.1%.

Even here, however, the direction is softer. The percentage of stocks up at least 25% over three months slipped to 9.7%, while those down at least 25% rose to 9.9%. The previously positive spread has essentially disappeared.

So the breadth picture has developed a clear time-frame divergence: short-term participation has weakened substantially, while the intermediate-term structure remains modestly positive.

Market Context

Tuesday’s market news. The 30-year Treasury yield moved toward 5.3%, its highest level in roughly 19 years, while semiconductor stocks reversed sharply after Monday’s rally. The Nasdaq absorbed most of the index-level damage, with numerous semiconductor and AI-related leaders posting steep declines.

The breadth figures add useful context to that narrative. This was not merely a handful of expensive semiconductor stocks pulling QQQ lower: almost two-thirds of the broader stock universe declined, and the percentage of stocks falling more than 4% accelerated noticeably.

The pressure was therefore broader than the headline chip selloff alone.

Breadth Trend Rating

Current breadth trend rating: 5 - Neutral, positive bias

This is a downgrade from 6 - Positive, which was the assessment carried into Tuesday.

One weak session would not justify changing the rating, especially while all four moving-average breadth measures remain above 50% and the weekly index trends remain positive.

But the combination of two consecutive sessions with roughly 64% decliners, sharply weaker weekly participation, falling moving-average breadth, new lows overtaking new highs and all three index ETFs losing their EMA9s represents a meaningful deterioration.

A neutral-positive rating better captures the current balance: the established uptrend has not broken, but breadth is no longer strong enough to provide clean confirmation of the index trend.

Conclusion

Tuesday shifted the breadth picture from straightforwardly positive to constructive but vulnerable. SPY, QQQ and IWM remain above their EMA21s and retain positive weekly trends, while a majority of stocks are still above each major moving average.

The short-term evidence is considerably weaker. Downside participation is broadening, larger decliners are becoming more common and weekly breadth has flipped negative. A quick recovery in the EMA9s and short-term participation would keep this episode contained; continued deterioration in the SMA50/SMA200 readings would point to a more meaningful loss of internal market support.

ChartMill Market Desk

This daily Market Breadth Report is prepared by ChartMill for informational purposes only and does not constitute investment advice. Always do your own due diligence before making investment decisions.

On August 17, Paramount Skydance filed a motion asking a US court to force twelve state attorneys general and the Writers Guild of America to post a bond of nearly $1.9 billion for challenging its $81 billion acquisition of Warner Bros. Discovery.

The stated reason: Paramount wants those challengers to be financially liable for the damages it’s incurring while the case is pending. Specifically, the $7 million per day “ticking fee” it owes to Warner shareholders if the deal isn’t closed by October 1. That works out to $650 million per quarter, and the trial doesn’t even start until March 2027.

Reading this news, I had one question. “Where would Paramount get $1.9 billion if the request for this bond were denied?”

Not from free cash flow. Even at the time of the original agreement, I already knew that far too much was at stake with that “ticking compensation structure.” After all, even then it was already clear that the balance sheet wouldn’t be able to support the aggressive terms that were being offered.

What I didn’t have back then was the analysis I have today. I do now.

The Balance Sheet Reality

I pulled the fundamental snapshot for PARAMOUNT SKYDANCE CL B (PSKY | ▲1.38%) from ChartMill this morning. The overall rating is 3 out of 10. Health specifically: 2 out of 10. The individual numbers tell the story.

• Start with the Altman-Z score of 1.00. That puts PSKY in the distress zone. The scientific threshold for elevated bankruptcy risk is 1.81, and at 1.00 you’re not near the line, you’re well past it.

• The debt-to-free-cash-flow ratio comes in at 32.25 years. That’s how long PSKY would need to pay back its debt from current free cash flow. Not a decade. Three of them.

• Return on invested capital sits at 5.34% against a weighted average cost of capital of 8.89%, which means every dollar the company puts into the business earns less than the cost of financing it. The ratio is 0.6. In plain terms, PSKY is destroying value with each investment cycle.

• Add the liquidity picture. Current ratio of 1.04, barely enough short-term assets to cover short-term liabilities, with no meaningful buffer.

• Debt-to-equity of 1.23, and worse than a year ago; the direction is wrong.

• Shares outstanding are higher than a year ago and higher than five years ago, so dilution is already ongoing before any new capital raise.

• EPS is down 84.96% over the last twelve months. The forward P/E is 12, the trailing P/E is 60. The market is pricing in a substantial earnings recovery. If that recovery slips because of legal costs or integration friction, the multiple compresses fast.

None of this is a matter of interpretation. These are reported numbers, from a company that closed the Skydance-Paramount merger in August 2024 and is now trying to close another $81 billion transaction on top of it.

Where the Money Could Come From

Back to the original question. If a judge did award the $1.9 billion bond, PSKY would have four options, and none of them are good.

• Draw down cash. With a current ratio of 1.04, this means immediate liquidity risk right before the largest financing operation in the company’s history.

• Issue new debt. On the current credit profile, and with more than $40 billion of acquisition financing already in motion, this pushes the rating toward junk. Higher coupons, tighter covenants, and less operational flexibility.

• Issue equity. Dilute existing shareholders (including Ellison and RedBird, who just recapitalized the company) at a stock price of $10.28. Painful, and a public signal that the balance sheet couldn’t handle the cost.

• Sell assets. Pluto TV, parts of CBS, or the film library. Which undermines the entire strategic logic of the WBD deal in the first place. You can’t argue for combining two studios to unlock synergies while selling off pieces to fund the litigation of the combination.

The motion itself is likely going nowhere. US courts almost never impose damage bonds of that size against state attorneys general or a labor union. But the fact that Paramount felt the need to file it says something about the internal pressure.

When management starts spending legal budget on aggressive procedural motions rather than closing the deal, something is off.

The Netflix Contrast

The Warner Bros. Discovery board originally chose the Paramount bid over Netflix’s, in part because of that ticking fee. It was pitched as a signal of commitment: pay us $7 million a day if you don’t close on time. Warner shareholders liked the certainty.

That same fee is now working against Paramount. And the deeper point is that NETFLIX INC (NFLX | ▼2.74%), on the other side of the table, walked away. Not because it couldn’t compete on price. Because it wouldn’t take on the balance sheet exposure to fund a deal at those terms.

Netflix generates roughly $9.6 billion in annual net cash. No integration debt, no ticking fee, no lawsuits from twelve states. When the Warner opportunity came up, the discipline was to say no rather than to match a bid carrying this kind of financial footprint. Decisions like that may not be the boldest, but that’s why it’s called “capital discipline.”*

Bottom line:

Paramount’s balance sheet was already stretched before the WBD deal. ChartMill’s diagnostic makes clear that PSKY sits in the distress zone by any classical measure, and this $1.9 billion bond motion is what a company under real financial pressure looks like when it starts using the courts as leverage against its critics.

Netflix chose not to be in this position. Six months ago that looked like a missed opportunity. Today it looks like one of the better non-transactions of 2026.

ChartMill Market Desk - Kristoff

Disclaimer: As of the time of writing, the author owns shares in Netflix.

Participation improved across nearly every timeframe, SPY pushed to another high and IWM finally broke above its recent range, while QQQ continued its recovery toward major resistance.

The breadth trend remains positive (6/7).

Index Overview (SPY, QQQ, IWM)

Short Term (Daily)

Long Term (Weekly)

SS SPDR S&P 500 ETF TRUST-US (SPY | ▲0.70%) continued to lead the broader market higher, gaining 0.7% to 777.88 and reaching an intraday high of 779.37. The ETF remains comfortably above its rising EMA9 at 768.37 and EMA21 at 759.44, while the former 760-area resistance has now been left clearly behind.

The weekly trend remains positive and Thursday’s price action further confirms the recent breakout rather than showing signs of immediate exhaustion.

INVESCO QQQ TRUST SERIES 1 (QQQ | ▲1.16%) gained 1.2% to 732.07, the strongest move among the three major ETFs. It has rebuilt considerable distance above its EMA9 at 717.63 and EMA21 at 710.74 after the late-July correction.

The next test is more obvious here than in SPY: QQQ is approaching the 740-749 resistance zone created by its June highs. The short-term trend is strong, but unlike SPY, it has not yet cleared that overhead supply.

ISHARES RUSSELL 2000 ETF (IWM | ▲0.26%) added a more modest 0.26% to 303.50, but its chart may contain the more meaningful development from a breadth perspective. The ETF closed above the recent 302.72 high and remains above its rising EMA9 and EMA21.

After spending much of the rally lagging SPY and QQQ, small caps are now participating more convincingly. The weekly ChartMill Trend Indicator also remains positive.

Compared with Wednesday, the index picture therefore improved in the area where confirmation had been most useful: IWM is no longer simply following higher inside its range; it is attempting a genuine breakout.

Breadth Strengthens Again

The internal numbers strengthened further from Wednesday’s already-positive readings.

Daily advancers increased from 59.7% to 63.4%, while decliners fell from 37.3% to 33.7%. Stocks gaining more than 4% accounted for 5.5% of the market compared with only 2.7% falling more than 4%, maintaining a healthy positive skew without suggesting indiscriminate speculation.

Moving-average participation improved across all four timeframes:

• Above SMA20: 66.0% → 71.1%

• Above SMA50: 58.7% → 61.8%

• Above SMA100: 56.6% → 57.9%

• Above SMA200: 52.5% → 53.3%

The SMA20 reading above 70% shows strong short-term participation, but the improvement is no longer limited to short-term measures. More than half of the market is now above each of the 20-, 50-, 100- and 200-day averages.

That makes the current advance substantially healthier internally than the late-July period, when strong index performance repeatedly conflicted with weak participation underneath.

New Highs, Pocket Pivots and Momentum

New highs rose sharply from 5.3% to 8.7%, while new lows slipped from 1.4% to 1.2%. The spread between the two is now decisively positive.

Pocket pivots also improved from 18.9% to 21.9%. That remains well below the kind of reading normally associated with highly speculative or overheated participation, but the direction supports the broader picture of more individual stocks developing constructive setups.

This combination is preferable to an index rally driven solely by a handful of mega-cap names: more stocks are above their moving averages, more are reaching new highs and fewer are registering meaningful downside pressure.

Participation Broadens Across Timeframes

The strongest change is visible beyond the daily numbers.

Weekly advancers jumped from 60.1% to 70.0%, against only 29.1% decliners. Monthly breadth improved to 60.8% advancing versus 38.6% declining, while the three-month measure reached 60.4% versus 38.8%.

That progression removes much of the short-versus-medium-term divergence that was visible earlier in the rebound. Short-term strength is increasingly being confirmed by the broader one- and three-month participation data.

The three-month extreme-mover figures remain constructive as well: 12.9% of stocks have gained at least 25%, compared with 8.9% declining by that amount.

Market Context

Thursday’s market backdrop was supportive without being uniformly bullish.

Cooler July producer-price data reduced some of the immediate rate pressure, while SanDisk’s Investor Day produced a sharp rally in NAND-related stocks including Western Digital and Micron. At the same time, Cisco and Cerebras suffered substantial declines despite their AI exposure, showing that investors are still differentiating heavily between individual companies rather than simply buying the entire growth complex.

The 30-year Treasury auction at 5.216%, the highest yield since 2001 according to the supplied market report, provided a notable counterweight. Yet the breadth figures strengthened despite that pressure from the long end of the bond market.

The Applied Materials results and Reddit’s after-hours index-related move occurred after Thursday’s closing breadth data and are therefore not used to explain the breadth readings in this analysis.

Breadth Trend Rating

6 - Positive

The assessment remains unchanged at positive, making August 13 the fifth consecutive trading session rated 6.

There is considerably more confirmation behind that rating now than there was several sessions ago. More than 70% of stocks are above their 20-day average, all four moving-average participation readings exceed 50%, weekly participation has reached 70%, new highs are expanding and IWM has joined the index breakout.

Breadth trend: Positive (6/7).

The main change from Wednesday is therefore not the rating itself, but the quality of the confirmation underneath it. Thursday produced another meaningful step toward broader participation, with small caps and medium-term breadth increasingly supporting what the major indexes are already showing.

ChartMill Market Desk

This daily Market Breadth Report is prepared by ChartMill for informational purposes only and does not constitute investment advice. Always do your own due diligence before making investment decisions.

Just as the main indices finished modestly lower on nervous positioning ahead of Wednesday’s CPI print, two AI-infrastructure heavyweights dropped quarterly numbers that told the same story from different angles: the spending cycle behind this cycle is not slowing down.

Meanwhile, the oil market kept pushing higher on a fourth consecutive session of Hormuz stress, and Washington quietly negotiated a pause in Ukraine’s drone campaign against Russian oil tankers to try to steady the barrel. A busy setup ahead of a data-heavy Wednesday.

The Rundown

• After-hours earnings from two AI-infrastructure players confirmed the capex cycle still has real legs.

• The main indices drifted lower for a second session as traders waited on the July inflation report.

• Oil climbed for a fourth straight day, with global fuel inventories tightening further.

• Washington intervened to reroute Ukraine’s drone campaign away from Russian oil tankers.

The Session Nobody Wanted to Own

Tuesday was one of those tapes where nobody wanted to take a strong view before the July CPI print.

Consensus is looking for headline inflation of 0.1 percent month-on-month after June’s 0.4 percent drop, with core prices firming by 0.2 percent. On a year-over-year basis, headline CPI is expected to cool from 3.5 to 3.4 percent and core from 2.6 to 2.5 percent.

Sounds benign in isolation, but with WTI having tacked on close to 11 percent over the last four sessions, the market knows the real question is what oil does to inflation expectations, not what already printed for July.

The 10-year Treasury yield eased slightly on the session and the euro traded around 1.1544 against the dollar. Small-business optimism ticked higher in July, while existing-home sales moved the other way as mortgage rates continue to chew into affordability. None of it moved the needle much, because everything is subordinated to Wednesday’s data.

Super Micro Delivers a Genuine Blowout

The after-hours highlight was SUPER MICRO COMPUTER INC (SMCI | ▲0.45%).

The AI-server specialist posted adjusted EPS of $1.70 on $11.1 billion in revenue for its fiscal Q4, against a Street looking for $0.92 on $11.6 billion.

