Micron ( ) surged +15% the day after earnings. By the end of the week it had given every point back, closing right where it started before the report. That price action tells you something important — not about Micron’s business, which is extraordinary right now, but about where we are in the cycle and what the market is starting to understand about it.

The earnings were real. Revenue up 346%. COGS up 11%. Operating income up 1,436%. Those are not normal numbers. They happen when a company has pricing power so acute that customers have no alternative but to pay whatever is asked. That’s the position Micron and the broader memory industry sit in today. It’s also, if you understand how this industry works, almost precisely the moment to be most careful.

What’s Actually Driving the Numbers

Before getting into cycle mechanics, it’s worth being precise about what’s happening — because most coverage misses the most important detail.

Memory chips specifically — DRAM and HBM — store and move the data that AI systems process. Every model trained, every query answered, every token generated runs through memory. The AI buildout created a demand surge that existing supply couldn’t meet, and when supply can’t meet demand, prices go up.

HBM and DRAM pricing is up 60-70% in a single cycle. Micron is signing 3-5 year contracts to lock in revenue visibility. Manufacturing capacity is constrained because you can’t build a new fab in six months. So supply stays tight, pricing stays elevated, and the margin numbers look like nothing this industry has produced before.

The three memory companies: Micron, Samsung, and SK Hynix, could soon rank among the ten largest market caps in the world simultaneously. Three memory companies in the top 10(!) That has never happened. Ever.

The detail that matter the most is that this is a price-driven boom, not a volume-driven one. Units aren’t exploding. Prices are. That distinction is critical when you start thinking about what happens on the other side, because price increases are far more fragile than volume growth. Volume reflects structural demand. Price reflects a temporary supply-demand imbalance. One of those sustains a cycle. The other ends it.

The Stock Peaks Before the Earnings Do

This is the trap. And it catches sophisticated investors every single cycle.

My prediction is that Micron’s stock price will likely peak in late 2026 or early 2027. Earnings will likely peak in 2027-2028. That gap — roughly two years between when the stock tops and when the business tops — is where most investors lose money in semiconductor cycles. They look at the earnings, see them still growing strongly, conclude the stock is cheap relative to those earnings, and buy. Then the stock goes sideways for 18 months while earnings remain strong and analysts keep raising targets. Then the cycle turns and the stock falls 50-60% while everyone is still pointing to the fundamentals.

Nvidia ( ) is already showing you this in real time. The stock is roughly flat compared to pre-Halloween 2025, despite analysts revising 2027 earnings estimates up materially — from mid-20s percent growth to 40% growth. The earnings story got better. The stock went nowhere. When a stock underperforms despite significant upward earnings revisions, the market is telling you the cycle is maturing. That signal is worth taking seriously.

The reason stocks lead earnings by two years is straightforward. Markets price future expectations, not current results. By the time Micron is printing those 1,400% operating income growth numbers, the market is already asking what 2028 looks like — and the 2028 picture is considerably less comfortable.

The Hyperscaler Math

To understand why the cycle turns, you have to understand who’s buying all this memory and what their financial situation actually looks like.

The four major US hyperscalers — Amazon, Microsoft, Alphabet, and Meta — are collectively projected to spend $700-725 billion on capital expenditure in 2026 alone. Amazon at $200 billion. Microsoft at $190 billion. Alphabet at $180-190 billion. Meta at $125-145 billion.

Now look at what they earn. Meta’s projected net income this year is $74-85 billion against $125-145 billion in capex. Amazon earns $90-105 billion against $200 billion in spend. The gap between what these companies earn and what they’re spending is not a rounding error. It’s $50-100 billion per company, funded through cash depletion and debt issuance. Wall Street projects up to $1.5 trillion in total debt issuance across hyperscalers over the next several years to sustain this buildout.

That pace cannot continue indefinitely. Hyperscaler capex is growing 80% in 2026. Consensus estimates have it growing 22% in 2027. Then flat to single digits in 2028. You cannot keep doubling from $700 billion. At some point the boards of these companies look at the gap between what they’re earning and what they’re spending and decide the rate of increase has to slow. That deceleration — not a collapse, just a deceleration — is the trigger for the chip cycle turning.

There’s a compounding problem that I don’t see a lot of people talking about. When Meta spends $100 billion on chips it doesn’t expense it immediately. It depreciates over 5-6 years, roughly $20 billion in annual charges against earnings. But they spent more last year than the year before, and more this year than last year. Those depreciation layers stack on top of each other. GAAP earnings for Meta, Amazon, and Microsoft will be materially weaker than revenue growth implies for the next several years, regardless of how strong the top line looks. The forward PE looks reasonable on revenue multiples. Run the depreciation math properly and it looks considerably less so.

Subscribe now

The Zero-Sum Nature of the Memory Boom

On the day Micron reported earnings, something interesting happened in the market. Memory producers — Micron, Samsung, SK Hynix, Seagate, Western Digital — all gained market cap. Apple, Microsoft, Meta, Amazon, and Alphabet all lost market cap. The net result was negative. More value was destroyed on the consumer side than created on the producer side.

That’s not a coincidence. Memory pricing is a direct transfer of margin from the buyers to the sellers. When Micron charges 60-70% more for DRAM, that cost lands directly on hyperscaler capex budgets. The memory producers win exactly as much as the hyperscalers lose — and right now, the hyperscalers are already spending more than they earn. Every dollar of pricing power Micron captures is a dollar that accelerates the moment when the hyperscalers decide to slow their spend.

The boom contains the mechanism of its own ending. That’s what makes it a cycle rather than a trend.

What Happens Next and Where the Opportunity Actually Is

Late 2027 into 2028, Micron’s new fab capacity comes online. New supply hits the market exactly when hyperscaler capex growth is decelerating from 80% toward flat. Supply surge meets demand deceleration. Margins compress. The stock, which (from my projections) peaked 18-24 months earlier, is already well into its down cycle by the time the earnings confirm what the price action already told you.

Buying at current valuations likely means dead money through most of 2027 followed by a painful drawdown when the turn arrives. Not a crash next week. A long grinding period where the earnings look fine, the analysts stay bullish, and the stock goes nowhere — and then the cycle does what cycles always do.

The more interesting question is where the opportunity is on the other side of this. All that AI infrastructure spend eventually generates returns — but it does so through the software layer, not the hardware layer. Every AI agent deployed inside an enterprise needs governance, orchestration, compliance tracking, and identity management. That’s ServiceNow ( ). That′s Salesforce( ). The enterprise software names that have been left completely behind while the chip trade absorbed every available dollar of attention and capital.

The rotation from infrastructure to software is coming. It comes when hyperscaler capex growth decelerates and the market starts asking which companies benefit from the AI that’s already been built rather than which companies sell the components to build more of it. I believe the window to build those positions — before that narrative shift happens — is roughly the next 6-9 months.

The memory industry has been cyclical for 50-70 years. Long-term demand has grown exponentially the entire time. So has the pattern of 50-60% drawdowns when the cycle turns. Both things have always been true simultaneously. The current excitement seems to have forgotten the second one.

Disclaimer: Nothing here is financial advice. This is original research and personal opinion. Always do your own work before making any investment decision.

So that’s what this is. I’m expanding it to five names, adding ServiceNow and Premium Snacks Nordic to the mix, and walking through why I’m genuinely excited about each of these at current prices. Not excited in the way that every stock always looks good in a bull case — but excited in the way where I’ve done the work, I understand the downside, and I still think the risk/reward is compelling enough to own.

These aren’t momentum plays. Most of them are in sectors the market actively dislikes right now. That’s part of why they’re interesting.

1. e.l.f. Beauty ( )

The selloff was real. Down 65% from its highs. But the reason behind it was a huge overreaction and highly irrational.

In August 2025, management raised prices by $1 across all SKUs to offset tariff costs that briefly hit 170% on Chinese imports. Unit volumes fell. The market looked at slowing growth numbers and concluded the brand was structurally broken. Then management cut a single SKU from $18 to $14 and saw a 38% unit lift almost immediately. That is not what structural demand destruction looks like.

The underlying business has 29 consecutive quarters of net sales growth. 920 basis points of US market share gained over seven years. Four brands each generating over $200M in annual retail sales. Rhode — Hailey Bieber’s skincare brand, acquired last year — is already the #1 brand at Sephora North America, and it’s in less than 20% of Sephora’s global stores. A 19-country European launch is coming this September and isn’t in any guidance numbers yet.

ELF currently trades at roughly the same EBITDA multiple as Ulta Beauty ($ULTA), which is growing revenue at 3%. ELF is growing at 3-4x that pace with higher margins and a brand portfolio with far more international runway. International is only 21% of sales today versus 70%+ for legacy peers. On my FY2028 numbers, using multiples still 40-50% below this business’s own historical range, I get to a price target implying 100%+ upside from here.

I’m long and added them to my portfolio for $54

2. Premium Snacks Nordic ( $SNX.ST )

This is easily the most niche name on this list, and probably one that almost none of my North American readers have heard of.

That’s exactly why I find it interesting. Premium Snacks is a Swedish small-cap snack company with a market cap of just ~458 million SEK, but I believe it has one of the most attractive risk/reward profiles in my portfolio.

The company has quietly transformed from a niche producer of chips and pick-and-mix products into a scalable branded snacks platform. Revenue has compounded well above the broader snack market, management is now targeting 10%+ annual sales growth through 2029, and, more importantly, the heavy investment phase appears largely complete. That means future growth should require far less incremental capital while operating leverage begins to show through.

The real story isn’t gross margin expansion. It’s SG&A leverage. As more products secure central listings across Sweden’s largest grocery chains, fixed costs are spread across a much larger revenue base. Management recently raised its long-term EBITA margin target from 6% to above 8%, and after listening to management discuss the business, that target doesn’t strike me as overly ambitious.

There are still risks. Commodity costs, foreign exchange, and execution all matter, while international expansion is still in its early innings. But that’s also what makes the opportunity asymmetric. If management simply executes on the targets they’ve already laid out, I believe the business deserves to trade closer to 10–12x EV/EBITDA, supporting a fair value around 45–50 SEK per share, and I see the stock more than doubling over the next year or so, from todays compressed levels at 33 SEK per share.

This isn’t the type of stock that will make headlines on CNBC. It’s a small, underfollowed business that continues to execute quarter after quarter, and those are often the kinds of opportunities I like the most.

Subscribe now

3. Celsius Holdings ( )

Celsius dropped 40% year to date despite reporting record Q1 revenue of $783M. The market is punishing it for integration complexity, trough margins, and a Texas AG investigation into Alani Nu’s marketing practices. I think the selloff has created one of the better risk/reward setups in consumer right now.

One in every five energy drinks sold in America is a Celsius Holdings product. The company controls 21% of the US energy drink market through three brands — Celsius, Alani Nu, and Rockstar — distributed exclusively through PepsiCo’s national network. PepsiCo doesn’t just move the product. They own 11% of the equity, granted Celsius exclusive energy category captaincy, and have every financial incentive to keep growing the portfolio. That distribution moat took decades to build and cannot be replicated quickly.

The Alani Nu acquisition for $1.8B is already looking cheap. One year in the brand is doing $1.5B annualized revenue growing triple digits at scanner. It has overtaken Celsius core as the largest brand in the portfolio, has a loyal female consumer community built entirely on social media, and still has massive white space in convenience channels completing their first full reset this summer. International Alani Nu hasn’t started yet — that is a decade of runway not yet in any financial model.