Revenue undershot slightly but earnings almost doubled the consensus, and the margin story is where things got interesting. Gross margin recovered from 9.5 percent a year ago to 17.5 percent, and adjusted margin followed suit at 17.6 percent. Net income for the quarter more than doubled sequentially and multiplied roughly sixfold year-on-year to $1.18 billion.

The guide is where the real “wake up and pay attention” number sits.

Management is pointing to fiscal Q1 revenue of $14.5 to $15.5 billion - the Street had $11.82 billion - and full-year revenue of $65 to $72 billion against a consensus closer to $53 billion. That is not a minor rerating of expectations, that is a full re-pricing.

CEO Charles Liang flagged over $60 billion in new orders during the past year and a record backlog to open fiscal 2027. This is exactly the sort of print that reminds you why the AI-infrastructure trade keeps refusing to die on cue.

CoreWeave: The Top Line Impresses, the Bottom Line Deserves a Look

Then came COREWEAVE INC-CL A (CRWV | ▲2.42%), and the reaction says everything.

• Revenue growth of 112 percent for the second quarter,

• guidance for further acceleration to $3.45–$3.60 billion in the current quarter, the upper end well above where Bloomberg’s consensus was clustered.

The stock immediately shot 13 percent higher after hours, and I understand why: on the top line, this is a beat and a raise on top of a growth story that already looked white-hot.

But I would encourage a closer look at the rest of the P&L before getting swept up...

Strip out roughly $1.5 billion in infrastructure investment and $640 million in interest expense, and you land at a net loss of $626 million, or $1.14 per share. That is more than double the loss of a year ago.

Financial debt has climbed to over $35 billion from $21 billion a year earlier, with an additional $16 billion or so in lease obligations and nearly $10 billion of deferred revenue sitting on the balance sheet, cash already collected against services still to be delivered.

When CoreWeave IPO’d in the spring of 2025, the debt profile was already a debate. That debate has not gone away; it has just been drowned out by the top-line noise.

Intel Cranks the Raise to $20 Billion

INTEL CORP (INTC | ▲0.19%) confirmed: the equity issuance is being upsized from $15 to $20 billion. That is a meaningful bump, and it tells you the book was comfortably oversubscribed.

The stock only drifted marginally lower on the news, which for a raise of that size is a strong tell about how the recovery narrative is being received.

Elsewhere in semis, the tape was mixed.

NVIDIA CORP (NVDA | ▼0.02%) drifted fractionally lower even after reports that it is pulling together a $500 billion AI investment vehicle to be executed with large investors and banks, the follow-up story to Monday’s report.

ADVANCED MICRO DEVICES (AMD | ▲1.01%) and BROADCOM INC (AVGO | ▼1.50%) finished split, in what you might call a “no-conviction” session for the chip complex.

When even a $500 billion capital construct barely moves Nvidia, you know the market is fatigued of the circular-financing debate and simply wants to see Wednesday’s inflation print before doing anything decisive.

Riot, Rocket Lab, and the Rest of the Roster

The most interesting individual-stock story of the day, in my view, was RIOT PLATFORMS INC (RIOT | ▲4.33%). The bitcoin miner has struck a $9 billion cloud deal with Anthropic, a genuinely strategic pivot for a company still primarily known for hashing power.

Miners with credible AI-compute pipelines are quickly becoming a distinct sub-category, and Riot is now firmly in that camp. Worth watching how the peer set reprices around it in the coming days.

ROCKET LAB CORP (RKLB | ▼0.04%) posted a bigger quarterly loss on growing revenue and announced plans to open a German operation, a smart expansion given Europe’s push for sovereign space capacity.

Sector peer AST SPACEMOBILE INC (ASTS | ▲4.17%) also posted a wider loss on revenue that fell short of expectations and still climbed 4 percent, which tells you how much of the space-connectivity trade is being priced on optionality rather than near-term numbers.

SPACE EXPLORATION TECHN-CL A (SPCX | ▼3.93%) dropped back below its June IPO price of $135, closing around $133, a symbolic level that will attract attention if it sticks.

On the negative side of the ledger, APPLE INC (AAPL | ▼1.09%) shed just over a percent following the announcement that Jennifer Bailey, head of Apple Pay, is leaving the company.

HIMS & HERS HEALTH INC (HIMS | ▼3.97%) sold off almost 4 percent after a bigger-than-expected Q2 loss that even a raised outlook could not offset.

TRUMP MEDIA & TECHNOLOGY GRO (DJT | ▼5.11%) reported after Monday’s close that its quarterly loss had ballooned tenfold on digital-asset markdowns, and Tuesday brought another 5 percent slide on top of Monday’s 8 percent drop.

The Hormuz Situation Just Got More Complicated

Oil is telling its own story. WTI added over 1 percent Tuesday evening after Monday’s near-5 percent gain, extending a run that has lifted a barrel almost 11 percent over four sessions.

Sentiment briefly softened on comments from Pakistan suggesting that Washington and Tehran were close to some form of understanding on the Strait, but the hardened rhetoric on both sides makes any near-term breakthrough look unlikely. Traffic through Hormuz remains sharply reduced and global diesel inventories are visibly tighter.

Against that backdrop, one of the more interesting geopolitical wrinkles: Ukraine has paused its drone campaign against oil tankers using the Russian Black Sea port of Novorossiysk, at the direct request of Vice President J.D. Vance.

Washington grew concerned that Kyiv’s strikes were destabilising oil markets and hurting US commercial interests, particularly around the Caspian Pipeline Consortium’s terminal that handles Kazakh crude.

Ukraine has agreed to steer clear of CPC infrastructure and non-Russian vessels, provided those vessels aren’t carrying Russian cargo or already under Ukrainian sanctions.

Kazakh oil production fell 14 percent in July as those strikes bit, so the pause has real market implications. Worth watching whether it holds.

Conclusion

Tuesday was a tape of two very different personalities.

The regular session was cautious, drifting sideways-to-lower on CPI nerves. The after-hours session, by contrast, delivered a pair of AI-infrastructure prints that will feed the bull case for weeks.

Super Micro’s guide is the kind of number that resets a peer group; CoreWeave’s growth pace is undeniable, even if the balance sheet still deserves scrutiny. Meanwhile, oil is quietly compounding higher and Wednesday’s inflation print will land on a market already trying to work out how quickly Hormuz stress feeds through to prices.

I would keep a close eye on how the AI-infrastructure enthusiasm from tonight interacts with tomorrow’s CPI reading, the setup could either extend the risk-on tone or complicate it very quickly.

ChartMill Market Desk - Kristoff

With regard to the stocks discussed in the article above; the author owns individual shares in Nvidia.

This daily update is prepared by ChartMill for informational purposes only and does not constitute investment advice. Always do your own due diligence before making investment decisions.

👉 This article is also available at ChartMill.com.

Recently, I received an email following a few posts I shared on X:

“How do you always manage to find those clean setups? Doesn’t it take a massive amount of time to scroll through all those charts?”

The short answer to that second question is “no”. Finding solid swing trade setups isn’t difficult at all and only takes a few minutes.

How do I do it in ChartMill? I’ll break it down for you short and sweet below.

The person who sent me that message was referring to this setup in Ares Management (ticker: ARES). Here is the post I shared on X:

The bullish flag pattern you see on the chart - formed by the two purple lines - is automatically recognized and drawn by ChartMill.

Are these bull flag patterns not showing up on your chart?

In the stock screener, head over to the ‘Chart Settings’ tab. Click the yellow plus icon next to Main Chart, then click ‘Select Overlay To Add’. Choose ‘ChartMill Flag’ from the drop-down menu and finish by clicking ‘Add Indicator’.

Here’s step-by-step how I found this setup:

Step 1 – Trading Idea

On the homepage main menu (left), click on ‘Trading Ideas’.

Once you’re on the ‘Trading Ideas’ page, select the following options in the filter fields at the top:

On the left, you’ll now see the ‘Bull Flags’ trading idea. Click ‘RUN’, and it will automatically take you to the stock screener page.

Step 2 – Setting additional basic filters

In a bullish market environment, you’ll quickly end up with more than 100 results.

Let’s refine that a bit further.

I always apply the following core filters:

Under the ‘General’ tab:

- Price: > $5 (to avoid stocks priced too low)

- Exchange: US Only

- Market Cap: At Least Small

- Type: Common Stocks Only

Under the ‘TA Indicators’ tab:

- Volatility: ADR(20)% Above 3 (we only want stocks with sufficient movement).

At the time of writing, that narrows the number of setups down to 54.

Going through these takes just a couple of minutes. You can add interesting candidates directly to a watchlist by clicking the ‘+’ icon in the top right corner of any chart.

Step 3 – Saving your bull flag screen

Want to use this screen again later? Save it with your current parameters by clicking the floppy disk icon in the top right corner. Give it a name and confirm with ‘Save’.

Whenever you want to run this screen again in the future, you can do so directly within the stock screener. Click the folder icon and select your saved bull flag screen from the drop-down list.

Finally, using ARES as an example, here is a potential setup structured with the stock position size calculator.

Please note: this is for educational purposes only and not a recommendation to buy the stock!

Based on the parameters you enter in the top fields, this tool calculates the exact number of shares you can (and, from a risk management standpoint, should) buy.

Any excessive risk parameters will be flagged with a red X instead of a green checkmark.

You can read more about this very handy tool here.

Trade Safe!

Kristoff - ChartMill

The rule in one sentence

The 7-8% sell rule says this: never let a loss on any stock exceed 8% below the price you paid for it. No exceptions, no averaging down, no “waiting for the bounce”. The stop is set the moment you buy, and if it triggers, the position is closed the same day.

William O’Neil laid the rule out in ‘How to Make Money in Stocks’ and repeated it in almost every interview he ever gave. It is not a suggestion. In the CANSLIM framework, it is the single mechanical rule that keeps the strategy alive during the periods when the other seven letters are working against you.

If you have already read the anchor exit article on ChartMill - Sell Rules for Growth Stocks — When to Get Out - you have seen the rule referenced briefly.

The CANSLIM screener documentation mentions it in passing.

This article is the deep dive: where the number comes from, what it does not mean, and how to actually apply it on a live position.

Who this rule is (and isn’t) for

The 7-8% rule belongs to a specific style of investing: actively managed growth and momentum, where positions are entered on technical breakouts and the timing of the entry is half the edge. That is the context O’Neil wrote it for, and it is the context in which the rule is non-negotiable.

It does not apply to every strategy. A long-term quality portfolio - Terry Smith, Charlie Munger, the classic “buy wonderful businesses and hold them” school - operates on completely different logic. Entries are made on fundamental grounds (maybe even refined with a bit of technical timing like I do myself) and exits happen only when the quarterly or annual numbers invalidate the original thesis. In that framework, a 15% price drawdown on a stock whose fundamentals are still intact is not a sell signal, it is often an invitation to add. Averaging down at a lower price on a preserved thesis is a feature of the strategy, not a mistake.

Both approaches work. What does not work is mixing them: applying growth-style entries with quality-style holding, or quality-style conviction with growth-style position sizing. Pick a lane. If you’re in the growth lane, the 7-8% rule is the mechanical guardrail that keeps the strategy alive. If you’re in the quality lane, this article is background reading, useful to understand, not meant to be applied.

Where the 7-8% number comes from

O’Neil did not pull the number out of a hat.

His research team at Investor’s Business Daily studied decades of the biggest winning stocks in the US market and looked at what happened around their eventual failed setups. The pattern was consistent: the best CANSLIM stocks, when they broke out from a proper base and were bought at a proper pivot, rarely retraced more than 7-8% before resuming the trend.

Anything worse than that was, statistically, an early tell that the setup had failed.

The number is anchored in the recovery math too. A 7% loss requires a 7.5% gain to break even. A 25% loss requires 33%. A 50% loss requires a 100% gain, you need to double your money just to get back to where you started. The 7-8% ceiling exists because it is the largest single-position loss you can absorb without the recovery math turning against you.

The point of the table is not that a 50% loss is bad, everyone knows that. The point is that the difference between an 8% loss and a 15% loss is not “a bit worse”. It is more than double the ground you need to make up.

The math is not linear. The 7-8% cap sits at the last point on the curve where recovery is still cheap.

Measured from the buy price, not from the pivot, not from the high

The 7-8% is measured from the price you actually paid. Not from the pivot point. Not from the intraday high. Not from your average cost after adding to the position. From your entry fill.

Take a concrete example.

$BPOP breaks out of a base with entry filled at $153.08. The O’Neil ceiling sits at $140.83, 8% below entry. But the daily ATR on $BPOP is $3.93, so a 2×ATR stop lands at $145.22, which corresponds to $7.86 of risk per share, or 5.13% below entry. That is the operative stop.

It is well inside the 8% ceiling, and it is anchored on the stock’s own volatility rather than on a round number. The Position Size Calculator confirms the setup: on a $25,000 account with 1% risk per trade, that stop distance produces a 31-share position sized at roughly $4,745, about 19% of capital, with total dollar risk of $243.66, right at the intended 1%.

The 8% ceiling never came into play. It did not need to.

That is how it is supposed to work: the ATR-anchored stop is tighter, so the ATR-anchored stop wins. The ceiling exists for the sessions when a trader is tempted to widen a stop past 8% because “the stock is just volatile” and that is exactly when the ceiling refuses.

If the stock runs to $135, then rolls over and comes back, the 7-8% rule is no longer the operative exit, a stock that has moved 14% in your favour and then given it all back is a different signal, covered in the trailing-stop and profit-protection sections of ‘Sell Rules for Growth Stocks’.

The 7-8% rule is about the initial stop only. It is the maximum acceptable loss on a fresh position that has not yet had a chance to work. Everything after that first move belongs to the trailing-stop framework.

The rule is a ceiling, not a target

Here is the part where a lot of traders get it wrong in the other direction: 7-8% is the maximum, not the standard. It is the widest stop you are allowed to use, not the stop you should always use.

The right stop distance for any individual position is whichever is tighter:

7-8% from your buy price (the O’Neil ceiling), or the stop that keeps your total account risk on the trade at whatever percentage you have decided in advance (typically 0.5% to 1.5% of account value per trade).

On a low-beta large-cap breakout, those two often converge. On a high-beta small-cap, the account-risk stop is almost always tighter, and it is the one that governs. The 7-8% ceiling is what stops you from ever widening a stop past that point on the argument that “the stock is just volatile”.

ChartMill’s Position Size Calculator is the tool that runs this arithmetic directly. Enter your account size, your risk-per-trade percentage, and the entry and stop prices, and it returns the share count and the total capital at risk. The workflow is: pick the tighter of the two stops first, then size the position around that. Not the other way around.

The stop distance itself should be anchored to the stock’s own volatility, not to a round percentage number. The Average True Range (ATR) indicator is the standard input.

common rule of thumb: the stop should sit at least two times the current daily ATR below the entry, so normal intraday noise does not take the position out on a session where nothing is actually wrong. If a 2×ATR stop already lands you inside the 7-8% ceiling, use it. If 2×ATR would put you past 8%, the position is too volatile for the size you were planning, either shrink the size, or skip the trade.

Going tighter than 2×ATR is a legitimate choice, but a deliberate one.

A stop at 1.5×ATR or even 1×ATR keeps the same dollar risk on a smaller price move, which means a larger share count and a bigger initial position, fine on a setup where the structure genuinely justifies a tight stop (a well-defined pivot, a tight consolidation, a low-volatility breakout). The trade-off is a much higher probability of being stopped out on normal intraday noise. 1×ATR is roughly the floor: anything tighter than that is not really a stop, it is a coin flip on the first tick against you.

The Position Size Calculator raises a warning below 2×ATR for exactly this reason, treat it as a prompt to check whether the tightness is deliberate, not as an automatic block.

Why the rule stays intact even when it “feels” wrong

The two most common reasons traders talk themselves out of the 7-8% rule on a specific trade are always the same, and they both sound reasonable in the moment.

“The fundamentals are still great.” They usually are. That is exactly why you bought it.

A stock breaking down 8% from a proper pivot is not doing so because earnings suddenly deteriorated, it is doing so because the technical setup has failed, and the technical setup was half the reason for the trade. Cutting the loss on a fundamentally strong company you still like does not mean you are done with the name. It means this specific entry was wrong. You can re-enter on the next valid setup, at a better price, with the risk reset. The rule does not stop you from owning the stock; it stops you from owning it from the wrong price.

“It will bounce right back.” Sometimes it does.

Most of the time it does not and even when it does, the entries that trigger the 7-8% stop and then reverse are the same entries that produce the biggest disasters when they do not reverse. This is exactly the situation the rule was written for. If your rule bends the first time it feels inconvenient, it is not a rule. It is a preference. Preferences do not survive drawdowns.

There is also a survivorship bias in the “it always bounces back” argument. Traders remember the times they held through an 8% drawdown and got saved. They forget the times they held, watched it slide to 15%, then to 25%, then quietly stopped checking the account. Those positions do not make it into the mental highlight reel. The 7-8% rule is designed for the second category, which is far larger than the first.

The averaging-down trap

Averaging down on a growth stock that has hit its 7-8% stop is a specific mistake worth naming, because it is dressed up as a virtue. You bought at $118. It is now at $102. Buying more at $102 lowers your average cost, so the position “only” needs a small rally to be back at breakeven.

That framing is wrong for two reasons.

First, you are increasing the size of a position that has already told you the setup was wrong. The stock crossed the line you drew at purchase. You are not “buying the dip”, you are doubling exposure to a technical failure.

Second, you are turning a small, manageable loss into a larger position sized around a lower average cost, which means the dollar loss if the stock continues to slide is now substantially bigger than the original single-position loss would have been. Your average cost went down, your total risk went up. That is a bad trade dressed as a good one.

O’Neil’s phrasing was blunt: cut all losses at 7-8% without exception, and never average down on a losing position. Not because averaging down is always wrong in every context - value/quality investors do it deliberately, and can defend it - but because averaging down inside a growth strategy is a category error. The two systems are not compatible.

In a bad market, the rule tightens

The 7-8% cap is calibrated for normal market conditions. It is not calibrated for the periods when growth as a factor is being distributed en masse. In those regimes, the same setup that would have worked six months earlier fails at a much higher rate, and holding to 8% on every position means eating the maximum drawdown on a much larger share of trades than usual.

This is where the “M” in CANSLIM comes in, the market direction letter that most retail readers skim past. O’Neil’s own guidance was to tighten stops in a weakening market environment, typically down toward 3-5% instead of the full 7-8%. Some CANSLIM disciplines go further and pull growth exposure back to zero when the market is in a confirmed correction.

The specific signals that define “confirmed correction” - distribution days, follow-through days, breadth deterioration — sit in the ChartMill Market Breadth (CMMB) documentation and are addressed at portfolio-level in ‘Risk Management for Growth Stocks — What’s Different?’.

The point for the sell-rule article is narrower: the 7-8% number is a maximum for normal conditions. When the regime shifts, the maximum shifts with it. A trader running the rule mechanically at 7-8% through a full-blown correction is being disciplined about the wrong thing.

How to set the stop on ChartMill

The mechanics on ChartMill are direct.

1. Compute the stop before you place the order. Use the Position Size Calculator. Enter account size, risk-per-trade percentage (start with 1% if you have no calibration), entry price, and stop price. Take the tighter of two candidates: 8% below entry, or the ATR-anchored stop.

2. Check the ATR distance. Pull the ATR indicator on the daily chart. If 2×ATR is already inside the 8% ceiling, use 2×ATR. If 2×ATR blows through 8%, the stock is too volatile for the position size you were planning, cut the size or skip the trade.

3. Place the stop with the order. Not as a mental note. Not as an alert. As an actual stop order at the broker, sized for the full position, active from the moment the entry fills. Mental stops fail during the exact price action the rule was written to protect against, the fast, ugly, out-of-nowhere gap-and-slide morning.

4. Do not move the stop against yourself. The initial stop moves in exactly one direction; up, as the position works. It never moves down. If the stop gets hit on the first day because the breakout immediately reversed, that is not a bug. That is the rule doing its job.

What the rule is not

The 7-8% rule handles exactly one situation: the initial-stop loss on a position that never worked. It does not handle:

• Trailing stops on profitable positions. Once the stock is well above the entry, you are in trailing-stop territory. The framework for that is in ‘Sell Rules for Growth Stocks’ — breaks of the 50-day moving average, climax patterns, CRS deterioration, head-and-shoulders tops.

• Profit-taking on winners. Related but different problem. O’Neil’s guidance was to sell partial size into strength in the 20-25% profit range on most positions, with the exception of stocks showing the specific characteristics of a much larger move. That framework belongs in a separate treatment.

• Position-level sizing and portfolio-level risk. The 7-8% rule tells you where to exit; it does not tell you how large the position should be in the first place. That layer sits upstream, in ‘Risk Management for Growth Stocks — What’s Different?’ and the underlying ‘Swing Trading Risk Management guide’ it builds on.

• Regime risk. Covered in the same risk-management article. In a bad regime, tighten the rule, or step aside entirely.

Treating the 7-8% rule as the whole exit framework is another common mistake. It is one line of defense - the first one - and it needs the others to actually work.

Bottom line

The 7-8% sell rule is boring, mechanical, and unglamorous. That is exactly why it works.

It does not require you to be right about the market. It does not require a view on the sector. It does not require any judgement in the moment, the number is fixed as a maximum distance the second you place the order. Its entire job is to remove your ability to make a bad decision when the position is going against you and your instincts are the least reliable input available.

Every trader who has stayed in this game for a decade or more has some version of this rule. The specific number differs - 6%, 7%, 8%, occasionally slightly wider for lower-beta names - but the structure is the same: a hard, pre-committed, non-negotiable exit set at the moment of entry. O’Neil’s contribution was to publish the number, defend it in print for forty years, and refuse to add caveats.

If you take one rule from CANSLIM and apply it consistently, take this one. Everything else in the framework is easier to fix later. This one, if you get it wrong, is the reason people quit.

Trade Safe!

Kristoff - ChartMill

I’ve tried day trading a few times over the years. Never stuck. I do fine with swing trading and my long-term portfolio, so it’s not a knowledge gap. Position sizing and risk management matter just as much in swing trading as they do intraday, and I know that side of the game well.

My problem is different. Day trading moves too fast for me. There’s no gap between sessions to sit down, screen calmly, work through the charts, and have orders ready for the next day. That pause is where I do my best thinking, and day trading doesn’t give it to me.

That’s part of why we asked Cory Mitchell, a Canadian day trader (CMT), to write a guest article for ChartMill. We knew he was already using our screener to find candidates, and it felt like a good match: a real trader, showing his actual process, on the tool we built.

What struck me in his approach wasn’t some clever new indicator. It was the opposite. Cory’s screen comes down to two numbers: volume and ADR% (average daily range, a pure measure of how much a stock moves intraday, stripped of overnight gaps). That’s close to the whole thing.

And because ADR% doesn’t swing wildly from one day to the next, he doesn’t scan daily. Once or twice a week is enough to build his short list for the days ahead, which, ironically, gives him exactly the kind of breathing room I can’t seem to find in day trading.

He laid out the full method, including how he narrows the list down to the four or five stocks he actually watches for the week, here:

Read the article: How to Find Top Canadian Day Trading Stocks (Or Other Markets)

We built the two screens straight from the article, ready to run:

Canadian Day Trading Screen by Cory Mitchell, CMT — CAD stocks only.

US Day Trading Screen by Cory Mitchell, CMT — US stocks only, with tighter volume and ADR% thresholds since the US universe is so much bigger.

Bottom line: worth running this weekend if intraday trading is on your radar, and worth studying even if it isn't. Simple, well-designed screens have a way of teaching you something you didn't expect.

Trade Safe!

Kristoff - ChartMill

The story wasn’t a single macro shock.

It was a stack of company-specific reactions all pointing in the same direction: the market is done paying up for AI-adjacent names that merely beat consensus.

SpaceX’s first-ever quarterly report as a listed name took the hardest hit despite a profit beat, Alphabet lost its most senior AI engineer to a startup you had not heard of yesterday, and AMD got knifed on a chip-supplier decision from Elon Musk. Let me walk you through it.

The Rundown

• The Nasdaq gave back gains as post-earnings reactions in the AI complex turned selective, while the Dow held firm on cyclical strength.

• A newly public space-and-satellites name took a double-digit hit on its inaugural quarterly print despite beating profit estimates, heavy AI capex and an incoming lock-up expiry did the damage.

• A major cloud-and-search platform lost a senior AI executive to a new startup, feeding fresh doubt about competitive positioning in the AI race.

• Oil stabilised near $79 on progress toward a Hormuz agreement, and a soft private-payrolls print shifted attention toward Friday’s employment number.

The AI-Capex Tape Just Repriced Itself

Wednesday’s biggest story was not one stock, it was the market deciding, in real time, that beating consensus is no longer enough when you are trading on hyperscaler-tier expectations.

SPACE EXPLORATION TECHN-CL A (SPCX | ▼13.61%) took the harshest read.

The first quarterly report since June’s IPO delivered a profit beat, but the number that mattered to the tape was the size of the AI-infrastructure spend, and management’s willingness to keep running at that rate.

Beneath the reaction there is a second overhang: Thursday marks the end of the first lock-up window, releasing roughly $101 billion in employee and early-investor shares into the market. When you combine a stretched capex disclosure with imminent supply hitting the tape, a 14% drop starts to look almost measured.

What to remember from SpaceX’s first earnings report (a plain-language guide for beginner investors)?

ADVANCED MICRO DEVICES (AMD | ▼7.04%) got hit from two directions. The Tuesday-night print beat and guided above consensus, but the sell-side ranked it “excellent, not exceptional”, the same read I flagged in yesterday’s note.

Then Musk made it worse by disclosing that SpaceX will source AI-infrastructure silicon exclusively from NVIDIA CORP (NVDA | ▲3.43%), citing the new Vera Rubin architecture as the technology of choice.

That single decision closes a marquee reference customer for AMD’s Instinct line and moved Nvidia up 3.4% on the same headline. This is a zero-sum reallocation dressed up as a technology comment, and the market priced it that way.

Alphabet Loses A Founding-Era Engineer

ALPHABET INC-CL A (GOOGL | ▼4.03%) shed 4% on news that Jeff Dean - at Google since 1999 and arguably its most influential engineer - is leaving to found a startup called Discovery Loop, taking three colleagues with him.

On the same day, Demis Hassabis stepped back from operational leadership of Google DeepMind, moving into a chairman role at the AI lab and a chief-scientist title at Alphabet. Koray Kavukcuoglu takes over day-to-day.

The market read this as a leadership vacuum at exactly the moment the AI race is hardening around OpenAI and Anthropic. Alphabet is putting a good face on it, the company is investing in Discovery Loop and becoming its cloud partner, but a leadership reshuffle plus a marquee exit is not the signal you send when things are going well internally.

I am not calling a thesis break here; Alphabet’s compute and data advantages are structural. But you cannot pretend this is a nothing-burger either. The overhang is real and will need at least one quarter of solid execution to lift.

Disney Delivered The Cleanest Print Of The Day

WALT DISNEY CO/THE (DIS | ▲3.65%) climbed 3.7% on the kind of quarter that reminds you why the operational-execution playbook still matters. Fiscal Q3 EPS came in at $2.06, well above the $1.86 consensus, a 28% jump from the year prior. Revenue landed at $25.2 billion, up 7%, though slightly below the $25.4 billion analysts were modelling.

The Consumer Products segment is where I would spend the most time. It grew 7% to $1.1 billion, driven meaningfully by “Toy Story 5” merchandise. The film opened June 19 and has already cleared $1 billion in global box-office receipts. That is the kind of franchise flywheel Wall Street stopped giving Disney credit for two or three years ago, and it is quietly grinding back into the numbers.

Management reiterated the fiscal 2025/2026 guide of roughly 12% EPS growth and pre-guided double-digit growth for the following year starting in October. That combination - a clean beat, a franchise tailwind and a re-affirmed multi-year growth trajectory - is what a boring, well-run print looks like. It is also becoming rare enough to be worth noting.

Memory Names Delivered A Warning Shot

Two after-hours prints from the memory complex told the same story: current-quarter results are fine, forward guides are the problem.

SANDISK CORP (SNDK | ▼5.40%) saw revenue quintuple and profit swing to $6.9 billion (from a $23 million loss a year prior), riding the ongoing memory-price spike.

The next-quarter guide of $10.3 to $10.8 billion undershot the $11.2 billion consensus, and the stock dropped 8% after the bell.