Gross margins are at trough — 48.3% in Q1 versus 52% pre-acquisition. The compression came from Alani Nu and Rockstar both carrying lower margin profiles upon acquisition. Management has a specific path back — raw material alignment complete, North Carolina manufacturing line contributing H2, orbit model optimizing freight. They have missed this timeline before which is why I model conservatively at 50.5% by 2027 rather than management’s more optimistic guidance. Even at my conservative assumptions the margin recovery story adds significant earnings power independent of revenue growth.

The numbers I am working with: 2028 EPS of approximately $2.60, applying 20-22x gives a fair value of $57-74 per share from today’s $29.

At $29 you are buying a business with 21% market share, a locked-in PepsiCo distribution moat, and two growth brands at trough margins with a market that has completely lost patience. The risk/reward at this price is the most attractive it has been since the stock was building its initial distribution gains in 2021.

4. ServiceNow ( )

ServiceNow dropped 60% from its highs after the market decided AI agents would eat their lunch. I think that’s the wrong conclusion for this specific company — and the gap between that fear and the underlying reality is why I bought at $89.

The business is genuinely exceptional. $27.7 billion in contracted backlog. 97% renewal rate. 85% of the Fortune 500 as customers. Revenue growing at 20%+ at $15 billion in scale, which almost no enterprise software company has ever sustained. Operating margins at 31% and expanding. This is not a business in trouble.

The AI disruption argument misses something fundamental. Every AI agent deployed inside an enterprise needs governance, identity management, compliance tracking, and orchestration. That’s exactly what ServiceNow provides. More agents means more ServiceNow, not less. Jensen Huang said this directly at their conference a few weeks ago — agentic AI requires ten times more computation than simple generative AI, and all of it needs the governance infrastructure that ServiceNow has spent 22 years building.

Their AI product suite — Now Assist — crossed $750 million in annual contract value in Q1. Management raised the full year target from $1 billion to $1.5 billion on the earnings call. The 2030 vision is $9-10 billion in AI revenue alone, representing 30% of total contract value. That’s the hockey stick forming in real time.

Three recent acquisitions complete the picture. Moveworks gives them the AI conversational front door. Veza governs identity for AI agents in real time. Armis discovers every connected asset across the enterprise — IoT, operational technology, medical devices, shadow IT. Together they form the only complete AI governance stack in enterprise software. No competitor has all three layers.

The stock trades at 22x non-GAAP forward earnings. Salesforce trades at 27x growing at 9%. Workday at 25x growing at 15%. ServiceNow is growing faster than both at larger scale and trading at a discount. That gap closes one of two ways — the business slows down, or the market re-rates it.

I’m betting on the re-rating. Fair value today is $145-155. In two to three years I think this is a $200+ stock.

I’m long and added them to my portfolio for $89

5. Thermal Energy International ( $TMG.V )

Just like Premium Snacks, most people have never heard of this one.

Thermal Energy is a $28M market cap Canadian company that helps Pepsi, Danone, AstraZeneca and Bristol Myers Squibb cut their fuel bills by recovering heat that would otherwise be wasted. Simple concept. Enormous market.

Here’s the setup. Two thirds of all energy consumed in the US is simply wasted — most of it as industrial heat. The entire clean energy movement is focused on electricity. Solar, wind, batteries. None of that touches industrial thermal waste. You cannot put a solar panel on a pharmaceutical boiler. That gap is exactly where Thermal Energy operates, and it’s a gap that gets more urgent every year as companies face carbon reduction targets with no renewable solution for their heat emissions.

The business is already profitable — three consecutive years in the black, debt free, $4M cash, buying back shares. Revenue has compounded at roughly 20% coming out of COVID and is now at record levels. They’ve penetrated around 100 sites across their top customers who collectively have around 1,000 sites globally. The runway is obvious.

What makes this asymmetric is the combination of a tiny market cap, a proven and recurring customer base of household name multinationals, a fixed cost base that’s done growing, and multiple new revenue channels — an indirect sales rep network and simplified turnkey projects — that don’t yet appear in the numbers.

There are definitely risks here. It’s a micro-cap. Revenue is lumpy. One bad order intake quarter looks ugly. Liquidity is thin.

But when you find a profitable, growing, debt-free small company solving a structural problem for Pepsi and AstraZeneca — trading at under 1x revenue — the asymmetry is hard to ignore.

If you missed Bloom Energy at $20. Missed Rocket Lab at $5. Missed Micron before the memory supercycle. You know the feeling of watching a name run 10x while you were waiting for a “better entry.”

You wait for a cleaner chart. A bigger company. More coverage. More certainty.

And then the stock runs while you are still waiting.

The best small-cap opportunities usually do not look perfect at the beginning. They look boring. Too small. Too illiquid. Too lumpy. Too easy to ignore.

Well, there’s a stock that almost no one is talking about, and the setup looks strikingly familiar.

I’m going to lay out the full case. You can be the judge.

The company is Thermal Energy International Inc. Ticker: $TMG.V

Thermal Energy is a tiny Canadian energy-efficiency company trading around CAD $0.160 per share with roughly 171 million shares outstanding.

It does not make solar panels.

It does not make batteries.

It is not an AI story.

It helps large industrial customers recover wasted heat, reduce fuel use, lower emissions, and save money.

That last part is the key.

This is not “green tech” built on hope. Thermal sells a product that pay for itself through lower energy costs.

Lower fuel bills. Lower emissions. Better plant efficiency.

That is a much stronger pitch than asking customers to spend money just to check an ESG box.

What this company actually does

Factories waste a lot of heat.

Boilers lose energy. Steam systems leak efficiency. Exhaust streams carry usable heat out of the building. Industrial plants burn fuel, lose heat, and repeat the cycle every day.

Thermal Energy sells systems that capture some of that waste and turn it back into usable energy.

Its products include heat recovery systems, steam-efficiency solutions, and turnkey industrial energy-efficiency projects.

The end markets are exactly where this matters most: food & beverage, pharmaceuticals, and manufacturing.

These are not customers trying to look green for a press release.

These are customers with real energy bills.

The customer base changes the story

This is not a random microcap trying to convince the market that its product might work one day.

Thermal has already sold to large multinational customers.

That matters because the biggest opportunity is not just winning a first project. It is expanding inside customers that already know the product.

A company like Pepsi or Danone does not operate one facility. These are global businesses with many industrial sites.

If Thermal proves the economics at one site, that can become a reference point for more sites, more regions, and more projects.

That does not make the business recurring in the traditional software sense. There is no automatic subscription renewal.

But it can become repeatable in a different way.

One site becomes two.

Two becomes five.

Five becomes a multi-year relationship.

That is the part I think the market may be underappreciating.

The numbers behind the story

Revenue is starting to move.

In Q3 FY2026, Thermal generated CAD $9.4 million of revenue, up 62% year-over-year. For the first nine months of FY2026, revenue reached CAD $26.5 million, up 15% versus last year. For a company trading around 0.8x p/s, that alone is worth paying attention to.

But the real story is margins.

Thermal has already built much of the sales, engineering, and operating platform. So if revenue keeps scaling, more of each incremental dollar should start falling to the bottom line. That is where the thesis gets exciting.

By FY2030, based on my revenue projections and margin assumptions, I expect EBIT to be more than 5x higher than FY2026 levels, with net income almost doing the same. The upside comes from operating leverage.

Revenue grows.

Operating expenses grow slower.

Margins expand.

Earnings compound much faster than sales.

That is the part the market may be missing.

And on my FY2030 numbers, at today’s price level, Thermal would be trading at roughly 3.5x P/E, 2.3x EV/EBITDA, and 0.45x sales. By simple math, that is where the rerating potential becomes obvious. If the company gets anywhere close to those numbers, applying even fair small-cap industrial multiples creates a very different valuation picture.

The order book also supports the forward-looking story. Q3 order intake was CAD $8.7 million, up 26% year-over-year. Year-to-date order intake reached CAD $26.5 million, roughly matching year-to-date revenue. On a trailing twelve-month basis, order intake reached CAD $31.3 million, which management described as an all-time high.

That matters because this is a project-based business. Reported revenue can be lumpy. But order intake tells you whether customers are still signing new work.

So the story is simple:

Revenue is growing. Order intake is at record trailing levels. Backlog remains healthy. The balance sheet has around CAD $4 million of cash and virtually no bank debt. And if operating leverage shows up, EBIT and net income can grow several times faster than revenue.

That is a rare setup for a microcap.

At today’s price, the stock is still being valued like a small, lumpy, ignored business.

The opportunity is that it may be turning into something much better.

Read my full deep dive of Thermal with my model assumptions, valuation work, upside case, and the risks I am watching here:

The energy transition actually helps the story

One question I had to answer was simple:

Does the renewable transition eventually hurt Thermal?

My conclusion: not in the near or medium term.

Renewables solve the electricity problem first. Industrial customers can buy green power, sign renewable agreements, or install solar.

But the harder problem is heat.

Food processing, pharma, and manufacturing still need steam, boilers, and process heat. Cheap, large-scale electrification of that heat is not easy today.

This actually makes Thermal more relevant.

As companies clean up their electricity supply, the remaining emissions problem becomes more concentrated in industrial heat.

And that is exactly where Thermal operates.

The real risk is not renewables killing demand tomorrow.

The real risk is customer capex cycles.

The risks are real

This is still a tiny company.

Revenue is lumpy. Projects can move between quarters. Gross margins depend on mix. Order intake matters. Customer capex budgets matter. Liquidity is limited.

The biggest risk is simple:

If order intake slows, the thesis weakens fast.

The second risk is operating leverage.

If expenses rise as fast as revenue, the earnings upside becomes much smaller.

The third risk is project execution.

Thermal does fixed-price turnkey projects. If costs are estimated poorly, margins can get hit.

But compared to many microcaps, Thermal avoids some of the worst problems.

It is not drowning in debt.

It is not pre-revenue.

It is not constantly issuing shares just to survive.

It has real customers, real profits, and a balance sheet that gives it room to execute.

That combination is rare at this size.

The thesis in one paragraph

I wrote the full deep dive with my model assumptions, valuation work, upside case, and the risks I am watching here:

Thermal Energy International Inc. ($TMG.V) is not the kind of company that screens perfectly at first glance.

It is small. It is illiquid. Revenue can be lumpy. It does not have classic recurring software revenue. It operates in industrial energy efficiency, which is not exactly the sexiest corner of the market.

But when I look deeper, I think the setup is very interesting.

Thermal Energy is currently trading around CAD $0.160 per share with roughly 171 million shares outstanding. On my FY2026 estimates, that puts the company around 16x earnings, 8.8x EV/EBITDA, and 0.8x sales.

For a normal slow-growth microcap, that would not necessarily be special.

But Thermal Energy is not a stagnant industrial company. It is a profitable, debt-free business selling energy efficiency solutions to large global customers, with growing order intake, a large installed customer runway, and what I believe could become meaningful operating leverage over the next several years.

That is the core of the thesis.

I am buying it because I believe the company has a real chance to compound revenue, expand margins, and grow earnings much faster than the current valuation implies.

In simple terms:

If Thermal Energy can grow revenue in the mid-teens, keep gross margins stable around the low-40s, and let operating expenses grow much slower than revenue, earnings can grow much faster than sales.

That is where the asymmetry comes from.