WESTERN DIGITAL CORP (WDC | ▼5.36%) delivered a 44% revenue lift to $3.75 billion and adjusted EPS of $3.56, both above consensus, but its Q1 revenue guide of roughly $4.1 billion only matched the sell-side model, and the stock dropped another 10% after hours on top of its 5% loss during the regular session.

The message across the group is that the memory upcycle is still real, but the market has priced a steeper ramp than the companies themselves are willing to underwrite.

Elsewhere In Company News

ELI LILLY & CO (LLY | ▲4.86%) beat Q2 on both lines and raised its 2026 outlook. The GLP-1 franchise continues to be the dominant driver, and the raise removes one of the near-term concerns that had been building around volume-versus-price dynamics in the class.

ARISTA NETWORKS INC (ANET | ▲3.57%) posted a strong quarter as AI-driven demand for high-performance networking gear continues to pull through the order book.

UBER TECHNOLOGIES INC (UBER | ▼5.29%) lost 5.3% despite CEO Dara Khosrowshahi highlighting the strongest new-user growth in five years.

PINTEREST INC- CLASS A (PINS | ▼8.68%) dropped 8.6% after a quarterly loss driven by higher costs, including a reorganisation charge. Reorg noise is usually forgivable if the underlying user metrics are intact; the reaction suggests the market is not yet convinced they are.

EBAY INC (EBAY | ▲1.40%) closed up 1.4% and traded slightly higher in the after-hours after a Q2 beat - adjusted EPS of $1.60 versus $1.50 consensus, revenue up 15% to $3.1 billion - but the forward EPS guide of $1.36 to $1.42 came in below the $1.45 FactSet consensus.

Modest positive reaction overall, which tells you positioning was not stretched going in.

The mobile-carrier and internet-provider names took a broader hit on SpaceX’s disclosure that it intends to take meaningful share via direct-to-cell satellite connectivity. This is the second-order competitive threat to the traditional telco stack, not a today problem, but one worth tracking as the Starlink service tiers expand.

Macro: ADP Miss Sets Up Friday, Hormuz Steadies Oil

The ADP employment report showed just 44,000 private-sector jobs added in July, half of June’s 98,000 pace and well below the 75,000 economists were expecting. Wage growth for job-changers stayed firm, which softens the read a bit, but the headline is a soft print heading into Friday’s non-farm payrolls number. That is the swing input for the September FOMC.

Oil steadied. Brent traded down 0.2% to just above $79 after briefly touching $81 during the session, on progress toward a Hormuz reopening, President Trump flagged that a deal with Iran could land Wednesday or Thursday, and Iran confirmed an agreement with Oman on a possible route through the strait. Roughly one-fifth of global oil trade moves through that waterway, so any durable normalisation is a genuine tailwind for the inflation picture.

The broader earnings backdrop stayed constructive. Roughly 80% of S&P 500 companies are beating expectations this season versus the typical 73%, which is the read that lets the market lean less heavily on the mega-cap tech complex and still find bid elsewhere.

The euro/dollar closed at 1.1552.

Conclusion

The Dow (DJI | ▲0.49%) closed higher, the S&P 500 (SPX_X.SP | ▼0.17%) sat flat, and the Nasdaq (COMPX | ▼0.83%) gave up ground on post-earnings selling in AI-adjacent names.

The signal underneath is that after four record-adjacent sessions, the market is willing to punish anything that merely meets the hype bar. SpaceX and AMD got that treatment in the harshest form; Alphabet caught a governance-and-talent overhang on top; Disney and Eli Lilly earned their bids the old-fashioned way.

Friday’s payroll number is the next input that materially reshapes the setup, until then, breadth is doing more work than the mega-caps, and that is a healthier tape than the one we came into the week with.

ChartMill Market Desk - Kristoff

With regard to the stocks discussed in the article above; the author owns individual shares in Nvidia.

This daily update is prepared by ChartMill for informational purposes only and does not constitute investment advice. Always do your own due diligence before making investment decisions.

👉 This article is also available at ChartMill.com.

SPACE EXPLORATION TECHN-CL A (SPCX | ▲9.43%)‘s first-ever public quarter looks strong on the surface: revenue nearly doubled, the operating loss shrank, and Adjusted EBITDA tripled. But three details deserve attention before drawing conclusions.

• Starlink is genuinely profitable and scaling fast, though revenue per user is dropping sharply.

• Space, the historical flagship, is actually losing external customers and burning cash on Starship development.

• AI suddenly delivers billions in revenue, but nearly 20% depends on a single anonymous customer whose contract can be cancelled on 90 days’ notice.

Add a capex pace that could exhaust the $100 billion cash pile in roughly 18 months, and you get a company priced on story more than on cash flow.

Below: what each segment actually did, what the headlines skipped, and the metrics to watch each quarter from here.

The key numbers, in plain English

For the quarter ending June 30, 2026, SpaceX reports:

• Revenue: $7.81 billion, nearly double a year earlier ($4.07 billion). Revenue is what the company brings in from sales, before any costs are subtracted.

• Operating loss: $(143) million, versus a loss of $(970) million a year earlier. This is the loss from day-to-day operations — from running the business itself.

• Net loss: $(541) million, versus $(1,008) million. This is what actually remains at the bottom of the page, after all costs, interest and taxes.

• Adjusted EBITDA: $3.54 billion, nearly triple. EBITDA is a popular profit metric that ignores depreciation (the accounting spread of the cost of machines and buildings), interest expense, and taxes. “Adjusted” means the company also strips out other items it considers one-off. Useful for spotting trends, but it is NOT cash.

• Capital expenditures (capex): $18.37 billion in a single quarter. Capex is money SpaceX sinks into physical assets: rockets, satellites, data centers, buildings. Of that amount, $15.83 billion went to the AI division alone.

At first glance an impressive quarter: revenue nearly doubled, loss cut in half, EBITDA sharply up. But anyone looking only at the headlines misses the real story. Under the hood the picture is more nuanced, and three things deserve extra attention.

What is actually happening in each segment

SpaceX today consists of three divisions: Connectivity (satellite internet via Starlink), Space (rockets and launches), and AI (artificial intelligence, since the merger with xAI). We’ll walk through them one by one.

Connectivity (Starlink): the goose laying the golden eggs

This is currently the only part of SpaceX that turns a profit. Revenue grew 66% to $4.29 billion, with operating income of $1.66 billion. The number of subscribers doubled to 12 million. Strong performance.

But there’s a catch.

ARPU (Average Revenue Per User, the average revenue per customer per month) fell from $85 to $66. That’s a 22% drop. SpaceX attributes this to international expansion and the introduction of cheaper subscription plans. That’s a defensible strategic call - trading margin per customer for more customers - but it’s something to watch closely. If network costs rise faster than new customer growth, margins will come under pressure over time.

The enterprise segment (businesses, governments, airlines) grew 108% to $1.81 billion. Contracts with American Airlines, Southwest, Virgin Atlantic, SoftBank and NTT Docomo prove there is real commercial traction. On top of that, more than $6 billion in Starshield contracts (military applications) was won from the U.S. military.

Space: the flagship is stalling

Space brought in $962 million in revenue in Q2, +29% year-over-year. But the operating loss widened from $(369) million to $(542) million. Even more striking: over the full first half of the year, Space revenue fell nearly 2%, from $1,611 million to $1,581 million. The number of launches for external customers over that same half-year period dropped from 21 to 17.

The total mass launched to orbit fell 26% in Q2 (from 652 to 485 metric tons). For a company that positions itself as the undisputed market leader in rocket launches, that’s a number that raises questions. SpaceX explains that more capacity was used for internal payloads (mainly Starlink satellites) and that Starship test flights replaced part of the classic Falcon launch cadence.

But the conclusion stands: today this is NOT a segment growing at an accelerated pace, it is a segment burning cash to get Starship ready. R&D expense for Space rose 55% to $1.08 billion in the quarter.

AI: the big surprise and simultaneously the biggest question mark

AI revenue jumped from $737 million to $2.56 billion, a 247% increase year-over-year. Adjusted EBITDA turned positive for the first time at $1.15 billion. Spectacular numbers and precisely why they deserve extra critical scrutiny.

Almost the entire story hinges on one thing: $14.1 billion in new Cloud Services agreements, of which $1.6 billion was already recognized as revenue in Q2. Two things matter here:

One, the risk is concentrated in one or two customers.

The 10-Q discloses that “Customer B” accounted for 19.5% of SpaceX’s total consolidated revenue in Q2 (in the prior quarter this customer was not material). All of it in the AI division. On top of that there is “Customer A,” accounting for 18.3% of revenue, spread across the three segments. Together, two customers represent nearly 38% of all revenue. Their identities are not disclosed.

For an investor this is a fundamental dependency: if one of them walks away, a hole opens up in the top line immediately.

Two, these contracts can be cancelled relatively quickly.

In the new risk factor added in this 10-Q, SpaceX literally states that the cloud services agreements “generally provide for monthly fees and, after an initial period inclusive of capacity ramp (generally, a number of months), may be terminated by either party upon 90 days’ notice.“ In other words, that $14.1 billion in “contracted sales” is not the same as a multi-year contract with, say, the U.S. Space Force. After the initial ramp, the customer can effectively walk away on a quarterly basis.

The AI division absorbed $15.83 billion in capex in Q2 alone. That’s more than 21 times what it was in Q2 2025. Compute capacity quadrupled from 0.4 GW to 1.4 GW.

Three things that did NOT make the headlines

1. Over the full half-year, the loss got substantially worse.

For H1 2026, SpaceX posted a net loss of $(4.82) billion versus $(1.54) billion a year earlier, more than triple. That contrasts with the “improvement” you see on a quarterly basis.

The main drivers: a $1.55 billion loss on the early repayment of legacy xAI loans (technical term: “debt extinguishment”), $539 million in unrealized losses on digital assets (likely crypto on the balance sheet), and a doubling of interest expense.

2. A lot of financing comes from “insiders.”

SpaceX has $13.3 billion in debt and lease obligations outstanding with Valor Equity Partners, an investment firm whose founder/CEO (Antonio Gracias) also sits on SpaceX’s board. These structures (technically: “failed sale-leasebacks”, SpaceX sells AI infrastructure and immediately leases it back) generated $327 million in interest expense in Q2 alone. For comparison: at year-end 2025 this exposure was “only” $4.5 billion.

That kind of concentration of AI financing with a party tied to a board member deserves at minimum a footnote in any valuation discussion: it raises questions about independent oversight and whether the terms are truly market-based.

3. Dilution: the share count nearly doubled.

The number of shares over which earnings are divided rose from 2.93 billion to 5.86 billion, a 100% increase. Dilution means that if you owned one share a year ago, that share today represents a smaller slice of the company, because many new shares have since been created. This is mainly the result of the xAI merger (February 2026), the IPO (June 2026), and the conversion of certain preferred shares to common.

The loss per share dropped from $(0.34) to $(0.09) as a result, but that is largely an accounting effect and NOT a real operational improvement.

What metrics should you actually track?

Anyone who wants to follow SpaceX seriously - without buying into the hype - should watch the following quarter after quarter.

For Starlink (Connectivity)

Net new subscribers per quarter.

In Q2, 1.7 million were added (from 10.3M to 12.0M). The question is whether the number of new customers per quarter stays high as wealthier markets saturate.

Trend in revenue per customer (ARPU).

From $85 to $66 in one year is a meaningful signal. Does it stabilize around $66? Or does it keep drifting toward $50 as Africa, South America and Southeast Asia become the dominant growth markets?

Margin in the enterprise segment.

SpaceX does not report this separately, but the +108% growth is likely coming in at higher prices than the consumer business. A concrete question to put to management: what is the margin at enterprise versus consumer?

Rollout of the new V3 satellites.

SpaceX announces these should deliver “meaningful increases in broadband capacity and data density.” Pace of rollout and impact on cost per gigabit per second are key.

For Space

Starship cadence and reusability.

Flight 12 (May 2026) and Flight 13 (July 2026) were successful. The company claims Starship will “reduce cost per orbit by 99% or more.” That’s one of the boldest industrial claims out there right now.

What to track: (a) number of Starship flights per quarter, (b) share of flights that are fully reused, (c) number of paying customers actually flying on Starship.

Recovery of customer revenue.

The drop from 21 to 17 external launches in H1 has to reverse. If it doesn’t, Space will shrink as a paying business while R&D losses explode.

For AI (the most critical)

Who is Customer B?

What share does this customer keep, and what happens once the initial ramp period of the contract ends? By far the single largest risk on the balance sheet.

AI revenue excluding cloud infrastructure.

The $2.19 billion in “AI Solutions & Infrastructure” masks whether underlying organic growth (Grok subscribers, X Premium, API access) is equally strong. Advertising revenue actually fell 14% year-over-year, from $426M to $367M, a sign that X as an ad platform is not growing along with the rest.

Utilization rate of the 1.4 GW of compute.

Installed capacity is not the same as used capacity. The ratio of “contracted GW / installed GW” and the payback period per invested dollar are metrics management should eventually disclose.

Capex pace.

Musk intends to hold the Q2 capex pace through the second half of 2026. That would work out to roughly $65 billion in capex for 2026. With $100 billion in cash and marketable securities, that means the cash position could be depleted in about 18 months without fresh financing or a sharp rise in operating cash flow. In the first half of the year, operating activities generated only $3.47 billion, less than one month of Q2 capex.

For the balance sheet

The order book technically “backlog” or “Remaining Performance Obligations”)

Stands at $47.46 billion, of which 56% is expected to convert into revenue within a year. Track quarter by quarter: how many new orders come in, how many convert to revenue, what share is cancellable versus fixed.

Evolution of “insider” debt.

Will Valor Equity remain the financing partner for AI infrastructure, or will this debt migrate to the regular capital markets?

New bond issuances.

The inaugural $25 billion bond at a weighted average coupon (annual interest rate) of 5.855% is a solid start at the investment-grade level (meaning: a reasonably strong credit rating). Future issuances and the spread versus comparable companies will show how the credit markets view SpaceX.

What is missing from this analysis?

This being SpaceX’s first public earnings report, there are inherent limitations that every reader should acknowledge:

• No historical quarterly track record. Only Q2 2026 versus Q2 2025 is available. Analyses over multiple quarters are impossible today. Q1 2026 is shown in the press release as a comparison point, but without underlying commentary.

• No guidance. SpaceX does not provide concrete numerical forward-looking guidance in this press release. Musk’s remark about $1 trillion in revenue by 2030 is a narrative claim, not an official corporate forecast.

• Anonymous customers. Customer A and Customer B remain unnamed. For a company where two customers together account for nearly 38% of revenue, that is a material information gap for external investors.

• The Cursor acquisition ($60 billion, closing in Q3 2026). The impact on revenue, dilution, and the balance sheet is not yet reflected in Q2 numbers.

• Spectrum deal with EchoStar ($19.6 billion). Partially reflected already, but the full impact on the competitive position in mobile satellite services will only become clear over several quarters.

Closing: facts versus story

SpaceX’s first public numbers confirm three things.

• First: Starlink is a fundamentally profitable business scaling quickly.

• Second: AI is suddenly generating billions in revenue thanks to a handful of large cloud contracts.

• Third: after the IPO, the company holds an unprecedented amount of financial firepower ($100 billion in liquidity).

What the numbers do NOT prove - and what should hold you back from premature conclusions - is (1) whether the AI revenue is structural, given that nearly 20% depends on a single customer with a 90-day cancellation clause; (2) whether Starship will be operationally AND economically viable in time to offset the losses in Space; and (3) whether the massive investment cycle will be paid back within the time horizon Elon Musk projects.

For an investor who wants to track this file objectively, the coming quarters come down to watching a specific set of gauges:

• the ARPU trend at Starlink,

• the utilization rate and customer retention in the AI cloud,

• Starship flight cadence and conversion to paying customers,

• and above all operating cash flow versus the capex pace.

If those parameters trend the right way, SpaceX effectively grows into its valuation story. If not, the $1.75 trillion valuation with which the company came to Wall Street - today already 20% lower - is a reflection of a story, not of cash flow.

ChartMill Market Desk - Kristoff

Disclaimer: This analysis has been prepared solely for informational and research purposes and does not constitute investment advice, a recommendation to buy or sell, or personal financial guidance. Readers should conduct their own research and, where appropriate, consult a registered advisor before making investment decisions. All figures are drawn from publicly available sources (SpaceX Q2 2026 press release and Form 10-Q filing with the SEC for the quarter ended June 30, 2026).

Sources:

• SpaceX-Reports-Second-Quarter-2026-Results.pdf (press release, 10 pages)

• UNITED-STATES-SECURITIES-AND-EXCHANGE-COMMISSION-10Q-SPACEX.pdf (SEC Form 10-Q, 150 pages)

• SpaceX Q2 2026 Earnings Webcast

The Dow just cleared 53,000 for the first time on the back of two things that don’t usually line up... A cratering crude oil market and an ISM Manufacturing print at the strongest level in more than four years.

Add Amazon crossing the three-trillion-dollar market cap threshold - the second US company ever to do it - and Palantir dropping an after-hours guide raise that CEO Alex Karp himself called “alien,” and you had a Monday session that quietly rewrote the setup for the rest of August. Let me walk through what actually moved and why the day matters more than the ticker on the front page suggests.

The Rundown

• The Dow closed at a fresh record on the back of a sharp oil selloff and the strongest ISM Manufacturing print in more than four years.

• A megacap tech name joined the three-trillion-dollar club, extending Friday’s cloud-driven rally into a second session.

• A high-profile data analytics company doubled its top line in Q2 and raised its full-year outlook by nearly a billion dollars in after-hours trade.

• Airlines and a jet-maker were the day’s cyclical winners; a hotel chain and a legacy marketplace went the other way on company-specific setbacks.

Oil Slid, Cyclicals Ripped, And The ISM Confirmed The Setup

Crude did the heavy lifting Monday. The oil-price drop pulled through directly to airlines and industrials, and it broadened the tape enough to matter: almost every S&P 500 sector closed green.

DELTA AIR LINES INC (DAL | ▲4.75%) and SOUTHWEST AIRLINES CO (LUV | ▲4.67%) were among the visible beneficiaries, and BOEING CO/THE (BA | ▲8.03%) tacked on 8%, the kind of size move you don’t see very often on a name Boeing’s scale, and one that suggests something more than reflex algo buying on the oil print.

Then came the ISM. The Manufacturing PMI printed 55.6 for July, its strongest reading in more than four years. That is well above the level that would fit a soft-landing frame. Anyone still writing a recession backdrop into Q3 needs to explain how you reconcile a four-year high in factory activity with the growth-slowdown narrative.

My read: you can’t, cleanly. The tape is telling you a different story than the bear case has been running with.

For anyone tracking the crude side, this is now the second session where oil has been the swing factor. The Iran-talks headline that broke into Monday’s open cracked the summer’s rally in WTI, and the ISM strength gave the market a demand-side reason to stay bid regardless. Both forces are pulling the same direction for equities, which is why the breadth looked as clean as it did.

Amazon Joins The Three-Trillion-Dollar Club

AMAZON.COM INC (AMZN | ▲4.58%) tacked on another 5% and formally crossed three trillion dollars in market cap, joining a very short list. Friday’s 15% AWS-driven pop set the stage; Monday’s extension was the market putting a second reading on the same print and saying yes, this actually is what it looks like.

The wider tech complex was strong. The Nasdaq closed up 2.13%, keeping the composite well within striking distance of a new record even without Apple’s ballast.

Palantir’s “Alien” Quarter Was Absurd By Any Reasonable Measure

PALANTIR TECHNOLOGIES INC-A (PLTR | ▲2.10%) delivered the after-hours line of the day.

• Q2 revenue up 93% year-on-year to $1.94 billion against a $1.80 billion consensus.

• Adjusted EPS of $0.41 versus $0.35 expected.

• Net income of $1.07 billion, more than three times last year’s $329 million.

• And the FY26 revenue guide got lifted from $7.2 billion to $8.15 billion, with adjusted operating income now expected around $4.89 billion.

The commercial line was the standout. US commercial revenue grew 149% to $764 million, which for a business whose historical spine was government contracts is the kind of number that reframes the story. Government revenue, the traditional core, still grew 90% to $809 million. So it’s not that one engine is cannibalising the other. Both are firing.

CEO Alex Karp called it “an alien quarter” in the earnings letter, which is on-brand for someone who’s never met a press release he couldn’t turn into a manifesto. He also used the release to press the “AI sovereignty” narrative, Palantir positions itself as the answer for customers who don’t want their proprietary data becoming free training material for third-party language models. The stock climbed roughly 10% in the immediate after-hours reaction and extended to around 15% as the tape absorbed the guide raise.

My take: at three hundred billion in market cap Palantir isn’t cheap on any traditional metric, and the tight political entanglement with the Trump administration is a real overhang for the international book. DGSI in France just walked, and that won’t be the last one. But the numbers this quarter are impossible to argue with, and the commercial acceleration is what actually justifies the multiple. A guide raise this large, this late in the year, is what a company does when it’s genuinely surprised by its own pipeline.

Snap’s Q2 Was A Reminder That Ad-Supported Social Isn’t Dead

SNAP INC - A (SNAP | ▲7.46%) also beat after the bell.

• Revenue of $1.60 billion, up 19% and ahead of the $1.54 billion consensus.

• Adjusted EBITDA of $250 million, six times last year’s number and well ahead of the $192 million analysts had penciled in.

• Daily active users came in at 493 million against a 487 million forecast, and ARPU ticked to $3.25.

The stock added better than 10% in the extended session. Q3 revenue guide of $1.70–$1.74 billion sits right at the top of consensus. The signal here is that Snap’s turnaround story - smaller loss, real EBITDA leverage, DAU growth outpacing the pessimistic case - is finally showing up in numbers investors can put in a model.

It doesn’t make the stock a buy at any price, but it does end the “structural decliner” thesis that has priced this name for the last two years.

The Casualties: GameStop, eBay, Marriott

Not everyone had a good session.

GAMESTOP CORP-CLASS A (GME | ▼12.25%) dropped 12.5% after converting $1.4 billion in convertible notes into equity, wiping out the year’s gains on the dilution. This is the kind of capital-structure event that reminds you the meme-stock era’s financial engineering doesn’t quietly amortise. It shows up on the share count in one lump.

EBAY INC (EBAY | ▼6.03%) fell about 5% on a Wells Fargo downgrade. The analyst’s concern: Depop is going to require material additional spend to defend against Vinted, which is winning US share fast. Marketplace competition is one of those variables that looks fine in the model until it isn’t, and once the sell-side flags the spending question, the multiple typically has to compress before it can rebuild.

MARRIOTT INTERNATIONAL -CL A (MAR | ▼6.97%) missed on revenue and pointed to the Middle East conflict as the drag.

That is a legitimate factor. It is also the kind of “external environment” answer that shareholders find satisfying exactly once. If it shows up again in Q3, the market will want more than a geopolitical footnote.

The Deal Story: Atkore Gets A Take-Out At $95

ATKORE INC (ATKR | ▲28.22%) jumped 28% on news that Italy’s Prysmian is bidding $95 per share for the US cable maker. Cross-border industrial M&A at meaningful premiums is a signal I take seriously, when European strategics are willing to pay up for US-listed peers in electrification-adjacent categories, it tells you private-market valuations for grid infrastructure assets are running well ahead of the public tape.

Watch the group; this is unlikely to be the last transaction of this shape.

Conclusion

Monday was a session where the macro backdrop, the sector rotation, and the earnings tape all pointed the same direction for once.

Oil down, ISM at a four-year high, cyclicals bid, hyperscaler cloud numbers reconfirmed by Amazon’s move above three trillion, and Palantir’s guide raise adding a second AI datapoint that isn’t a promise. It’s revenue.

The risks haven’t gone away. The long end of the curve is still elevated, three regional Fed presidents are still publicly hawkish, and Friday’s payrolls print is still the number that shapes September. But the setup coming out of Monday’s close is a lot more supportive than the one we walked in with.

ChartMill Market Desk - Kristoff

This daily update is prepared by ChartMill for informational purposes only and does not constitute investment advice. Always do your own due diligence before making investment decisions.

👉 This article is also available at ChartMill.com.

Last Friday, July 31, before the open, we published a case study on BRIGHTSPRING HEALTH SERVICES (BTSG | ▼18.07%). It walked through why the stock had cleared the Sector First Screen that morning, what the Q2 earnings print delivered, and the concrete trade setup that flowed from both.

Then the market opened, the stock gapped down, and closed the day 18% lower.

A handful of readers wrote in with variations of the same message: “You called this one badly wrong. Anyone who followed the article is now sitting on a 1% portfolio loss.”

Fair to raise. But the conclusion is wrong, and the reason it’s wrong is the entire point of how the setup was structured.

What the setup actually said

The trade plan in the article was specific:

• Entry: 74.28 (Buy Stop just above the 20-day high at 73.75)

• Stop: 69.14 (below the 68.16 horizontal support)

• Position size: 48 shares on $25K capital at 1% risk

A Buy Stop is an order that only becomes active - and only fills - once the price trades at or above the trigger level. It isn’t a market order to buy at the open. It isn’t a limit order sitting below the market. It’s a conditional order that says: I want to be long this stock only if the market first proves that buyers are willing to take price through 74.28.

BTSG closed the previous session at 72.88. The Buy Stop sat at 74.28, $1.40 above the close, deliberately parked just above the resistance zone that had capped the stock for weeks.

Friday’s open never got there. The stock gapped down and traded lower the entire session. The 74.28 trigger was never touched, which means the Buy Stop was never activated, which means no shares were ever bought, which means no stop was ever hit.

Portfolio impact of following the article’s setup: zero.

Why the entry sat above the market, not at it

This is where the choice of order type stops being theoretical.

If the article had recommended a Market Buy on the open, or a Buy Limit around 72.88, the story would be different. On a market buy, you’d have filled somewhere near the gap-down open, taken the 18% hit into the close, and blown through the 69.14 stop by a wide margin, meaning the pre-defined 1% risk would have expanded into something meaningfully worse. On a Buy Limit at the previous close, same story with a slightly better fill.

The Buy Stop exists precisely to filter out this scenario.

The premise of the setup was straightforward: the sub-industry is confirmed, the earnings print is clean, the CRS is top-decile, but the stock is sitting right at multi-week resistance, and I want the market to prove itself before I risk capital. The $74.28 trigger was that proof. Clear the resistance, the thesis gets confirmation and the trade goes live. Fail to clear it, no trade.

That’s the mechanism doing exactly what it’s designed to do.

The broader point

The choice of order type is one of the highest-leverage decisions in any setup, and it gets overlooked constantly because it feels like arithmetic housekeeping. It isn’t. It’s structural risk management that sits above position sizing.

The full breakdown of order types and when to use each is on ChartMill in the Money & Risk Management section. If you’re building setups from screens - Sector First, Bottom-Up, or anything else - that article is worth ten minutes of your time. The wrong order type quietly turns a well-conceived setup into a market order at whatever price the tape hands you.

Was the setup itself wrong?

Separate question, and worth answering honestly.

The setup was built on a top-decile leader inside a confirmed sub-industry with a fundamental beat and raised guidance. That’s the exact configuration the Sector First workflow is designed to surface. The subsequent 18% drop tells you the market read the print in a way I didn’t anticipate, likely a mix of profit-taking on a name up 271% over twelve months, plus specific concern about margin trajectory, guidance quality, or the reimbursement commentary embedded in the release. The thesis looked defensible on the data available Friday morning. Whether it stays defensible is a separate conversation.

But a defensible thesis is not the same as a filled trade. The order structure was there to distinguish between the two.

Bottom line

The Sector First screen did its job by surfacing a top-decile leader with fundamental confirmation. The order structure did its job by refusing to open a position without price confirmation of the breakout. The risk framework did its job by capping the worst-case fill at 1% of capital.

On Friday, none of those pieces failed. The trade simply didn’t happen, because the market didn’t give the setup the entry it required.

That isn’t a loss. That’s the framework working exactly as designed.

Trade Safe!

Kristoff

Related reading:

• From Screen to Setup: BrightSpring Health ($BTSG) Turns Up in the Sector First Output on Earnings Day

• Which Order to Use when Buying Shares

• Sector First or Stock First? Two Workflows for Growth Investors

Disclaimer: This article is for informational and educational purposes only and does not constitute financial advice or a recommendation to buy or sell any security. Always conduct your own research and consider your risk tolerance before making investment decisions.