Table of Contents

1. The One-Paragraph Version

2. What Thermal Energy Actually Doe

3. Why This Opportunity Exists

4. The Customer Quality

5. The “Recurring but Not Recurring” Revenue Dynamic

6. Order Intake, Backlog, and Why I Focus on TTM Numbers

7. My Base-Case Model Assumptions

8. Valuation: Why I See Asymmetry

9. Upside Potential

10. What Could Go Wrong

11. Does the Renewable Transition Hurt Thermal Energy?

12. What I Will Watch Going Forward

13. Final Thoughts

Read more

Table of Contents

1. The One-Paragraph Version

Celsius Holdings ( ) is a three-brand modern energy portfolio controlling approximately 21% of the US energy drink market through an exclusive captaincy arrangement with PepsiCo. The company acquired Alani Nu for $1.8B in April 2025 and Rockstar Energy for $1.8B in August 2025, transforming itself from a single-brand challenger into a scaled portfolio business virtually overnight. The stock has been volatile — down significantly from its 2023 highs — as investors wrestle with integration complexity, margin compression, recurring legal charges, and a decelerating flagship brand. My view is that the market is mispricing the normalized earnings power of this business two to three years out and that the Alani Nu acquisition in particular was a genuinely good deal that will look obvious in hindsight. This is not a trade. This is a three to five year hold.

2. The Business — What They Do and How They Make Money

Celsius Holdings sells energy drinks. But the mechanism by which it makes money is more nuanced than that simple statement implies.

The company manufactures product through a combination of owned manufacturing — they acquired a co-packer facility in North Carolina in 2024 — and third-party co-manufacturers. Product is sold to PepsiCo’s distribution system at a wholesale price. PepsiCo then distributes to retailers across the United States through its Direct Store Delivery network. This is the same trucks that carry Gatorade and Mountain Dew into every convenience store, grocery chain, and mass retailer in the country. CELH recognizes revenue when product ships to PepsiCo, not when it sells through to the consumer.

This matters for two reasons.

1. First, it means CELH’s reported quarterly revenue can diverge from consumer takeaway data depending on how PepsiCo manages its inventory positioning. We saw this dramatically in Q4 2025 when reported CELSIUS brand revenue declined 8% while scanner data showed +13% consumer growth — the gap was entirely timing.

2. Second, it means CELH is fundamentally dependent on PepsiCo as both its primary customer and its primary distribution partner. That dependency is simultaneously the company’s greatest strength and its most significant single-counterparty risk.

The three brands serve three distinct consumer segments:

CELSIUS targets active health-conscious consumers — predominantly 18-35, roughly gender balanced, fitness-oriented. Born in gyms and health clubs, the brand has expanded into mainstream retail on the back of zero sugar positioning and functional health claims — metabolism support, energy without crash. Priced at approximately $3.49 per can.

Alani Nu targets women 18-44 — 70% female demographic, community-driven, social media native. The brand was built on TikTok and Instagram before Celsius acquired it, with a loyal community that treats flavor drops like cultural events. Nostalgic fun flavors, zero sugar, beauty and wellness adjacency. Priced slightly above Celsius at approximately $3.65-3.70.

Rockstar targets men 18-44 — core energy consumer, motorsports, gaming, music. A 25-year-old brand with genuine heritage that Celsius acquired from PepsiCo itself. Currently in turnaround. Priced slightly below Celsius at approximately $3.30.

3. Why This Market Is Worth Owning

The US energy drink market is approximately $28.5 billion and is one of the fastest growing categories within all liquid refreshing beverages. Energy is taking shelf space from juice, premium water, and in convenience channels increasingly from beer. Retailers are actively expanding cold door allocation toward energy because the velocity and margin profile is superior to categories it displaces.

The category tailwinds are structural not cyclical. Zero sugar energy is growing faster than the overall category. The female consumer is entering energy drink consumption at accelerating rates — historically the category was 70%+ male; that is shifting meaningfully. GLP-1 adoption creates a genuine incremental occasion — users need energy while consuming fewer calories, which maps directly onto the zero-sugar functional positioning of Celsius and Alani Nu. Social occasions are increasingly replacing alcohol consumption among Gen Z, with mocktails and energy-based social drinks becoming mainstream — CELH has positioned both Celsius and Alani Nu explicitly for this occasion.

International is the long-term kicker. The US is $28.5B. Britain alone is $2.79B. Australia $2.06B. France $945M. CELH is present in approximately 10 international markets today through a disciplined partnership model — Suntory in key European markets being the cornerstone relationship. Alani Nu has essentially zero international presence today. That is not a problem. That is a decade of runway.

4. The PepsiCo Relationship — The Moat That Doesn’t Get Enough Credit

The single most underappreciated aspect of the Celsius Holdings investment case is not the brands. It is the PepsiCo captaincy arrangement.

As PepsiCo’s exclusive energy category captain in the United States, Celsius does not just get distribution. It gets category management authority — the ability to recommend shelf placement, planogram configuration, and space allocation across PepsiCo’s retail network. This is the mechanism behind the 17% shelf space gain for CELSIUS and the 100%+ space gain for Alani Nu in 2026’s spring resets. No amount of marketing spend can replicate what category captaincy delivers. You either have it or you don’t.

PepsiCo also owns approximately 11% of Celsius Holdings equity, paid for through its Series A preferred investment in 2022. Their financial interests are aligned with CELH’s success. When CELH’s brands grow, PepsiCo’s distribution revenue grows, its category captaincy position strengthens, and its equity stake appreciates. This alignment is structural and durable.

The moat this creates is real. A competitor attempting to challenge CELH’s US position would need either their own national DSD network — which costs billions to build and decades to establish — or a comparable partnership with a major distributor. Monster has Coca-Cola. Red Bull operates its own distribution partially. The remaining addressable distribution infrastructure is limited. CELH’s PepsiCo lock-in is not easily replicated.

5. The Acquisitions

Alani Nu: $1.8B, April 2025

At acquisition Alani Nu was generating approximately $600-700M in annualized revenue growing rapidly on a direct-to-consumer and regional DSD model. The purchase price of approximately 2.5-3x revenue looks reasonable for a brand of this quality and momentum — comparable premium beverage brands have traded at 3-5x. But the real value was not what you were buying at the time of acquisition. It was what the business would become once plugged into PepsiCo’s national DSD network with CELH’s purchasing scale, manufacturing infrastructure, and category captaincy leverage.

In Q1 2026 — one year after acquisition — Alani Nu generated $368M(!) in a single quarter. It is already the largest brand in the CELH portfolio by revenue. The $1.8B acquisition price looks very cheap in hindsight.

Rockstar: $1.8B, August 2025

Rockstar is the more complicated story. CELH acquired the US and Canada rights from PepsiCo — notably PepsiCo retained international Rockstar. At acquisition the brand was already declining — scanner data showed negative growth and that trend has continued with Q1 2026 at -13% year over year.

Why acquire a declining brand for $1.8B? Two reasons. First, PepsiCo was the seller and saying no to your most important business partner is strategically complicated, particularly when they are simultaneously offering you energy category captaincy. Second, Rockstar at stabilized $260-270M annual revenue with CELH’s cost structure brings gross margin improvement potential and portfolio breadth in the male core energy segment that complements rather than competes with Celsius and Alani Nu.

Whether this was a good deal depends entirely on whether the stabilization and recovery thesis plays out. I am not modeling significant Rockstar recovery in my base case — that upside needs to be earned not assumed.

Another position I recently initiated:

Down 65%. 100%+ upside potential. I just bought.
A category-defining brand with 29 consecutive quarters of growth, a recently acquired brand already ranked #1 at one of the most prestigious retailers in the world, and a stock trading at the same multiple as companies growing 4x slower.

6. The Numbers — What I Am Modeling

Let me be fully transparent about my assumptions and the reasoning behind every major line item.

Revenue Projections

I project revenue by brand rather than consolidated because the brands have fundamentally different growth profiles and the blended number obscures what is actually happening.

The 30.5% in 2026 is almost entirely acquisition math — Alani Nu and Rockstar are full year consolidations for the first time. The underlying organic growth story is CELSIUS at 8% and Alani Nu at 45% pro forma — the latter driven by genuine consumer demand plus the convenience channel distribution build completing.

One of the more important single number in this entire model is Alani Nu’s 2027 growth rate of 20%. That is where the distribution anniversary effect normalizes and you see what the underlying organic demand actually is. If that number comes in at 10-12% instead of 20% because velocity per store disappoints, the entire thesis needs reassessment.

CELSIUS Core at 8% declining to 5% by 2030 reflects a maturing brand at near-full domestic distribution (98.5% ACV) that grows through velocity improvement, usage occasion expansion, Fizz Free platform building, and modest international contribution. This is not an exciting growth assumption. It is a rather conservative one.

Rockstar at -5% in 2026 then 2-3% thereafter does not give management credit for the turnaround they are attempting. If it materializes that is pure upside to these numbers. My base case requires evidence before rewarding it.

Gross Margin — A Central Debate

This is the most important financial debate in the CELH investment case and the one where management has disappointed most consistently.

Management has guided toward low-50s gross margin as a target twice — first as a 2025 exit rate, then as an H2 2026 achievement. Q1 2026 came in at 48.3% with Q2 guided as a “sidestep.” The Midwest aluminum premium has been the most recent headwind but it sits on top of a structural compression from Alani Nu and Rockstar both carrying lower margins upon acquisition than the mature CELSIUS brand.

I am modeling more conservatively than management guidance.

The path from 49.5% to 52% comes from four specific things management has described with concrete timelines. The North Carolina second manufacturing line completing in H2 2026 delivering vertical integration savings. Alani Nu and Rockstar raw material costs fully aligning with CELSIUS purchasing structure by end of Q2 2026. The orbit model continuing to optimize freight and inventory positioning across the network. And mix improvement through price pack architecture as higher margin pack formats gain share.

The 52% ceiling I have placed on 2030 gross margin is deliberately conservative relative to management’s mid-50s aspiration. Rockstar will likely always carry below-average margins at the brand level. International markets in early stages carry higher logistics costs. And the track record of delayed margin recovery earns skepticism not generosity.

Operating Expenses & Margins

I want to be clear about something. Many sell-side models for CELH build in aggressive SG&A leverage as a primary driver of EBITDA margin expansion. I think that is wrong and here is why.

CELH is a brand-led business competing against the two most sophisticated marketing organizations in the beverage category — Monster and Red Bull. Cutting marketing spend to improve margins is a short-term decision with long-term consequences. The company has explicitly guided increased summer marketing investment for 2026. Rockstar needs marketing support to stabilize. CELSIUS needs investment to defend and grow share in new channels. Alani Nu’s entire business model depends on maintaining cultural relevance through continuous marketing investment.

My S&M assumption is 22% of revenue in 2026 tapering to 21.5% by 2027 and holding there. That is higher than Q1 2026’s 19.2% — deliberately. Q1 is the seasonally weakest quarter and management said summer spend increases. The flat 21.5% from 2027 onwards reflects a business that needs to maintain investment intensity to sustain the growth rates I am projecting.

G&A at 8% in 2026 tapering to 6.5% by 2030 captures genuine operating leverage on a fixed infrastructure cost base. This is where scale genuinely helps — the G&A required to run a $3.4B business is not dramatically different from running a $5B business. I have also embedded a recurring legal friction cost in the G&A assumption rather than treating every legal charge as a one-time add-back — more on this in the risks section.

The resulting EBITDA margins:

The expansion from 19.6% to 24.9% over five years is driven primarily by gross margin improvement and secondarily by G&A leverage. S&M is essentially flat as a percentage. This is the more conservative and more honest version of the margin story.