That is the story from Friday’s close, and it dragged the entire memory-chip complex down with it.

What kept the tape from getting ugly was on the other side of Big Tech, where Amazon delivered the kind of cloud print that pays for a lot of AI capex. And by the time Asia opened on Monday, oil had shed more than four dollars a barrel on the announcement of fresh talks with Iran, which is quietly the most consequential development for how the next PCE print lands.

Let me walk through what actually mattered from Friday’s session and why the week ahead now hinges almost entirely on payrolls.

The Rundown

• Wall Street closed a volatile July mixed, with a soft outlook from an outgoing Big Tech CEO carving the memory-chip trade in half.

• A cloud blowout on the other end of the megacap complex kept the Dow green and reset the AI-hyperscaler debate again.

• Oil fell sharply into Monday’s open on the announcement of new US–Iran talks, easing one macro pressure point ahead of the next inflation print.

• The week ahead is a jobs-report week, with three regional Fed presidents already publicly restless about inflation.

Cook’s Exit Interview Was The One Nobody Wanted

APPLE INC (AAPL | ▼7.35%) had rallied into the print because the sell-side had convinced itself the September iPhone cycle would rescue the September quarter. Cook effectively confirmed the demand story is still there and then explained why the company can’t fulfill it.

The AI build-out has vacuumed up every available memory chip, and Apple’s Q4 revenue guide of 9–11% growth landed below the 12%-plus the desk had penciled in. Gross margin, which is one of the three legs the Morgan Stanley team has always treated as Apple’s structural moat, is stepping down from north of 50% to a 47–48% range. Services growth is also cooling on unfavourable FX and softer mobile gaming.

The June-quarter numbers themselves were fine, revenue up sharply, EPS ahead of consensus, with a two-billion-dollar tariff refund flattering the print. But investors weren’t grading the past quarter. They were grading John Ternus’s inheritance. And the reaction is that the “startprice” his tenure will be measured against just got a lot more forgiving. That is a small mercy for the incoming CEO. It is not one Cook would have chosen to leave behind.

The knock-on move was in the memory names Cook publicly criticized on the call. MICRON TECHNOLOGY INC (MU | ▼5.90%) and SANDISK CORP (SNDK | ▼5.09%) both dropped in a 5–6% range, which is the kind of reaction you get when the world’s largest customer of your product tells you on a live earnings call that he wants “more suppliers.”

I take that comment seriously, because it tells you the pricing tension between hyperscalers, phone makers, and the memory-makers themselves is about to become a stated corporate priority, not a background one.

Amazon’s Cloud Was The Only Thing Keeping Big Tech Honest

The offset was AMAZON.COM INC (AMZN | ▲15.32%).

AWS accelerated to 37% growth, which is the kind of number that ends conversations about whether the AI capex boom is showing up on the revenue line. For anyone who’s spent the past three months trying to work out which hyperscaler is going to generate returns on the hundred-billion-dollar spending commitments, this print settled the argument for now: AWS is compounding at a rate that easily supports the invoice. The stock added 15%, which is not a print you see often on a business Amazon’s size.

Full disclosure: I’ve held a long Microsoft (MSFT) position since February, and the AWS print reinforces the broader “cloud is monetising the AI trade” thesis I’ve been carrying that position on.

The read for portfolios is that the split between “AI-monetising” and “AI-still-explaining” is now the dominant intra-tech trade, and it is not going away before the next round of guidance in October. If you own the hyperscalers, you want to see the AWS-style acceleration in every cloud print you read. If you own hardware, watch memory pricing, Cook’s comments were the loudest confirmation yet that the constraint has moved from GPU supply to memory supply.

Oil Cracked On The Iran Announcement

The macro news that will probably matter most this week is the one that broke over the weekend.

President Trump announced that fresh talks with Iran begin Monday, and the oil market reacted immediately: Brent slid more than four dollars to $83.70, giving back a chunk of the roughly 22% July gain that saw WTI push toward $85. The framework, as it’s been sketched, would target reopening the Strait of Hormuz and revive the fourteen-point Islamabad protocol that collapsed in early July.

Iranian military officials, characteristically, called Trump’s version of events “another lie”, while the Iranian foreign minister posted on Telegram that talks with Oman and Saudi Arabia on the Strait are in “final phase.” The two things can both be true. The market is treating the direction of travel as bullish for supply and bearish for the crude curve.

Energy has been the one macro variable capable of puncturing the disinflation trajectory that the PCE and CPI prints have started to show. If oil actually settles into the low $80s or lower, the Fed’s inflation-hawk argument gets weaker by the week. That does not mean cuts are coming, Kevin Warsh has been clear about wanting the inflation number below 3% before he considers easing, but it does thin the case for the hike the market’s front-end has been half-pricing.

Three Fed Presidents Went Public With Their Dissent

On Friday, Beth Hammack of Cleveland, Neel Kashkari of Minneapolis, and Lorie Logan of Dallas each explained on the record why they voted for higher rates at this week’s FOMC meeting.

Their shared view, in essence: absent a hike, the last mile of inflation back to 2% doesn’t finish. That is public pressure on Chair Warsh, and it is unusual for three regional presidents to coordinate their messaging this openly this quickly after a decision.

The bond market read it accordingly. The ten-year yield peaked at 4.75% on Friday, and the thirty-year printed 5.28%, which is the highest level in nearly two decades. EUR/USD closed at 1.1532.

My take: the long end is telling you it does not believe the disinflation story enough to price out term premium, and it is telling you the Fed is unlikely to move first. Both of those messages have to matter for how you position anything rate-sensitive going into a payrolls week.

Consumer Confidence Improved, But From A Low Bar

The University of Michigan’s consumer sentiment index rose from 49.5 in June to 55.2 in July, and the twelve-month inflation expectation eased from 4.6% to 4.2%.

So yes, the data is moving in the right direction. No, I would not read too much into a single month’s improvement from a genuinely depressed base.

The Rest Of The Tape

CHEVRON CORP (CVX | ▲2.35%) beat comfortably on higher crude and stronger production.

EXXONMOBIL HOLDINGS CORP (XOM | ▼0.97%) posted its highest quarterly output in more than two decades and, in a small paradox that says a lot about how oil equities are priced right now, still finished the day lower.

ROBLOX CORP -CLASS A (RBLX | ▼26.85%) was the day’s real casualty outside of Big Tech, cratering 27% after the gaming platform guided current-quarter revenue growth well below where the sell-side had modelled it. The Dow closed 0.53% higher at 52,485 despite the Apple drag, mostly on Amazon’s shoulders.

What I’m Watching This Week

The ADP payrolls read on Wednesday sets the expectation for Friday’s official non-farm print, which is the number that will shape the September FOMC path more than anything else on the calendar.

In corporate earnings, ADVANCED MICRO DEVICES (AMD | ▼1.90%) and PALANTIR TECHNOLOGIES INC-A (PLTR | ▲0.65%) report, both of them AI-adjacent enough that a guide-down would echo through the hyperscaler complex.

And Friday will bring the first-ever quarterly update from SPACE EXPLORATION TECHN-CL A (SPCX | ▼3.41%) as a listed name; the stock closed Friday around $108, still meaningfully below its $135 IPO print, which tells you the market wants to see the actual operating numbers before it commits to a valuation.

Conclusion

The market cares about hyperscaler cloud growth that scales with AI capex, Amazon delivered exactly that. It does not care about historically strong fundamentals if the next quarter’s guide steps down, Apple learned that on Cook’s way out.

It cares about the long end of the curve, where thirty-year yields at 5.28% are a real constraint on multiple expansion. And it cares, above everything else this week, about payrolls. The one macro tailwind is that oil is finally coming down. The one macro headwind is that three Fed presidents are publicly telling their chair to hike.

Position accordingly.

ChartMill Market Desk - Kristoff

With regard to the stocks discussed in the article above; the author owns individual shares in Microsoft.

This daily update is prepared by ChartMill for informational purposes only and does not constitute investment advice. Always do your own due diligence before making investment decisions.

👉 This article is also available at ChartMill.com.

That combination - screen hit, fundamental beat, muted price reaction - is exactly the setup the top-down workflow was designed to catch. This piece walks through how the name got there, what the earnings print actually said, and where the trade sits now.

Where the name came from

The Sector First workflow, laid out in ‘Sector First or Stock First? Two Workflows for Growth Investors’, runs six steps: check the sector dashboard, isolate the sub-industries with high price strength, cross-check on fundamental growth, verify the trajectory, drill into the stocks inside, build the watchlist.

The saved screen encodes those steps as filters:

• Industry Strength ≥ 70 (top 30% by GICS sub-industry price strength)

• Industry Growth ≥ 60 (top 40% by fundamental group growth)

• HGM ≥ 6 (High Growth Momentum)

• CRS ≥ 80 (individual relative strength)

• Setup ≥ 7 (technical structure ready)

• Avg Volume (50 SMA) > 300K, US only

BTSG cleared all six. That already tells you something: it isn’t an isolated leader in a dead sub-industry. It sits inside a Health Care sub-group that scored on both price and growth before the earnings print arrived.

The individual profile

The stock’s own numbers are the reason it passed the individual filters:

• Technical Rating: 10/10

• Setup Rating: 7/10

• CRS (ChartMill Relative Strength): 98.85 — top 1.15% of the US market

• 3-month RS: 96.33 / 6-month RS: 97.90

• 12-month performance: +271%

• Trend: long-term rising, short-term rising, every SMA from 5 to 200 rising

• Outperforming 96% of 102 peers in the Health Care Providers & Services industry

The Fundamental Rating is 5/10 overall, but the breakdown matters more than the headline: Growth 8/10, Health 4/10, Profitability 5/10. In other words, the growth engine is strong; the balance sheet is still recovering from the 2022–2024 stretch of negative net income. That’s the exact profile the HGM filter is designed to accept, accelerating fundamentals, not pristine ones.

What the earnings print actually said

The Q2 2026 release hit before the open:

• Adjusted EPS $0.45 vs $0.40 consensus (+12.5% beat)

• Net revenue $3.873B vs $3.731B consensus (+23% YoY)

• Adjusted EBITDA $206M, +44.2% YoY

• Pharmacy Solutions +22% to $3.407B; Provider Services +30% to $466M

• Net income $87M vs $9M in Q2/25

Guidance was lifted to $15.10–$15.43B in revenue and $820–$845M in Adjusted EBITDA, implying roughly 33–37% EBITDA growth for the full year. Management also flagged a $300M debt paydown that dropped leverage from 2.27x to 2.15x.

There’s nothing soft in that print. Revenue beat, margin beat, EBITDA acceleration, guidance up, balance sheet improving. On paper it’s a clean quarter.

Why the muted reaction is the interesting part

Pre-market showed the stock up about 0.2%. On a beat this size, that looks anticlimactic. It isn’t.

The stock closed yesterday at 72.88, up 5.06% on the day, sitting right at the multi-week resistance at 72.91. It’s already up 38.6% over the past three months and 271% over the past twelve. A lot of the good news was priced in during the run into the print.

What you’re seeing now is a fundamental beat landing at a technical inflection point, with sentiment already stretched. That’s a different setup from a beat that gaps a stock 8% higher. It’s the configuration where the market absorbs the print quietly, the base holds, and the next leg comes from continuation rather than reaction.

Two things to watch from here:

Whether the stock takes out 72.91 on volume in the sessions immediately after the print. That’s the resistance line that’s capped price for weeks and it also happens to be the automated CR1 level on the chart. Whether 68.16 - the horizontal support that lines up with the 20 EMA - holds on any pullback. That’s the level the technicals break on if the print is going to be a top rather than a launchpad.

The trade setup example

Using ChartMill’s Position Sizing Tool on the daily chart:

• Entry: 74.28 (Buy Stop just above the 20-day high at 73.75, giving the resistance zone room to clear)

• Stop: 69.14 (below the 68.16 horizontal support with a small buffer)

• Risk per share: $5.14 — 6.92% of entry, or 1.64 ATR

• Capital: $25,000, risking 1%

• Position size: 48 shares → $3,565 invested, 14.3% of capital

The 1.64 ATR stop is on the tight side of comfortable, the tool flags it because it sits below the 2 ATR guideline for a stock with 3.9% daily true range.

Two ways to handle that: accept a slightly higher hit rate of noise-driven stop-outs in exchange for a better R:R, or widen the stop to 2 ATR (roughly 66.75) and cut share count to bring the dollar risk back to 1%. Neither is wrong. The first suits shorter-horizon swing traders willing to re-enter; the second suits position traders who want the setup to have room to breathe through normal volatility.

On the upside, the next real resistance sits at the psychological 80 level, and above that there’s no historical price data, the stock is at a two-year high. A move to 80 from 74.28 would deliver roughly 1.1R on the tight stop and 0.8R on the wider one. That’s the R:R you’re accepting on the initial position; trailing rules take over from there.

What can invalidate this

Three things to watch, in order of severity:

Reimbursement pressure. Medicare, Medicaid and managed care programs remain the largest revenue exposure. Any adverse policy move on rates hits the whole industry, and a Sector First trade dies with its sub-industry. Watch the Industry Strength Score for BTSG’s GICS sub-industry on the sector dashboard, if it starts rolling over, the tailwind that put $BTSG on the screen in the first place is fading.

Integration risk. The Amedisys and LHC Group branch acquisitions are expected to add $35M of EBITDA this year. Execution slips or margin pressure during integration would hit the guidance credibility that’s currently supporting the price.

Extension risk. The stock is up 271% in twelve months. That’s a lot of unrealised gain looking for an excuse to be taken. A muted reaction to a great print is fine; a muted reaction that turns into distribution days on rising volume is a different signal. The 68.16 support break is the mechanical trigger to reassess.

Bottom line

$BTSG is a working example of what the Sector First workflow is supposed to deliver: a name pre-screened against a sub-industry that already has both price strength and fundamental growth on its side, arriving at your desk on a day when a fundamental beat gives you a clean catalyst. The muted pre-market reaction isn’t a rejection of the print; it’s the market digesting a strong quarter into an already-extended trend and deciding where price wants to go from here.

The setup gives you a defined entry, a defined stop, and a defined position size with the Position Sizing Tool doing the arithmetic. What it doesn’t give you is a guarantee. That part is on the sub-industry holding up and on the 68.16 support not breaking.

If it works, you caught continuation on a top-decile CRS name inside a confirmed sector, with a fresh catalyst. If it doesn’t, the 1% dollar risk keeps the damage inside the framework.

Trade Safe!

Kristoff

Related reading:

• Sector First or Stock First? Two Workflows for Growth Investors

• Sector First Screen (trading idea)

• BrightSpring Health Services (NASDAQ:BTSG) Posts Q2 Beat and Raises Full-Year Guidance

Disclaimer: This article is for informational and educational purposes only and does not constitute financial advice or a recommendation to buy or sell any security. Always conduct your own research and consider your risk tolerance before making investment decisions.

Microsoft walked in as the biggest question mark in Big Tech and walked out as the answer. Meta did the reverse. Then Amazon reopened the AI-capex debate at a whole new price tag, and Apple chose to close out the day with a guide-down that its outgoing CEO would rather not have delivered as his final act.

That is a lot of narrative for one session, so let me take it in the order that actually matters for how I’d want to be positioned into Friday.

The Rundown

• The AI-cloud trade got its cleanest positive proof-point in months, and the whole Nasdaq benefited.

• One megacap ad company reopened every question this market has been asking about the payoff on hyperscale AI spend.

• Guidance from the world’s most valuable phone maker leaned on supply constraints and currency, not demand, but the market only heard “guide down.”

• Cooling PCE inflation and a slower Q2 GDP print did nothing to shift a Fed that has clearly signalled it plans to sit on its hands.

Microsoft Delivered Exactly The Number The Market Needed

I’ve been writing about the “AI capex vs. AI revenue” split for weeks, and MICROSOFT CORP (MSFT | ▲15.51%) just gave the cleanest answer we’ve seen this year.

Q4 revenue and operating income both climbed 18%, cloud growth accelerated to 43%, and - this is the part I care most about - operating cash flow of $55.4 billion was 30% higher than a year ago.

When a business spending this aggressively on data centres still generates cash flow that strong, the argument that the AI build-out is hollowing out the core simply doesn’t hold. Add to that a Copilot user base that jumped from 20 million paying seats a quarter ago to more than 30 million now, and you get why the stock tacked on 15%.

For me the more important line was Satya Nadella’s tone on the call. He wasn’t defensive about the spend, he was matter-of-fact about the return on it. That’s the tone the market wanted to hear from any hyperscaler defending three-digit-billion capex plans.

Analysts I follow flipped from “watch the free cash flow line” to “this changes the story”, and I think that shift is deserved. This was the AI proof-point of the earnings cycle so far.

Meta Reopened Every Question Microsoft Just Closed

META PLATFORMS INC-CLASS A (META | ▼7.95%) did the opposite in almost every way that matters. Revenue was fine, ad prices were fine, engagement was fine, the problem is that the Q3 guide underwhelmed, capex remains stubbornly high, and management’s language around when the AI infrastructure spend actually starts pulling its weight was, at best, vague.

Zuckerberg is still all-in on renting out compute alongside serving Meta’s own model training needs, but “vague” isn’t what a market wants when it’s being asked to fund the second-largest capex line in tech.

Several sell-side desks trimmed price targets while keeping the buy rating, the polite way of saying “we still believe, but we’re not making the case any louder until you show us something.” Meta is now firmly in the “show me” bucket for the second earnings cycle in a row. I’d note that this is exactly the fate Microsoft escaped tonight; the difference between the two stocks’ reactions is a near-perfect map of what this market is willing to pay for right now.

Amazon Went Bigger On Cloud And Bigger On The Capex Number

Then, after Thursday’s close, AMAZON.COM INC (AMZN | ▲3.90%) came out with what was arguably the most important earnings report of the day.

• AWS grew 36.7% to $42.2 billion, the strongest print for the cloud division since Q4 2021

• Total revenue of $200.6 billion beat the $196.9 billion consensus

• Advertising kept sprinting at 26% growth

• And AWS delivered a 39% operating margin, which is the kind of number you can build a valuation around.

The stock added 8% in extended trading on top of the 3.9% it already put on during the session.

But the headline number I want to flag is not any of the above. On the analyst call, Andy Jassy raised 2026 AI capex guidance from $200 billion to $220 billion, blaming rising memory-chip prices, the same “hundred-year flood” language Tim Cook used on the Apple call an hour later. The message from Jassy was that demand still exceeds supply, that 2027 and even portions of 2028 capacity are already reserved, and that “we think AI demand is still in an early stage.”

Twelve-month free cash flow flipped to a $7.6 billion outflow from a $18.2 billion inflow a year ago. If the market took a $220 billion capex number in stride, it’s because AWS delivered exactly the growth acceleration required to justify it. Microsoft made this trade look easy tonight; Amazon just extended the invoice by another twenty billion dollars.

Apple’s After-Hours Guide-Down Was The Ugly Note

The last earnings report of the day was also the roughest. APPLE INC (AAPL | ▼1.41%) headline June-quarter print was actually decent, revenue up 16% to $109.4 billion, EPS of $2.02 beating the $1.89 consensus (though 11 cents of that came from the tariff refund the Supreme Court forced the US government to hand back). iPhone revenue rose almost 22% to $54.25 billion, and Mac had a genuinely strong quarter at $10.35 billion, up 29%.

Where it went wrong was in the guidance and the after-hours desk reaction: Apple lost more than 8% in extended trading.

The Q4 revenue guide of 9–11% growth came in below the 12%+ the sell-side had penciled in, with the shortfall blamed on FX and - more meaningfully - supply-chain constraints. The AI boom is pulling every available memory chip toward the hyperscalers, and Apple can’t get enough of them to build all the phones it’s being asked for. Cook was explicit that demand exceeds expectations; the constraint is on the manufacturing side.

He also called out that gross margin will step down to 47–48% in Q4, hit by the same memory-price surge that Amazon just cited. Services disappointed slightly at $30.7 billion versus a $31.2 billion consensus, and China revenue of $18.8 billion undershot the $19.6 billion analysts wanted. This is Cook’s last conference call before the September handover to John Ternus, and it wasn’t the send-off he’d have chosen.

PCE Cooled, GDP Slowed, The Fed Wasn’t Moved

Away from the megacaps, the macro data did the Fed’s messaging work for it.

• Headline PCE inflation dropped from 4.1% in May to 3.7% in June, with core easing from 3.4% to 3.3%.

• Q2 GDP came in at 1.5%, down from 2.1% in Q1.

• Initial jobless claims rose by 9,000.

On paper, that’s a slowing-but-not-broken economy with disinflation moving in the right direction, which would normally start pulling forward the first cut. It didn’t. A majority of the market, per CME FedWatch, still expects the next Fed move to be a hike in September or later, and Kevin Warsh’s signalling this week gave no hint of any easing bias.

The bond market read the setup the same way. The ten-year yield eased slightly to 4.672% after Wednesday’s surge, but the level itself remains uncomfortable for high-multiple growth. EUR/USD sat at 1.1533.

My read: absent a genuine growth scare, this Fed is going to sit on its hands well into Q4, and the equity market has already stopped pricing in help from the front end of the curve.

Oil Slipped, Iran Kept Rumbling, And The Rest Of The Tape

WTI settled at $83.59, down about 1% on the day, even as US strikes on Iranian targets resumed following an Iranian attack in Jordan and Houthi activity in the Red Sea kept the shipping-lane premium in the price. On a monthly basis, crude is still tracking for a roughly 20% rise, which is why I keep flagging energy as the one macro variable that could puncture the PCE trajectory that just started cooling.

Around the edges, the picture was mostly constructive.

ORACLE CORP (ORCL | ▲8.34%) rallied on the announcement of an expanded partnership with ALPHABET INC-CL A (GOOGL | ▼0.91%), with the Microsoft halo effect on enterprise-tech names doing the rest.

LAM RESEARCH CORP (LRCX | ▲17.98%) blew past Q2 estimates as AI demand continues to soak up chip-equipment orders. SANDISK CORP (SNDK | ▲25.99%) put in its biggest one-day rally since early January - on a stock that had lost 45% earlier this month - as the memory-price story finally started working for the specialists rather than against them. QUALCOMM INC (QCOM | ▼2.62%) went the other way after missing Q3 estimates.

Outside tech, YUM! BRANDS INC (YUM | ▲3.34%) more than doubled Q2 profit even while managing through the parasite scare at one of its brands, and MASTERCARD INC - A (MA | ▲2.49%) beat on both lines and raised full-year guidance.

Elon Musk’s SPACE EXPLORATION TECHN-CL A (SPCX | ▼0.31%) closed slightly lower despite winning a $1.6 billion contract from the Space Force for eighteen Falcon 9 launches through 2027, a reminder that the recently listed name is still finding its trading base.

Conclusion

Thursday’s session put the current market’s grading rubric on a whiteboard for anyone who needs to see it.

If your revenue and cash flow scale with your capex, this market will fund whatever you want to spend, Microsoft and Amazon just proved that. If your capex outruns visible payoff, you get punished immediately, Meta learned that in the daylight, Apple learned it after hours.

My bias into Friday: the split between “AI-monetising” and “AI-still-explaining” is now the dominant trade, and I don’t see a macro catalyst on the calendar that flips it before the next round of earnings. Watch memory pricing, Cook and Jassy both flagged it, and if that “hundred-year flood” language turns out to be even half accurate, the hardware side of the AI trade gets a lot more interesting than it looks right now.

ChartMill Market Desk - Kristoff

With regard to the stocks discussed in the article above; the author owns individual shares in Microsoft.

This daily update is prepared by ChartMill for informational purposes only and does not constitute investment advice. Always do your own due diligence before making investment decisions.

👉 This article is also available at ChartMill.com.

Chair Kevin Warsh kept the policy rate on hold, exactly as the market expected, and stocks briefly liked it. Then the details of the vote surfaced, the bond market re-priced in real time, oil ripped 7% higher on renewed Trump-Iran language, and by the close the Dow was down more than 2% with the Nasdaq not far behind.

After the bell, Microsoft delivered the kind of AI number that shuts sceptics up, while Meta did the opposite and reopened every question investors have been asking about the sustainability of hyperscaler spending.

The Rundown

• The Fed held rates steady but three dissenters wanted a hike, and the long end of the curve responded by pushing yields to multi-year highs.

• Renewed geopolitical language around Iran sent oil up sharply, feeding the inflation worry that the Fed’s hold was supposed to soothe.

• Semiconductors and AI-adjacent industrials led the selling as investors kept trimming capex-heavy names.

• After-hours earnings split the megacap AI trade cleanly down the middle, one blowout, one disappointment.

The Fed Decision That Didn’t Land Where It Was Supposed To

The headline was benign. The Federal Open Market Committee left the target range at 3.50%–3.75% and Chair Kevin Warsh’s opening remarks emphasised the 2% inflation target as, in his words, “not soft.” That should have been a green light for equities. It wasn’t, and the reason sits in the vote itself: three governors dissented, all in favour of raising rates to push back on inflation that hasn’t obliged the Fed by rolling over. That’s a hawkish tilt inside a nominally neutral decision, and the bond market saw it immediately.

The ten-year yield jumped nine basis points to 4.70%. The thirty-year added eleven basis points to 5.21%, the highest level in years. That is the market pricing in the possibility that the Fed is now visibly behind the inflation curve — and importantly, that the political and internal appetite for another hike is real. Warsh’s aside that market participants “should learn to play, not to referee” reinforced the message: this Fed is not going to cushion any tantrums.

Oil Went Up, Everything Else Went Down

The macro side got worse from there. President Trump used a Fox News interview to warn Iran of a hard response to attacks on US troops in the region, and the oil market did exactly what you would expect. WTI jumped 7.4% to $88 a barrel, one of the largest one-day moves in months, on top of a stronger-than-expected drawdown in US crude inventories despite 3.8 million barrels being released from the Strategic Petroleum Reserve.

A supply-side shock landing on top of a bond market that just re-priced inflation risk is not a friendly combination for equities carrying premium multiples.

That is roughly the moment the tape gave up. The Dow Jones (DIA | ▼2.19%) shed 2.2%, the Nasdaq 100 (QQQ | ▼2.06%) closed 2.1% lower, and the S&P 500 (SPY | ▼1.9%) got dragged down with them. The one place investors could hide was, predictably, the energy patch.

EOG RESOURCES INC (EOG | ▲4.51%), APA CORP (APA | ▲5.07%) and gas producer EXPAND ENERGY CORP (EXE | ▲4.45%) all closed sharply higher, riding the crude move directly into the print.

The Chip Complex Got Sold Again And So Did The AI Industrials

The semiconductor sell-off from earlier in the week broadened.

ADVANCED MICRO DEVICES (AMD | ▼5.51%) dropped another 5.5%, and APPLIED MATERIALS INC (AMAT | ▼8.40%) took an 8.4% hit, the kind of one-day move that usually needs a specific catalyst but here really didn’t. ARM HOLDINGS PLC-ADR (ARM | ▼8.11%) fell 8.1% during the regular session even though its Q1 results printed a record 22% revenue growth to $1.29 billion after the bell, with data-center royalties more than doubling year-over-year. The stock barely moved in extended trading, which tells you more or less everything about what the market is currently willing to pay for AI-infrastructure exposure.

The more telling name was CATERPILLAR INC (CAT | ▼6.91%). Cat has been treated as an AI derivative for months thanks to its exposure to power generators for data centers, but Baird downgraded the stock on the view that tightening local regulation and NIMBY opposition are set to slow new data-center construction.

This is the second signal in a week - the SK Hynix capex bomb was the first - that the market is starting to interrogate the assumed straight-line growth in hyperscaler build-out. GOLDMAN SACHS GROUP INC (GS | ▼5.09%) also weighed on the Dow with a 5.1% drop, though there was less of a story attached to that one beyond the broader risk-off tone.

Garmin And GE Healthcare Prove Earnings Still Matter

Away from the wreckage, two names showed what a clean beat still looks like. GARMIN LTD (GRMN | ▲16.23%) exploded 16% higher after reporting a Q2 that felt almost from a different market.

• Net profit climbed 30% to $616 million on a record top line, driven almost entirely by the fitness division, the sport watches and cycling computers business, where consumers still aren’t cutting back.

• CEO Clifton Pemble raised full-year revenue guidance to above $8 billion and, importantly, launched the Cirqa Smart Band, a screen-less wellness tracker without a subscription requirement.