Below the Line — Debt, Interest, Tax

CELH carries $670M of term loan debt at approximately 6.78% floating rate maturing 2032. Annual interest expense is approximately $45M. There is no scheduled amortization — bullet maturity in 2032. Net debt at Q1 2026 is approximately $121M ($670M debt minus $549M cash) implying roughly 0.5x net debt to projected 2026 EBITDA. The balance sheet is not a concern.

The more important below-the-line item is the $1.76B in preferred equity — Series A held by PepsiCo and Series B issued to fund the Alani Nu acquisition. These generate approximately $56M in annual cash preferred dividends plus additional participating security allocations that reduce net income attributable to common shareholders. The total annual drag on common EPS is approximately $100M.

Tax rate normalizes toward 20-21% from the distorted 13.6% in 2025 which was artificially suppressed by distributor termination fee deductions and elevated SBC. Q1 2026’s 20% is the clean signal. The Dublin global headquarters provides modest structural benefit as international revenue grows — Ireland’s 12.5% corporate rate applies to international earnings routed through the Irish entity.

Working Capital

At Q1 2026, net working capital (AR + Inventory - AP) was approximately $998M — massively elevated due to the PepsiCo distribution transition creating a large related-party AR balance of $378M, plus Alani and Rockstar inventory builds. This is transitional not structural.

The normalized net working capital intensity for a business of CELH’s structure — selling primarily through one major distributor with growing purchasing leverage over suppliers — should be approximately 8-12% of revenue. I project NWC stabilizing at 9-11% for FY2026 with a cash inflow of $60-120M from the normalization.

At 10% NWC to revenue against 2026 projected revenue of $3.4B, implied normalized NWC is approximately $343M. The unwinding from $998M toward $343M through 2026 represents a significant working capital release that flows directly into operating cash flow. This is why 2026 FCF will look dramatically better than GAAP net income implies — the working capital release is a one-time but real cash benefit that makes 2026 a much stronger FCF year than the earnings line suggests.

From 2027 onwards NWC is modeled as a stable percentage of revenue declining gradually from 9.5% to 8.5% as purchasing leverage improves and AR terms with PepsiCo normalize.

7. The Valuation — What I Think This Is Worth

I value CELH on 2030 projections applied at multiples I believe the business deserves once the transformation is complete. I use three methodologies with specific reasoning for each.

A quick note before the numbers. Current share price is $29.12 with 256M shares outstanding, implying a market cap of approximately $7.5B. Total enterprise value adding $670M debt and $1.76B preferred, minus $549M cash equals approximately $9.4B. I am not comparing implied values to current price here — that changes daily and you can do that arithmetic yourself with the fixed targets below.

My 2030 projections:

Primary methodology: P/E at 20-25x 2030E normalized EPS

This is my anchor valuation. On normalized forward earnings that strip out one-time items — which our model already does — P/E captures every real cost including interest on the term loan, the normalized tax rate, and preferred dividend drag on common earnings. It tells you exactly what common shareholders own after all senior claims are satisfied.

At 20-25x on $3.23 EPS:
Implied fair value: $63-79 per share (16-22% CAGR over the next 5 years)

At 20x the implied 2030 fair value is $63 per share.
At 25x the implied 2030 fair value is $79 per share.

From today’s $29.12 that implies a +17-22% annual CAGR over five years.

The near-term picture is equally compelling. By 2027 the model implies $43-54 per share — a +21-36% annual return over just two years if the earnings trajectory holds and the market begins re-rating as integration complexity resolves.

The 20-25x range is justified by a business growing revenue at 7-10% annually with dominant PepsiCo captaincy distribution that no competitor can replicate. It is a premium to consumer staples at 15-18x reflecting the superior growth profile, and a modest discount to Monster’s historical 25-28x reflecting the preferred equity complexity and Rockstar drag. If the legal situation resolves cleanly and Alani Nu proves durable by 2028, 25x becomes the base case not the optimistic scenario.

Cross-check: EV/EBIT at 20-22x 2030E EBIT

EV/EBIT is my preferred enterprise value multiple because it forces D&A accountability that EV/EBITDA obscures. The $40M annual brand amortization is not truly without economic substance — brands require continuous marketing investment to maintain value. EV/EBIT deducts it.

Note: preferred equity is included in my enterprise value calculation alongside debt, meaning the implied share prices below already reflect the full $1.76B preferred claim above common shareholders. No separate bridge adjustment needed.

At 20-22x on $1.24B 2030E EBIT:

At 20-22x on 2030E EBIT the implied fair value is $81-90 per share — a +21-24% CAGR from $29.

The near-term picture is equally striking. Applying the same multiple to 2027E EBIT alone implies $55-61 per share. This is a 90-100% increase from today’s $29.12 in approximately 18 months if the earnings trajectory holds and the market begins assigning a normalized multiple as integration complexity resolves.

EV/EBIT produces higher implied values than EV/EBITDA for a straightforward reason. CELH’s D&A is modest relative to EBIT — approximately $40M annually — so the gap between the two multiples is meaningful but not extreme. The higher implied value versus P/E reflects the enterprise value approach capturing the full business earnings power before the debt and preferred claims are netted out. With relatively modest net leverage, most of that enterprise value flows directly through to common equity.

Cross-check: EV/EBITDA at 16-18x 2030E EBITDA

The industry standard metric for peer comparison. Monster trades at 17-19x NTM EBITDA currently. CELH at proven execution deserves a comparable multiple — I use a slight discount reflecting residual preferred complexity.

Note: preferred equity is included in enterprise value, so implied share prices already reflect the full $1.76B preferred claim.

At 16-18x on 2030E EBITDA the implied fair value is $66-75 per share — a +18-21% CAGR from $29.

Monster currently trades at 17-19x NTM EBITDA with cleaner balance sheet, no preferred complexity, and decades of proven execution. I apply a slight discount to that range reflecting the residual Rockstar drag and ongoing legal uncertainty. If those two overhangs resolve by 2028 the discount narrows and 18-20x becomes defensible — pushing the implied value toward $75-83.

EV/Revenue at 3-3.5x 2030E Revenue

Least precise methodology for a business with evolving margins. Included as a sanity check only.

Note: preferred equity included in enterprise value.

At 3-3.5x on 2030E revenue the implied fair value is $48-57 per share — a +10-14% CAGR from $29.

The lower implied values relative to earnings-based multiples are expected and logical. EV/Revenue applies a multiple to the top line before any costs are deducted — it cannot differentiate between a business running at 49% gross margins today versus 52% in 2030. It therefore misses the entire margin expansion story that is the core of the CELH investment thesis. The 3-3.5x range is calibrated to peers — Monster trades at 4-5x revenue at full maturity with higher margins, KDP at 2-3x with lower growth. CELH sitting between those benchmarks is appropriate but the multiple compression from earnings-based approaches reflects that revenue multiples simply cannot capture where the real value creation is happening in this business.

Use this as your downside reference. If margins disappoint and the earnings story does not materialize, EV/Revenue is approximately where the stock finds support.

DCF: 10% WACC, 2.5% terminal growth

Most conservative methodology by design. The 10% WACC reflects CELH’s beta above 1.0, preferred equity complexity, and ongoing legal uncertainty. The 3% terminal growth rate is deliberately below what a global beverage brand with Alani Nu’s international runway probably deserves — I use it to build in margin of safety rather than optimism.

Implied fair value $35-40 per share.

The DCF is the floor not the target. It answers one specific question — what is this worth if multiple expansion never materializes and I discount everything aggressively? The answer is still above today’s $29 on the conservative end, which provides some baseline downside comfort. As the integration resolves and earnings visibility improves, the appropriate WACC compresses and the DCF value moves up meaningfully — but I will not assume that in advance.

My base case fair value anchor: $72-76 per share on 2030 numbers

My base case fair value anchor on 2030 numbers is $70-80 per share, implying a 17-22% annual CAGR from today’s $29. This is what I believe the business is worth once the integration complexity has resolved, gross margins have recovered toward 52%, and Alani Nu’s growth durability is established and the market is pricing it accordingly.

But 2030 is five years away and I want to be transparent about the nearer term picture too.

2027 is the first year where the integration noise is largely behind us, Alani Nu faces its first clean comparable period, and gross margin should be approaching 50.5%. Applying the same methodologies to 2027 numbers implies $43-61 per share — representing a +48-110% return from today’s $29 in approximately 18 months. The wide range reflects exactly how much the Alani Nu velocity question and gross margin trajectory matter in the near term. Get both right and the stock re-rates sharply. Disappoint on either and the re-rating takes longer.

2028 is my preferred primary valuation anchor for the text because it is far enough to capture the full margin recovery story but close enough to be credible. By 2028 the business should be in steady state — integrations complete, margins established, Alani Nu durability either proven or disproven. Implied fair value on 2028 numbers is $50-71 per share — a +70-140% upside from today if that trajectory holds.

The honest summary is this. At $29 per share you are buying a business at trough margins, mid-integration, with a legal overhang, and a market that has lost patience with the story. If you believe — as I do — that the normalized earnings power of this portfolio in 2027-2028 is significantly above what today’s price implies, the entry point is compelling. If the Alani Nu velocity disappoints or gross margins stay stuck below 50%, the thesis takes longer to play out but the long-term compounding case does not break — it just requires more patience.

8. The Legal Situation (Potential Bear Case?)

This deserves its own section because it has evolved into a genuine bear case driver not background noise.

The pattern: Four consecutive years of material legal charges — SEC disclosure violations in 2024-2025 ($3M settlement plus ongoing investigation), a $24.6M litigation accrual in Q1 2026 for undisclosed ongoing litigation, and now a Texas Attorney General investigation announced June 4 2026 into whether CELH and Alani Nu marketed high-caffeine products to minors in violation of the Texas Deceptive Trade Practices Act. The Pomerantz securities law firm announced a securities fraud investigation two weeks later on June 16.

Why the Texas AG investigation is the most important item:

This is not a private plaintiff lawsuit. This is a state attorney general investigation with enforcement authority to seek injunctive relief — meaning the AG can force changes to marketing practices not just extract a financial settlement. Alani Nu’s entire growth model depends on social media influencer marketing to an 18-24 female demographic. The line between that demographic and minors on TikTok and Instagram is genuinely blurry because platform algorithms distribute content based on engagement not age verification. Regulators are not wrong about the ambiguity.

If the Texas investigation produces findings that restrict how Alani Nu can market on social platforms, the top of the brand’s consumer acquisition funnel is impaired. That is a business model risk not just a legal cost.

How I have handled this in the model:

I have embedded a recurring legal friction cost in G&A — approximately $25M per year — rather than treating every legal charge as a non-recurring add-back. This is more honest than management’s adjusted EBITDA presentation which strips out every legal cost. Beyond the friction assumption I have not built in a worst-case scenario regulatory outcome — that would belong in the bear case not the base case. But the Texas situation is the single item most likely to cause me to trim the position if it escalates.

What would make me exit: Multi-state coalition investigation producing platform-level restrictions on energy drink marketing to under-18s. That scenario impairs Alani Nu structurally and changes the growth math fundamentally.

9. The Bear Case

The bear case is not that CELH is a bad business. It is that you pay a growth multiple for a business that underdelivers on the two assumptions that matter most.

The specific scenario: Alani Nu 2027 pro forma growth comes in at 10-12% rather than 20% as velocity per store in new convenience doors disappoints once the distribution build anniversaries. Gross margin fails to reach 50% by Q4 2026 for a third consecutive time as aluminum costs stay elevated and the integration savings arrive later than guided. The Texas AG investigation expands to additional states forcing meaningful changes to Alani Nu’s social media marketing model. Legal costs run at $50M+ annually rather than the $25M friction assumption embedded in our model.

In this scenario 2027 EBITDA comes in around $700M rather than $830M. The market re-rates the stock toward 10-12x that disappointing EBITDA — the multiple a business deserves when execution credibility is questioned for the third consecutive year. That gives you an implied price of $20-25 per share — a 25-30% decline from todays $29.

That is the honest downside. It is real and worth taking seriously.

10. Why I Am Still Holding

There are three sources of upside that are genuinely real but that I have deliberately excluded from my base case because they require evidence not assumption.

Alani Nu international. Zero international revenue today. A brand with 70% female demographic, zero-sugar positioning, and social media native community is arguably more naturally suited to European and Asia-Pacific markets than the traditional Monster/Red Bull positioning. If Alani Nu international launches in 2027-2028 and captures even 5-10% of the UK or Australian market, that is $150-250M of incremental revenue not in any of my numbers.

Rockstar recovery. I model Rockstar at 2-3% growth from 2027. The brand has 25 years of heritage, genuine male consumer loyalty, and now the benefit of CELH’s operational expertise and PepsiCo captaincy. If the motorsports and gaming partnerships actually convert to velocity improvement and the SKU rationalization works, 5-8% growth is achievable. The management team has already demonstrated they know how to grow energy brands — Rockstar is not a fundamentally broken brand, it is a mismanaged one that found better owners.

Preferred equity resolution. The $1.76B preferred equity overhang is the most significant structural complexity depressing common equity value. If CELH uses its growing FCF to redeem or convert the preferred in 2027-2028, common equity value increases by approximately $7 per share instantly and the EPS trajectory improves materially as the $100M annual preferred drag disappears. Management has not signaled this intention but the financial capacity will exist.

Before You Go — One More Stock To Read

If you have read this far you understand that my Celsius thesis rests heavily on brand-led consumer growth through dominant distribution, operating leverage on a scaling revenue base, and a long multi-year compounding story. There is a company I have analyzed in equal depth that I believe is one of the most asymmetric setups in small-cap markets right now — a 95% gross margin monitoring annuity with 90%+ annual renewal rates, zero debt, virtually no federal taxes for the next several years, and a brand new product just hitting the market this week that the market has not priced in at all.

The base case implies roughly 100% upside over the next years on conservative assumptions. The bull case is a different conversation entirely.

It has a $41M market cap. No analyst coverage. 35% insider ownership. And a management team that has compounded revenue at roughly 20% on average over the long run from a fixed cost base that barely grows.

I published the full deep dive — every assumption, every risk, every number — for subscribers a few days ago.

I am not saying this is the same kind of business as the one you just read about. I am saying if you are building a concentrated portfolio of high conviction long-term positions, this analysis belongs on your desk before you finalize your sizing.

11. The Bottom Line

Celsius Holdings is a genuine long-term hold for me at current prices. The PepsiCo moat is real and durable. Alani Nu was an excellent acquisition that will look cheap in three years. The gross margin recovery is delayed not derailed. The normalized earnings power of this business in 2028-2030 is meaningfully higher than today’s numbers suggest and the market is not yet fully pricing that trajectory.

The risks are real — the Texas AG investigation, the preferred equity overhang, the recurring legal pattern, and the Alani Nu velocity question are all legitimate concerns that I monitor actively. If any of the load-bearing assumptions break I will tell subscribers immediately and explain why I am trimming or exiting.

But the base case — $76-85 per share fair value on 2030 normalized earnings — represents a compelling return from today’s $29.12 if execution holds. That is why I own it.

Grab a Celsius. This was a long one.

Disclaimer: This is not financial advice. This represents my personal analysis and investment thesis. Do your own research. I may hold positions in securities mentioned.

Earnings recaps. Price targets. Analyst upgrades. Twitter threads that age like milk.

None of it makes you a better investor.

What actually moves the needle is doing the work nobody else wants to do. Reading the footnotes. Understanding why a stock is down. Separating the temporary from the permanent. And having the conviction to act when everyone else has already moved on.

That is the only edge that has ever consistently worked. And it is the only edge I try to build here.

I publish one deep dive at a time — no filler, no hot takes, no recycled consensus views. Just original research, written clearly, grounded in the numbers, with my own money on the line every single time.

I do not cover 50 stocks. I cover the ones I would actually buy.

Which brings me to this week.

I have spent the last several months deep inside the filings, earnings transcripts, and investor presentations of a company most people have never heard of. It has a market cap of $41M. No analyst coverage. Trades roughly 10,000 shares on a good day. The kind of stock that never shows up on a screener because it is too small to matter to anyone managing real money.

I think that is exactly why the opportunity exists.

This is a business with 95% gross margins on its recurring revenue stream. That renews at above 90% annually with no contracts required. That has compounded its monitoring base steadily for years while quietly pivoting toward larger and stickier commercial customers. That is debt-free, cash-generative, and pays virtually zero federal income tax for the next several years.

And it just launched a product that could change the size of this business entirely.

The core business alone is worth owning at the current price. The new product — just hitting the market this week — is an option the market has not priced in at all. The addressable market is 6x the average sale size of their existing product. The urgency driver is a $500M annual theft problem that is getting worse every month copper prices rise.

My base case implies roughly 100% upside over three years. The bull case is a different conversation entirely.

Below I walk through every assumption, every risk, and every number behind that conclusion.

The stock is…

Read more

There’s a specific investing problem that doesn’t get talked about enough.

You’ve done the research. You understand the macro tailwind. You’re genuinely convinced a sector is going to be significantly larger in five to ten years than the market is currently pricing. But when you look at the individual companies — the pure plays, the early-stage names, the ones with the most upside potential — the execution risk is enormous. Long timelines. Heavy regulation. Complex operations. Multiple companies competing for the same opportunity with no clear winner yet.

This is exactly where a sector ETF becomes the right tool.

Not because you want average returns. Because you want the sector’s upside without concentrating everything on a single company’s ability to execute a decade-long business plan.

The Nuclear Setup

I’ll use nuclear energy as a concrete example because it’s one I’ve been researching deeply and recently added to my portfolio through a targeted ETF position.

The thesis is straightforward.

US data center power demand is growing from roughly 75 GW today to 134 GW by 2030. That’s the equivalent of adding Japan’s entire electricity grid in five years! This isn’t a projection built on optimism — it’s the direct consequence of AI infrastructure spending running at $400 billion per year with no signs of slowing. Training frontier models and running inference at scale requires gigawatts of stable, always-on power.

Nuclear is the only proven firm, carbon-free baseload power source that meets that requirement. Wind produces power roughly 35% of the time. Solar roughly 25%. Nuclear runs at a 90%+ capacity factor — 24 hours a day, 365 days a year, regardless of weather. For a hyperscaler running AI inference at 3am in January with no wind and no sun, there is exactly one carbon-free answer.

The hyperscalers have already reached this conclusion. Microsoft, Google, and Amazon have collectively signed over $200 billion(!) in nuclear power purchase agreements. Three Mile Island was commercially restarted for Microsoft. Dead reactors are being brought back to life specifically to feed AI data centers. When the three most cash-rich companies in the world make the same infrastructure bet simultaneously — that’s not a trend. That’s a structural shift.

Then add the uranium supply picture. New mines take 10–15 years from discovery to production. Demand is accelerating now. The IEA confirms the supply-demand gap widens structurally through the 2030s and into the 2040s. Uranium spot price sits at $90/lb — historically only seen at the back end of a major contracting cycle, never the front end.

Cameco’s CEO — four decades in the uranium industry — called this the best environment he’s ever seen.

The thesis is clear. The macro tailwind is real. The demand is contracted and signed. The supply can’t respond fast enough.

So why not just buy the individual stocks?

The Problem With Picking Individual Nuclear Names

I do own individual nuclear names. Cameco ( ), Constellation Energy ( ), and others where I have specific high conviction in the company’s positioning and financials.

But the nuclear value chain is wide. And parts of it carry enormous individual company risk.

Take the innovation layer — NuScale ( ), Oklo ( ), X-energy. These are genuinely exciting companies. NuScale is the only SMR design certified by the US Nuclear Regulatory Commission. Oklo has Sam Altman backing and is targeting criticality on test reactors by July 2026. X-energy has Amazon committed to 5 GW through AWS by 2039.

But these are pre-revenue or early-revenue businesses with 5–10 year timelines to meaningful operational scale. Heavy regulatory oversight. Complex construction. Historically nuclear projects run 102% over budget on average. The Utah project that NuScale was building got cancelled in 2023 after inflation pushed the cost per megawatt hour from $58 to $89 — making it uneconomical overnight. The business recovered. Investors who were concentrated in that single name felt every moment of that.

The question isn’t whether SMRs work. The question is which company executes, on what timeline, without a catastrophic cost overrun. That’s genuinely hard to predict a decade in advance. And if you’re concentrated in one name and they hit a major setback — a cancelled project, a regulatory delay, a funding crunch — the drawdown is severe while the broader sector thesis remains completely intact.

This is exactly the scenario where an ETF solves the problem. You’re not abandoning the thesis. You’re expressing it more efficiently.

Why the ETF Was the Right Call Here

The nuclear ETF I bought has delivered over 150% in the last three years. More importantly — it recently pulled back significantly along with the broader nuclear sector selloff in 2026.

The combination of a strong long-term track record and a meaningful drawdown is exactly when a sector ETF becomes interesting as an entry point. The individual companies sold off on sentiment, not fundamentals. Data center demand didn’t slow. Hyperscaler PPAs didn’t get cancelled. Uranium supply didn’t suddenly catch up with demand. The stocks moved. The thesis didn’t.

Buying the ETF on that drawdown means I’m positioned across the entire nuclear value chain — uranium miners, fuel processors, operating utilities, equipment manufacturers, and SMR developers — without having all my exposure concentrated in any single company’s operational timeline.

If Cameco executes perfectly — I capture that. If Constellation keeps printing cash from its 22-reactor fleet and long-term Microsoft contract — I capture that. If NuScale’s TVA project delivers and the stock re-rates — I capture that. And if one of the SMR developers hits a setback that sends their stock down 40% — it’s a fraction of the portfolio rather than a catastrophic loss.

That’s the structural advantage of the sector ETF when you have high conviction in the theme but distributed uncertainty across individual names.

How I Chose the Right ETF

Not all nuclear ETFs are built the same. Before buying I ran several through a screener and comparison tool — looking at five specific things:

This Is Not a Buy and Forget Position

This is worth being clear about because it’s different from how most people think about ETFs.

A broad market index fund like VTI — tracking the entire US stock market — is genuinely a hold forever position. You’re not trying to time the market. You’re owning the market. Trimming or adding based on valuation makes little sense because the market is always fairly priced by definition.

A sector ETF is different. It’s closer to owning a stock in terms of how you should think about it. I have a thesis. I have a price target range in mind for the sector. I have a view on when the thesis matures. When the nuclear sector has run another 80–100% from my entry point — or when the fundamental drivers I’m tracking start to weaken — I’ll trim the position. If it pulls back further on continued sentiment weakness while the fundamentals stay intact, I might add.

That’s active management expressed through a passive vehicle. The ETF handles the stock selection within the sector. I handle the sector-level allocation decision.