Anyone doubting whether Garmin can still find whitespace in wearables got their answer.

GE HEALTHCARE TECHNOLOGY (GEHC | ▲12.15%) tacked on 12% after its own Q2 beat, with margins ahead of expectations. That’s two very different consumer- and healthcare-adjacent tech businesses telling the same story: when earnings visibility is high and capex is measured, this market will pay up.

PROCTER & GAMBLE CO/THE (PG | ▼1.87%) drifted 1.9% lower after its Q4 print showed higher sales but softer earnings, a reminder that “beat on revenue” alone doesn’t cut it right now.

Microsoft Delivered The Number, Meta Didn’t

Then the megacaps reported.

MICROSOFT CORP (MSFT | ▼0.71%) posted a genuinely strong Q4 for its broken fiscal 2025/’26:

• $90 billion in revenue against a $87 billion consensus

• Azure and AI revenue crossing the $100 billion annual mark for the first time

• Copilot at more than 30 million paying users (up from 20 million in April)

• Capex for the quarter climbed 70% to $41 billion, on a full-year investment pace of around $190 billion, but the top-line growth was strong enough to buy management the benefit of the doubt.

CEO Satya Nadella got exactly the print he needed. The stock, which is still down roughly 18% year-to-date and had been the laggard of the big cloud names, jumped more than 8% in after-hours trading. That is the market saying: fine, if the growth shows up, we’ll fund the spend.

META PLATFORMS INC-CLASS A (META | ▼1.31%) got the opposite reception.

• Revenue of $60.8 billion beat the $60.2 billion consensus and ad prices ran 12% higher year-over-year, so the core engine is intact

• EPS of $6.18 missed the $7.14 consensus

• capex climbed to $31.08 billion in the quarter

• free cash flow collapsed 91% year-over-year to just $784 million.

The stock fell 8% in after-hours trading. Mark Zuckerberg confirmed on the call that Meta will keep spending aggressively - the low end of the 2026 capex range was actually raised to $130 billion from $125 billion - and hinted that Meta will build out a rentable compute business alongside its own AI training needs. That’s a longer-dated payoff story than Microsoft’s, and this market isn’t in a mood to wait.

To me, the investors’ reaction seems clear: spending is fine, as long as revenue and margins grow along with it. What investors really can’t stand is a sharp decline in free cash flow. In that regard, the response to Meta’s results seems similar to the one to Alphabet’s results last week...

Robinhood And Starbucks, Two Different After-Hours Stories

ROBINHOOD MARKETS INC - A (HOOD | ▼3.15%) reported a strong Q2 - revenue up 32% to $1.31 billion, adjusted EPS of $0.48 versus $0.43 consensus - but the stock still slipped about 3% after hours.

Crypto trading revenue collapsed 38% year-over-year to $100 million, and while equities, options and the newer event-contracts business more than filled the gap (event contracts alone brought in $156 million, up 10x), the market is still figuring out how to price a diversified Robinhood. Management now runs thirteen product lines each doing over $100 million in annualised revenue, and the newly launched “Trump Accounts” already have 7 million signups and $1.5 billion in deposits. This is a very different Robinhood than the meme-stock-era version.

STARBUCKS CORP (SBUX | ▲1.01%) was the after-hours highlight, jumping about 7% on a Q3 that showed the “Back to Starbucks” playbook actually working.

• Same-store sales rose for the fourth consecutive quarter

• Adjusted operating margin improved 430 basis points to 14.4%, and North America comps ran up 8.1%

• Management raised full-year adjusted EPS guidance to $2.55–$2.65 (from $2.25–$2.45) and now expects global comps close to 6%.

The China deconsolidation into the Boyu Capital JV dragged reported revenue down 1% year-over-year, but strip that out and the underlying business is genuinely turning. CEO Brian Niccol earned some credibility here.

Conclusion

Wednesday was a tape where every piece of good news was outweighed by something worse.

The Fed’s hold got repriced by its own dissent count within an hour. Oil ripped on geopolitical language. Chip and AI-industrial names were sold on a market that no longer wants open-ended capex stories without matching revenue. Microsoft delivered the AI proof-point investors have been demanding; Meta didn’t, and paid for it immediately.

My bias into Thursday: the split between “clean-earnings names” and “capex-heavy names” is now the defining trade of this cycle, and until the long end of the curve stops rising, high-multiple growth is going to keep working against, not for, the tape. AMAZON.COM INC (AMZN | ▼1.82%) and APPLE INC (AAPL | ▼0.56%) report over the next 48 hours, both prints matter, but the market is now grading them on the same rubric that just crushed Meta and rewarded Microsoft.

ChartMill Market Desk - Kristoff

With regard to the stocks discussed in the article above; the author owns individual shares in Microsoft & APA.

This daily update is prepared by ChartMill for informational purposes only and does not constitute investment advice. Always do your own due diligence before making investment decisions.

👉 This article is also available at ChartMill.com.

The Dow climbed a full percent on the back of solid old-economy earnings, the S&P 500 nudged out a modest gain, and yet the Nasdaq slipped as the entire chip complex got taken apart before the print investors were actually worried about had even landed.

The SK Hynix quarter dropped overnight, well after the US close, and its US-listed shares are now sitting more than 11% lower in pre-market trade on Wednesday. Tuesday, in other words, was the market front-running the risk. Wednesday is the tape reacting to the confirmation.

The Rundown

• The Dow outperformed on strong industrial and consumer earnings while the Nasdaq drifted lower.

• US chip stocks were sold on Tuesday in defensive positioning ahead of a marquee memory-sector earnings print, a print that landed overnight and confirmed the AI-overinvestment worry.

• Cyclical winners in autos, payments and packaged goods punched above their weight; logistics did not.

• The Fed decision, cross-currents from consumer confidence, and Big Tech earnings all still ahead in the same 48 hours.

The Tape: A Dow Day, Not a Nasdaq Day

The scoreboard tells you almost everything about who was on offer and who wasn’t. The S&P 500 SS SPDR S&P 500 ETF TRUST-US (SPY | ▲0.24%) closed at 7,428.78, up 0.2%. The Dow Jones SS SPDR DOW JONES INDUS AVG (DIA | ▲1.08%) tacked on a proper 1% to 52,747.32, its heaviest lift in weeks. And the Nasdaq INVESCO QQQ TRUST SERIES 1 (QQQ | ▼0.97%) finished slightly lower at 24,876.91, dragged down almost single-handedly by the chip complex.

That divergence didn’t come out of nowhere. This is a market where investors are quietly rotating out of anything with an “AI capex” story and into names where the earnings math is easier to read. Coca-Cola raising guidance is a much cleaner signal than a chip manufacturer that has yet to justify the next round of spending. Under the surface, that’s the trade going on right now.

The euro/dollar drifted around 1.14, oil kept sliding for a second session, and the ten-year gently retreated, a set-up that should have been more supportive for tech than it turned out to be. The bond market is behaving. The equity crowd is picking sides.

The Chip Complex Sold First, Then SK Hynix Confirmed the Fear

The chip sell-off on Tuesday was, in a real sense, a pre-print trade. US-listed SK HYNIX INC-ADR (SKHY | ▼8.98%) shed 9% in the regular session, in nervous anticipation of the Korean giant’s Q2 numbers due overnight. MICRON TECHNOLOGY INC (MU | ▼8.85%) fell 9% and ended up among the ugliest names in the SOX.

TAIWAN SEMICONDUCTOR-SP ADR (TSM | ▼1.70%), INTEL CORP (INTC | ▼5.86%), SANDISK CORP (SNDK | ▼14.25%) and WESTERN DIGITAL CORP (WDC | ▼6.91%) all took hits of their own.

The iShares Semiconductor Index ISHARES SEMICONDUCTOR ETF (SOXX | ▼4.80%) shed more than 4% on the session, a big move for an index this large in a market that officially “closed higher.” What’s telling is that this all happened without anyone having seen the numbers yet. That’s positioning, not reaction.

Then the results dropped in the overnight hours, and the fear turned out to be well-placed, but not for the reason most people were bracing for. SK HYNIX INC-ADR (SKHY | ▼8.98%) printed what should have been a victory lap: revenue up more than 50% year over year on booming memory-chip demand, operating profit nearly seven times what it was a year ago at roughly 60.5 trillion won. Great quarter. Missed the highest end of analyst expectations, but a great quarter all the same.

The problem was tucked into the investor presentation. SK Hynix guided its 2026 capex above 40 trillion won - call it $24 billion - up from about 30 trillion in 2025. For a market that has spent weeks worrying about circular financing between chip suppliers and hyperscalers, that number landed like a match on a fuse. Seoul-listed shares got taken to the woodshed, dropping roughly 11%, and by Wednesday morning the US ADR was down more than 11% in pre-market. The stock now sits some 55% below the record high it printed on June 23.

My read: this is a structural setup, not a mood swing. As long as capex ambitions keep scaling faster than end-demand visibility, every strong memory print becomes a sell-the-news event and increasingly, a sell-ahead-of-the-news event. Tuesday’s pre-emptive sell-off in Micron and the rest of the complex is the tell. The market has already decided what to do with any AI-adjacent capex number that starts with a “3” or a “4”.

Dow Earnings Carry the Weight

The other half of the tape looked completely different, and it’s why the Dow put up the number it did. COCA-COLA CO/THE (KO | ▲5.00%) delivered exactly the kind of quarter you’d want in a market that suddenly cares about visibility again, strong Q2 revenue and profit growth and an upgraded full-year outlook. The stock reportedly had one of its best sessions in years. In a rotation environment, this is what “quality defensiveness” looks like on a chart.

BOEING CO/THE (BA | ▲4.76%) added to the industrial tone with a well-received print of its own, feeding straight into the Dow’s leadership. JOHNSON & JOHNSON (JNJ | ▲0.29%) edged up after saying it would pay $5.5 billion to settle a long line of talc-related cancer lawsuits, a big number, but the market read it as a lid on further tail risk rather than a hit to the thesis.

Not everyone in the “old economy” bucket held up. UNITED PARCEL SERVICE-CL B (UPS | ▼6.57%) sank almost 7% despite actually beating Wall Street’s numbers and raising its outlook for the full year. That’s a good reminder of what I’ve been saying for a few weeks now: this market is significantly harsher to earnings misses than it is generous to beats.

PayPal, Visa and Ford - Three Different Ways to Beat

The financials and autos also gave the tape something to work with.

PAYPAL HOLDINGS INC (PYPL | ▲4.01%) climbed on stronger revenue in Q2, even as margins pinched profit lower. The full-year outlook was lifted, which is what the market rewarded. Management again declined to comment on Stripe’s rumored $53 billion takeover bid, other than to say a buyout offer would be worth listening to. If that headline develops into an actual deal, this is a story that reshapes the payments landscape overnight.

VISA INC-CLASS A SHARES (V | ▲1.12%) beat both on revenue and adjusted EPS, with cross-border volumes up 13% and total processed transactions up 10% to 71.7 billion.

CEO Ryan McInerney called consumer and business spending “resilient”, which is a small but useful data point for anyone reading the tea leaves on the US consumer. The stock still slipped about 1.4% in after-hours trading, likely on the announced 7% workforce cut across tech and product teams. Wall Street tends to like cost discipline as an idea and dislike headline layoffs as a story. Both reactions are usually right.

FORD MOTOR CO (F | ▲1.91%) was the after-hours winner. Q2 operating profit of $2.5 billion easily beat the roughly $2.1 billion consensus, and management lifted the full-year operating-profit range to $10 to $11 billion (from $8.5 to $10.5 billion). Free cash flow guidance moved up too. The reported bottom-line loss of $1.3 billion is a red herring — that’s the previously announced $3.6 billion write-down tied to the wound-down BlueOval SK battery joint venture with SK On. Strip that out and Ford’s core business is running better than it has in a long time. The stock jumped almost 7% in after-hours trade.

APPLE INC (AAPL | ▲0.94%) meanwhile briefly touched a $5 trillion market cap during the session before slipping back, a milestone I don’t want to over-index on, but a nice data point for anyone tracking how much of this market’s whole cap is now concentrated at the very top of the pile.

Macro: Confidence Slips, Fed Up Next

The macro backdrop was quiet, with one thing worth flagging.

The Conference Board’s consumer confidence index dropped to 90.8 in July from 92.2 in June. As Dana Peterson of The Conference Board put it, consumers remain fairly negative on the labor market, though expectations for household income have gotten a touch more optimistic. Not a game-changer, but consistent with the picture of a consumer who is still spending but is becoming more picky about it. That aligns cleanly with what Visa and Coca-Cola are seeing at the register.

The Federal Reserve decides on Wednesday evening. Consensus is that Chair Kevin Warsh keeps the policy rate on hold, but the market is going to be leaning much harder on the tone of the statement and the press conference than the actual decision. What investors want is any signal on when - and under what conditions - the next move happens. Given how narrowly this market is trading between defensive winners and AI-capex worries, even a subtle hawkish shift in Warsh’s language would be enough to move things.

Oil kept sliding on Tuesday, and Trump repeated his willingness to give the Iran negotiations more room. Cheaper crude is helping the inflation narrative and hurting the energy majors, the same trade-off we saw on Monday. If Wednesday’s Fed language leans dovish, this becomes a more constructive setup for cyclicals. If it doesn’t, expect the rotation out of high-multiple tech to keep going.

Conclusion

Tuesday was a tale of two tapes stitched together into one confusing scoreboard.

Coca-Cola, Boeing and a raft of solid earnings pulled the Dow to a 1% gain. Underneath, the chip complex was sold hard on nothing more than nervous positioning ahead of the SK Hynix print and the overnight capex bombshell means Wednesday’s tape will spend the day digesting the confirmation. The Fed decision arrives with the market already halfway through this rotation, and Big Tech earnings will land into whatever tone Warsh sets.

My own bias: the tape is telling you what it wants, quality defensiveness with a clean earnings story, not another round of “spend more, trust us.” I’ll be watching whether Wednesday’s Fed language reinforces that rotation or reopens the door for the megacaps.

ChartMill Market Desk - Kristoff

With regard to the stocks discussed in the article above; the author owns individual shares in Microsoft.

This daily update is prepared by ChartMill for informational purposes only and does not constitute investment advice. Always do your own due diligence before making investment decisions.

👉 This article is also available at ChartMill.com.