That combination — sector conviction at the top level, diversification at the stock level — is what makes this approach genuinely powerful for investors who do their own research.

When a Sector ETF Makes Sense

Not every investment idea is better expressed through an ETF. Sometimes you have specific single-stock conviction that’s strong enough to justify concentration. Cameco’s unique positioning — mining 14–20% of global uranium supply while owning 49% of Westinghouse — is specific enough that I own it directly.

But a sector ETF makes sense when:

Nuclear energy right now fits every one of those criteria. SMRs are real but unproven at commercial scale. Multiple companies are competing for the same contracts. Regulatory timelines are long and unpredictable. Construction costs are historically unreliable. And yet the demand driver — AI infrastructure needing 24/7 carbon-free power at gigawatt scale — is one of the clearest structural trends in markets today.

The sector ETF lets me own that trend without betting the entire position on any single company’s ability to execute a decade-long plan without a major setback.

The Bottom Line

Getting the sector right is hard enough. Don’t make it harder by forcing yourself to also get the single-stock call right at the same time.

When you have genuine conviction in a macro theme — when the demand drivers are clear, the supply constraints are structural, and the market is still in the early stages of pricing the opportunity — a targeted sector ETF is often the most intelligent way to express that view.

You capture the sector’s upside. You diversify the execution risk. And you keep your powder dry for the individual names where you have specific, differentiated conviction on top of the macro call.

Right about the sector. Distributed on the stock. That’s the framework.

Disclaimer: Nothing here is financial advice. This is original research and personal opinion. Always do your own work before making any investment decision.

Nuclear stocks have cratered 30-70% from their peaks despite fundamentals that’ve only strengthened. The grid’s failing, AI data centers need power, and electrification isn’t slowing down. Markets are pricing in failure when the underlying thesis has gotten more compelling, not less.

From venture-stage micro-reactor plays trading at desperation levels to proven operators with actual revenue streams, this selloff has created entry points that won’t last. The disconnect between price action and reality rarely gets this wide.

I’m tracking five nuclear stocks right now, each representing different angles on what I believe will be the most significant energy shift in decades. Here’s why I’m bullish on nuclear as a sector and how I’m positioning for the recovery.

Why I Am Bull On Nuclear This Decade

The math is simple. U.S. electricity demand will grow 15-20% by 2030, driven by data centers, electric vehicles, and industrial reshoring. Wind and solar can’t provide baseload power, and natural gas faces increasing regulatory pressure. Nuclear is the only carbon-free technology that runs 24/7 regardless of weather.

Microsoft’s 20-year deal with Constellation to restart Three Mile Island proves tech companies are serious about nuclear. Google just signed agreements for 500MW of small modular reactor capacity. Amazon invested $500 million in nuclear development through its Climate Pledge Fund. When Big Tech commits this kind of capital, they’re not speculating, they’re solving infrastructure problems.

The regulatory environment has shifted dramatically. The ADVANCE Act passed with bipartisan support, streamlining nuclear project approvals. The Nuclear Regulatory Commission approved NuScale’s SMR design, the first in U.S. history. The Department of Energy allocated $3.46 billion for domestic uranium enrichment. Policy support is real, not theoretical.

1. Oklo ( )

Oklo sits at an $11.5 billion valuation for a company that’s still figuring out its micro-reactor designs. That’s venture-stage risk with public market liquidity. This is exactly the kind of setup that creates 10-baggers or zeros.

The company’s developing reactors smaller than SMRs that’ll deploy directly at data center sites. These replace the diesel generators currently powering AI facilities, solving the noise and pollution problems that make data centers unwelcome neighbors. Each reactor produces 15MW, perfect for edge computing applications where grid connections are expensive or unreliable.

Oklo’s business model is power-as-a-service rather than reactor sales. They’ll own and operate reactors while selling electricity under long-term contracts. If they execute, gross margins could hit 60-70% since fuel costs are minimal and maintenance is largely automated.

The risk is execution timing. It usually takes months between each meaningful progressive update, and their Aurora reactor hasn’t received NRC approval yet. The Department of Energy recently withdrew their site permit for Idaho National Laboratory, forcing them to find alternative testing locations. Any concrete development news should provide a significant bounce from these levels, but the path to commercial deployment remains unclear.

2. NuScale Power ( )

Nuscale offers a similar risk/reward profile at $11.40 versus $50+ peaks. Their certified small modular reactor design gives them a multi-year head start over competitors, plus they’ve got credible partnerships. Research agreements with Oak Ridge National Laboratory and a deal with TVA for 10 SMR units in Tennessee.

NuScale’s reactors produce 77MW each and can be deployed in modules of up to 12 units. Their VOYGR-12 plant would generate 924MW, comparable to traditional nuclear plants but with improved safety features and faster construction timelines. The modular design allows utilities to add capacity incrementally rather than committing to massive upfront investments.

The company has international momentum with Romania approving a 6-module plant and Poland evaluating SMR deployments. Their reactor design uses natural circulation instead of pumps, eliminating single points of failure that plague traditional reactors. If small modular reactors achieve commercial viability, NuScale has the regulatory approval and partnership network to dominate early deployments.

Both names trade together, suggesting sentiment-driven selling rather than company-specific issues. For investors comfortable with binary outcomes, current prices offer asymmetric upside if the nuclear thesis plays out.

3. Constellation Energy ( )

Constellation Energy dropped 21% over 90 days despite being the closest thing to a sure bet in nuclear. They’re restarting Three Mile Island for Meta, which is the first major reactor restart in U.S. history. If the grid needs power so badly that we’re bringing dead reactors back online, that tells you everything about supply/demand dynamics.

CEG operates the largest U.S. nuclear fleet with 21 reactors across 12 plants. Their capacity factor consistently exceeds 95%, meaning their reactors run at full power almost constantly. Unlike natural gas plants that ramp up and down based on demand, nuclear provides steady baseload power that’s increasingly valuable as grid complexity increases.

The company has locked-in power purchase agreements with major AI players, providing revenue visibility that speculative names lack. Their PPA with Meta runs for 20 years at undisclosed pricing, but industry estimates suggest $80-100 per MWh — well above their operating costs of $30-35 per MWh. That’s 60%+ gross margins on predictable volume.

CEG trades like a tech stock because of its AI data center exposure, not like a traditional utility. That correlation works both ways — it amplifies upside during bull markets, but creates volatility during selloffs that have nothing to do with nuclear fundamentals.

Scotia Bank’s $481 price target implies substantial upside from current levels, and the company generates actual cash flow from existing operations while investing in growth projects.

Beyond reactor restarts, CEG is exploring life extensions for existing plants and evaluating SMR deployments at current sites. They’ve got the regulatory relationships, grid connections, and operational expertise to capitalize on nuclear expansion more efficiently than pure-play developers.

4. BWX Technologies ( )

BWX is the Home Depot of nuclear. They make reactor components for Navy vessels and increasingly for commercial reactors. The stock’s near 52-week highs because defense spending is predictable, but the real opportunity is commercial nuclear expansion.

BWXT manufactures reactor pressure vessels, steam generators, and fuel assemblies. They’re the exclusive supplier of naval reactor components, generating $1.5 billion annually from defense contracts. This provides steady cash flow and technical expertise that translates to commercial applications.

The company supplies components for NuScale’s SMRs and will likely serve other reactor developers as the market scales. Each SMR requires a smaller pressure vessel than traditional reactors, but deployment numbers could be 10x higher. BWXT’s manufacturing capacity can support 10-12 SMR modules per year currently, with plans to expand based on order visibility.

They’re positioned to benefit from every reactor built, restarted, or retrofitted in the coming decade. The parts business has better margins and less execution risk than building reactors yourself. BWXT doesn’t need to prove reactor designs work, they just need nuclear capacity to grow.

With 95% institutional ownership, smart money recognizes BWXT’s leverage to nuclear growth without the binary risk of individual reactor projects. As production moves from prototypes to commercial deployment, parts demand accelerates exponentially. Their backlog hit $2.8 billion last quarter, up 15% year-over-year, suggesting momentum is building.

5. Cameco ( )

Cameco is the only uranium miner worth owning. Their Westinghouse acquisition provides diversification beyond pure mining, and they’re restarting tier-one assets like McArthur River. Most importantly, they’re North American-based, avoiding the geopolitical risks that plague other uranium producers.

Uranium prices have stabilized around $80+ per pound after hitting $100 in 2023. Cameco’s all-in costs are $35-40 per pound, providing substantial margins even if prices decline. They control 18% of global uranium production and have reserves to support 30+ years of mining at current rates.

The Westinghouse acquisition transformed CCJ from a pure commodity play into a nuclear services company. Westinghouse designs reactors, provides fuel services, and handles plant maintenance. This is higher-margin businesses with longer contract terms. This diversification reduces cyclical mining volatility while maintaining exposure to nuclear growth.

Mining stocks typically suck — flooded mines, permitting delays, cost overruns. CCJ sidesteps most of these issues through geographic diversification and the Westinghouse services business. They operate in Canada, Kazakhstan, and Australia with established infrastructure and regulatory relationships.

Canadian Imperial Bank’s $202 price target suggests significant upside from current levels. Every reactor restart, every new SMR deployment, every data center that goes nuclear needs fuel. CCJ controls enough supply to benefit from this demand inflection without the execution risk of unproven technologies. Their long-term contracts with utilities provide revenue visibility extending through 2030.

Disclaimer: This is not financial advice. Do your own research before making any investment decisions. These are analytical observations and a framework for thinking — not buy or sell recommendations.

The SpaceX IPO arrives June 12th at a $1.75–2 trillion valuation. Every obvious ecosystem play has already run. Rocket Lab is up 72% ( ), AST SpaceMobile up 65% ( TS ), Intuitive Machines up 54% ( ) , Redwire up 101% since early May.

Read my full Redwire writeup here:

The sector rerating is well underway. Capital is flooding into space before the IPO, not after. Analysts are scrambling to reprice the entire sector against a $2 trillion SpaceX comparable. Space ETFs are seeing inflows. Every name with a credible space story has been bid up aggressively.

One name still hasn’t moved much. At just over $7 per share is Momentus Inc ( ). $75 million market cap. Direct SpaceX integration. Active defense contracts with DARPA, the US Space Force, and NASA. A 9x revenue growth guide for the next 12 months. A freshly cleaned balance sheet. And a sector tailwind that is about to get considerably louder when the road show starts June 4th.

This is not a polished large-cap thesis. This is a high-risk, asymmetric micro-cap setup at a specific moment in time. The window is short. Treat it accordingly.

What Momentus Actually Does

Most space companies get lumped together as if they’re all doing the same thing. They aren’t. Understanding exactly where Momentus sits in the space economy is what makes the setup interesting.

Momentus operates orbital service vehicles (effectively space tow trucks). The mechanics are straightforward. SpaceX launches a Falcon 9 carrying dozens of payloads from multiple customers in a single rideshare mission. Momentus’s Vigoride vehicle is one of those payloads. Once the Falcon 9 reaches orbit and deploys its passengers, Vigoride detaches and becomes an independent spacecraft. From there it acts as a last-mile delivery driver in space — moving customer satellites from the Falcon 9’s drop-off orbit to their precise final operational orbits, hosting customer payloads for extended periods, and providing in-orbit services and propulsion that SpaceX doesn’t offer and has no interest in offering.

SpaceX is the highway. Momentus is the delivery van operating on it.

Amazon doesn’t drive every package to your door — UPS does. SpaceX doesn’t position every satellite in its final orbit — Momentus does.

The analogy isn’t perfect but the business logic is clean. Every additional SpaceX launch creates a potential Momentus revenue opportunity. Every increase in SpaceX launch frequency, payload capacity, and rideshare volume expands Momentus’s addressable market directly and proportionally.

This isn’t a theoretical future relationship. In March 2026, Momentus successfully launched their Vigoride 7 vehicle aboard SpaceX’s Transporter 16 rideshare mission carrying 10 customer payloads. The mission is live and operational. Vigoride 8 is already fully manifested for early 2027 — a confirmed, contracted seat on a future SpaceX launch. Momentus is one of the very few companies licensed and operationally integrated to ride SpaceX rideshare missions. That licensing and integration is a meaningful barrier to entry that the market is currently assigning zero value to.

SpaceX has publicly stated it wants to launch more frequently, carry more payloads per flight, and drive costs lower. Every one of those stated objectives is a direct tailwind for Momentus’s revenue opportunity. The highway is getting wider and the speed limit is going up. Momentus already has a van on the road.

In 2025, Momentus generated $1.1 million in revenue. Management is guiding $10 million for 2026. That is a 9x increase in a single year.

In absolute terms, $10 million is a small number. But valuing a company purely on absolute revenue at this stage misses the point entirely. What moves micro-cap stocks is the rate of change. The inflection from irrelevant to relevant, from pre-revenue to real revenue, from speculative to operational. A business going from $1.1M to $10M in twelve months while simultaneously securing defense contracts, cleaning its balance sheet, expanding its facilities, and operating confirmed missions on the world’s dominant launch platform is a fundamentally different business than the one the market priced six months ago.

At $10M in guided 2026 revenue against a $75M market cap, Momentus trades at 7.5x forward sales. Rocket Lab, who are the most credible small launch competitor in the US, trades at 90x sales. AST SpaceMobile trades at 500x sales. Redwire, after a 101% run, is at roughly 6x sales. Momentus is in the same neighborhood as Redwire on a sales multiple basis and a fraction of every other operational space name. The rerating math is straightforward if the revenue guide is anywhere close to accurate.

The $10M guide is also conservative relative to what the contract pipeline suggests. DARPA contracts, Space Force contracts, NASA partnerships, and confirmed Vigoride 8 bookings are all incremental to whatever base assumptions underpin that number. If defense task orders materialize in the second half of 2026, the gap between the guide and actual results could be significant.

The Defense Angle

The commercial rideshare story is the headline. The defense angle is what gives this thesis staying power beyond the SpaceX IPO cycle.

Government dollars are already flowing into Momentus. The Vigoride 7 mission was partially backed by a $4.2 million DARPA contract. The Defense Advanced Research Projects Agency doesn’t write checks to companies whose technology doesn’t work. A separate $1.9 million contract with Space Works, the US Space Force’s innovation arm, adds a second government revenue stream. NASA partnerships provide a third. None of these are enormous in absolute terms. For a $75M company they represent meaningful revenue validation and (more importantly) a proof of concept that the US government considers Momentus’s orbital service capability worth paying for.

The larger opportunity is what’s coming. The US Space Force FY2027 budget request stands at $71 billion, more than double the prior year’s allocation. That number reflects a genuine strategic shift in how the US military is thinking about space-based assets. Two specific programs within that budget are directly relevant to what Momentus does.

The Golden Dome Space-based Interceptor Program involves deploying and positioning space-based missile defense assets in precise orbits. MDA Shield similarly requires the movement, servicing, and ongoing support of space-based defense infrastructure. Both programs need exactly what Momentus’s orbital service vehicles provide, which is the ability to move assets from where a rocket drops them to where they actually need to be, and to service them once they’re there.

A task order win in either program would be a transformative catalyst for a $75 million market cap company. Management believes this is achievable within the next 6 to 18 months. The Space Force budget more than doubling suggests the demand is real. The question is execution and timing, both of which carry meaningful uncertainty at this stage.

What makes the defense angle structurally valuable beyond the near-term catalyst potential is contract visibility. Commercial rideshare revenue fluctuates with launch schedules and customer demand. Government contracts come with defined scopes, defined payments, and defined timelines. A single meaningful Shield or Golden Dome task order transforms Momentus’s revenue visibility from speculative to contracted — and contracted revenue at this stage of the company’s development is worth considerably more than the raw dollar amount suggests.

The Balance Sheet Cleanup

On April 20th, Momentus eliminated its convertible debt entirely. For a micro-cap this is more significant than it sounds.

Convertible debt on a small balance sheet is a structural problem. Converts create constant dilution fear. The market discounts the equity because it knows converts can be exercised into shares, flooding supply at inopportune moments. They create ceiling pressure on the stock because sophisticated investors short against the convert as a hedge. And they create creditor pressure on management that distracts from operations. All three of those dynamics suppress the stock regardless of what the underlying business is doing.

Eliminating the converts entirely removes that ceiling. The capital structure is now clean — straightforward equity, no overhangs, no creditor pressure, no dilution risk from structured instruments. That alone re-opens the stock to a class of investors who avoid companies with convert overhangs.

Simultaneously, Momentus moved into a 61,000 square foot facility in San Jose and quadrupled their clean room capacity. Clean room capacity is the physical constraint on how many vehicles they can build and prepare simultaneously. Quadrupling it is an operational statement. Management is building for significantly higher mission volume, not managing for today’s revenue base. Combined with the balance sheet cleanup, the company looks operationally and financially more like a business preparing for a meaningful step-up in activity than one managing a slow wind-down.

The timing of both moves (April 2026, six weeks before the SpaceX IPO road show) is not accidental.

What makes this setup more interesting rather than just speculative is the sequence of distinct catalysts running through the next 12 to 18 months. Each one is independent. Each one has the potential to be a meaningful mover on its own. Together they create a situation where the stock has multiple chances to rerate rather than one binary moment.

The Valuation Gap In Context

The sector multiple compression and expansion argument deserves more specificity than “Momentus is cheap relative to comps.”

When SpaceX lists at $1.75–2 trillion, every analyst covering the space sector has to remodel their comps. Rocket Lab at 90x sales looked aggressive six months ago. Against a $2 trillion SpaceX, it starts to look like a reasonable reference point for the only credible small launch competitor in the US. AST SpaceMobile at 500x sales is the most speculative end of the spectrum, direct-to-cell satellite internet competing with Starlink’s established 10 million customer base.

Momentus at 7.5x forward sales sits at the opposite end of that spectrum. It’s not a pre-revenue moonshot. It’s an operationally active company with confirmed government contracts, live missions, and a specific, defensible niche in the space economy that the dominant player (SpaceX) has no interest in competing with. The orbital last-mile delivery market is not a market SpaceX wants. It’s too small, too operationally complex relative to the volume of launches, and too focused on individual customer needs. That structural non-competition is what makes the SpaceX highway analogy durable. Momentus isn’t fighting SpaceX, it’s completing SpaceX’s service offering.

If the sector rerating pushes Momentus to even 20x forward sales, which is still a fraction of Rocket Lab’s multiple, that implies a market cap of $200M against today’s $75M. A 30x multiple implies $300M. Neither of those numbers requires anything heroic relative to where comparable operational space companies trade today. They just require the market to notice the company exists.

The Main Risks I see

Any honest treatment of this thesis requires equal time on the risks. There are several that are genuine and material.

The $10M revenue guide is management’s forecast, not a delivered result. If Vigoride 7 mission reports disappoint, if payload customers don’t follow through on commitments, or if the Space Force task orders don’t materialize on the expected timeline, there is limited fundamental support for the stock at current prices. A guidance cut on a micro-cap at peak sector excitement is a brutal combination.

Liquidity is a real risk on the way in and out. A $75M market cap with thin daily volume means meaningful position sizes move the stock. Getting in is easy when momentum is building. Getting out when the cycle turns is considerably harder. The bid-ask spread widens exactly when you most want to sell.

The SpaceX IPO euphoria is time-limited. The historical IPO cycle is consistent — lead-up excitement, peak around IPO day, cool-down as insiders begin selling in the staggered lockup structure SpaceX has designed. Ecosystem stocks follow that cycle with amplification. Momentus will likely see its sharpest move in the June 4–12 window and then face meaningful selling pressure as the broader SpaceX excitement normalizes. Investors who chase the peak rather than positioning ahead of it face asymmetric risk on the downside.

The defense thesis is real but early. DARPA and Space Force contracts validate the technology. They don’t guarantee task order wins in Golden Dome or Shield. Those programs involve competitive processes, classified evaluations, and timelines that management cannot fully control. Projecting 50–100% upside from a Shield win requires that win to actually happen.

Finally, this is a company that generated $1.1 million in revenue last year. The 9x growth guide is compelling on a rate-of-change basis. On an absolute basis, $10 million in revenue for a publicly traded company with operational complexity, facility costs, and mission expenses is still a very thin margin for error. Any operational setback hits disproportionately hard at this scale.

Position sizing is the single most important risk management tool here. This is not a core holding. It is a high-conviction asymmetric bet in a specific time window, sized to survive a 50% drawdown without material portfolio damage and sized to matter if the 200–400% scenario plays out.

The Thesis In Two Paragraphs

High risk. Asymmetric setup. Do your own work.

Disclaimer: This is editorial content for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any security. Do your own due diligence before making any investment decision. All figures sourced from company earnings releases and public filings.

Earnings recaps. Price targets. Analyst upgrades. Twitter threads that age like milk.

None of it makes you a better investor.

What actually moves the needle is doing the work nobody else wants to do. Reading the footnotes. Understanding why a stock is down. Separating the temporary from the permanent. And having the conviction to act when everyone else has already moved on.

That is the only edge that has ever consistently worked. And it is the only edge I try to build here.

I publish one deep dive at a time — no filler, no hot takes, no recycled consensus views. Just original research, written clearly, grounded in the numbers, with my own money on the line every single time.

I do not cover 50 stocks. I cover the ones I would actually buy.

Which brings me to this week.

While the S&P 500 ground out a modest +0.79%, my portfolio returned +7.04%. That is a 625 basis points beat in five trading days. That didn’t come from getting lucky on one position. It came from multiple separate holdings each moving 10–25% on distinct, researched catalysts. A UK clinical milestone in oncology software, back-to-back NATO defense contracts, a Q1 earnings beat in a seasonally weak quarter, and a defense stock reclaiming its highs after a technically-driven selloff that had nothing to do with the underlying business.

Full breakdown of every position, every driver, and what I’m watching moving forward.

Read more

Earnings recaps. Price targets. Analyst upgrades. Twitter threads that age like milk.

None of it makes you a better investor.

What actually moves the needle is doing the work nobody else wants to do. Reading the footnotes. Understanding why a stock is down. Separating the temporary from the permanent. And having the conviction to act when everyone else has already moved on.

That is the only edge that has ever consistently worked. And it is the only edge I try to build here.

I publish one deep dive at a time — no filler, no hot takes, no recycled consensus views. Just original research, written clearly, grounded in the numbers, with my own money on the line every single time.

I do not cover 50 stocks. I cover the ones I would actually buy.

Which brings me to this week.

I have spent the last several weeks inside the filings of a company most investors have already written off. The stock is down 65% from its highs. The headlines are bad. The sentiment is worse. Most people I speak to will not even look at it anymore.

I think that is exactly the wrong reaction.

This is a business that has grown net sales for 29 consecutive quarters. That has taken nearly 1,000 basis points of market share over seven years. That now owns one of the fastest-growing brands in its category — currently sitting in less than 20% of its largest retail partner’s global store count.

The slowdown that caused the selloff was real. The cause of that slowdown is not what the market thinks it is. And the fix is already being deployed.

My base case implies more than 100% upside from current levels. Not in ten years. On numbers that are two to three years out.

Below I walk through every assumption, every risk, and every number behind that conclusion.

The stock is…

e.l.f. Beauty (NYSE: ELF) — Initiating Coverage at $54

A Category-Defining Brand Trading at a Trough Multiple

The One-Paragraph Thesis

e.l.f. Beauty ( ) is one of the best-run consumer brands in the world, with 29 consecutive quarters of net sales growth, 920 basis points of US market share gained over seven years, and a portfolio that now includes four brands each doing over $200M in annual retail sales. The stock is down 65% from its highs because organic growth decelerated, mainly caused by a self-inflicted price increase that crushed unit volumes, not by any structural loss of consumer demand. The market is pricing in permanent impairment. I think it is temporary. At $54 the stock trades at roughly the same EBITDA multiple as slow-growth peers like Ulta and Coty despite growing revenue 3-4x faster and generating higher margins. My base case implies 85%+ upside on FY2028 numbers alone without needing the multiple to rerate anywhere near historical levels.

Table of Contents

1. Business Overview

2. The Brand Portfolio

3. What Happened & Why The Stock Is Down 65%

4. Why I Think The Market Is Wrong

5. The Financials

6. Valuation

7. Risks

8. Conclusion

1. Business Overview

e.l.f. Beauty is a multi-brand beauty company built on a simple but powerful idea: deliver premium quality products at extraordinary prices. Founded over 20 years ago with products starting at $1, ELF has evolved into one of the most culturally relevant beauty companies in the world. They are the number one purchased brand among Gen Z, Gen Alpha and Millennials simultaneously.

The business model has three interlocking pillars that management calls their competitive moat:

Value proposition. ELF products are priced at a dramatic discount to prestige equivalents (typically 70–80% cheaper) while matching or exceeding quality. The Power Grip Primer at $11 vs a prestige equivalent at $38. The Halo Glow Skin Tint at $14 vs prestige at $46. This is not a low-quality budget brand. It is a deliberate democratization of premium beauty.

Powerhouse innovation. ELF has a community-led innovation model — listening to what consumers want, identifying the prestige product that has gone viral, and rapidly bringing their own version to market at a fraction of the price. In 2025 they held 4 of the top 10 new products in all of mass cosmetics according to Nielsen. 6 of the top 10 the year before.

Disruptive marketing engine. ELF was digitally native before digital native was a concept. They built their brand on TikTok, YouTube, Instagram and influencer marketing before legacy competitors understood these channels. The results speak for themselves: unaided brand awareness has grown from 13% in 2020 to 45% in 2025.

The company operates as a single reportable segment with five brands: e.l.f. Cosmetics, e.l.f. SKIN, Rhode, Naturium and Well People. In FY2026 (fiscal year ended March 31, 2026) the company generated $1,636M in net sales, up 25% year-over-year, with $328M in adjusted EBITDA at a 20% margin.

Their four largest customers where Target (18% of sales), Walmart (13%), Amazon (11%) and Sephora (10%). These four customers accounted for just over half of revenue, with the remaining 45% spread across international retailers, drug stores, specialty retail and direct-to-consumer e-commerce.

2. The Brand Portfolio

Understanding the four brands is essential to understanding both the current situation and the opportunity.

e.l.f. Cosmetics - The Core Engine

The flagship brand generated approximately $1.1B in retail sales in FY2026 and holds 13% national market share in US mass color cosmetics, but 21% at Target, their longest-standing retail partner. That gap between 13% nationally and 21% at their most mature retailer is the single most important number in this thesis. It tells you the growth ceiling is far higher than the national share implies.

ELF Cosmetics gained 115 basis points of US market share in FY2026, which was the largest gain among nearly 1,000 cosmetics brands tracked by Nielsen. This was their 29th consecutive quarter of market share gains. Over seven years they have gained 920 basis points total. For context, the next highest brand gained 240 basis points over the same period.

e.l.f. SKIN

Generated approximately $200M in global retail sales in FY2026. Has risen from the #25 mass skincare brand in the US five years ago to #11 today, with approximately 2% share vs the category leader at 13%. Earlier in the growth curve than ELF Cosmetics with a longer runway ahead.

Rhode - The Crown Jewel Acquisition

Acquired in August 2025 for $897.5M. This is the most important strategic decision ELF has made and it has already exceeded every expectation.

Rhode generated approximately $390M in proforma net sales in FY2026, growing over 80% year-over-year. It is the #1 beauty brand at Sephora North America. This was achieved less than two years after launching in Sephora. It is currently in less than 20% of Sephora’s global stores.

In September 2026, Rhode launches with Sephora across 19 European countries simultaneously — a launch Sephora is described as going “all in” on based on North American performance. This is not in management’s FY2027 guidance. It is pure upside.

The brand was founded by Hailey Bieber and has 74% of its social followers outside the US. There is an extraordinary international demand signal that the retail footprint has barely begun to monetize.

Naturium — Quiet Outperformer

Acquired in October 2023 for $333M. Generated approximately $250M in global retail sales in FY2026, which is double its pre-acquisition levels in under three years. In Q4 FY2026 it was the fastest growing brand among the top 50 skincare brands. Management recently put awareness-building advertising behind Naturium for the first time and saw immediate acceleration. This brand is earlier in its marketing investment cycle than ELF Cosmetics, meaning the playbook is still being deployed.

3. What Happened — Why The Stock Is Down 65%

Understanding exactly why the stock fell is essential to evaluating whether the bear case is correct.

The Cause: A Self-Inflicted Wound

In August 2025, ELF raised prices across all SKUs by $1.00 — responding to tariffs that had reached as high as 170% on Chinese imports and general inflationary pressure. The price increase was rational given the cost environment. The consequence was not.

Tarang Amin, ELF’s CEO, said it plainly on the Q4 FY2026 earnings call:

“We saw a pretty big fall off on units from where before we took pricing to after we took pricing.”

ELF’s value proposition “best-in-class products at extraordinary prices” is the core of their brand identity. When they raised prices, they moved away from that identity, even if modestly. Consumers who buy ELF because it is the best value in mass beauty are highly price-sensitive. A $1 increase on a $9 product is an 11% price increase. On a $13 product it is nearly 8%.

The result: dollar sales held up because of the price increase, but unit volumes declined, which is the real measure of consumer demand. Market share in dollars continued to grow (+115bps in FY2026) but the underlying unit trend was concerning.

The Organic Growth Reality

ELF brand global consumption growth moderated from high single digits in FY2026 to low single digits in the last 12 weeks before the Q4 call. Organic revenue growth guided at just 4–5% for FY2027 with Q1 expected down high single digits.

This is what the market is pricing. And at face value it looks bad for a stock that historically traded at 30–60x earnings on the promise of 20%+ growth.

What The Market Is Missing

The market is treating a pricing-induced unit decline as a structural demand problem. The evidence suggests otherwise.

When management cut the price of the Halo Glow Skin Tint from $18 to $14, they saw:

• 38% unit lift on Amazon

• 36% lift across all retailers

• Triple-digit lift on TikTok Shop

That is not what structural demand destruction looks like. That is a price-elastic consumer who was temporarily priced out and came back immediately when value was restored.

4. Why I Think The Market Is Wrong

Management has already begun the remediation with four concrete actions:

Pricing: Selectively reducing prices on specific product families where unit velocity has fallen. The Halo Glow test is the template. More reductions coming across multiple families.

Innovation: Fall FY2027 innovation goes to market approximately one month from the Q4 call date. Additional fast-tracked innovation not originally planned for FY2027 being brought forward to before the holidays. Two of the top ten spring innovation launches in mass cosmetics already belong to ELF.

International: UK and Germany showing green shoots as they exited Q4 FY2026. Dedicated marketing activation underway. New markets being seeded selectively.

Leadership: New CMO, new President of ELF Brands, new Chief Technology and AI Officer — all appointed to sharpen execution on the core brand.

Critically, management said the pricing and innovation actions are not in guidance. They represent upside to the 4–5% organic growth baseline.

Rhode Changes The Equation

Even if the ELF brand takes two or three more quarters to recover fully, Rhode transforms the financial profile of the company. On a proforma basis Rhode contributed ~24% of total net sales in FY2026 at an 80%+ growth rate. The Sephora EU launch across 19 countries in September 2026, (which btw is not included in management’s base case guidance) could add hundreds of millions of dollars of incremental revenue over the next two years.

Rhode is in less than 20% of Sephora’s global stores. The brand’s social following is 74% international. The demand exists. The infrastructure is being built. The only question is the pace of execution.

The International Opportunity Is Structurally Underappreciated

International represents just 21% of ELF’s net sales. Legacy beauty peers derive 70%+ of their sales internationally. The gap is not because international consumers do not want ELF products. 50% of ELF Cosmetics social followers are outside the US. The gap exists because ELF is early in building international retail infrastructure.

In FY2026 international grew 38% vs US growth of 21.5%. In Q4 FY2026 international grew 75%. The trajectory is clear. Management launched in eight international retail customers across 14 countries in FY2026 alone.

The Multiple Does Not Reflect The Business

ELF currently trades at approximately 10-13x NTM EBITDA. Ulta trades at 8–10x growing at ~3%. Coty trades at 9–11x growing at ~4%. Inter Parfums trades at 14–17x growing at ~6%.

ELF is guided to grow revenue at 12–14% in FY2027 with 21% EBITDA margins and accelerating international exposure. The multiple does not reflect that growth profile by any reasonable comparable analysis.

5. The Financials

Income Statement Build

Revenue. I project total net sales of approximately $1,873M in FY2027 and $2,139M in FY2028, representing growth of ~14.5% and ~14.2% respectively. US growth of 9% annually reflects organic recovery as pricing actions restore unit velocity, combined with continued shelf space expansion at Walmart and Ulta. International growth of 35% in FY2027 and 30% in FY2028 reflects Rhode Sephora EU, continued ELF/Naturium international expansion, and market share recovery in UK and Germany.

Gross margin. I project 70.8% in FY2027 expanding to 71.0% in FY2028 and holding there. Tariff rate has declined from ~55% in FY2026 to ~35% currently, providing a tailwind. Partially offset by Rhode mix shift toward retail channels which carry lower gross margins than DTC.

SG&A. The detailed build:

The most important line is Other Operating Costs — inflated to 15.1% in FY2026 by the $57.6M Rhode earnout fair value adjustment which is non-cash and non-recurring in nature. It also consisted of legal efforts surrounding import tariff mitigation. Stripping out the earnout fair value, the underlying Other Operating Costs were approximately 11.6% in FY2026. My FY2027 assumption of 10.2% reflects clean run-rate costs as Rhode integration matures and ERP implementation costs wind down post-SAP go-live.

Adjusted EBITDA margin. I project 21.1% in FY2027, essentially in line with management’s guided ~21% and expanding to 22.6% in FY2028 as revenue scale drives leverage across compensation and distribution. This compares to ELF’s historical peak of ~23% in FY2024. By FY2029 and FY2030 I project margins of 23.5% and 24.4% respectively as international scale builds and the brand portfolio matures.

EPS. My base case adj. diluted EPS:

FY2027 EPS of $3.29 is essentially at the midpoint of management’s guided $3.27–3.32 — a deliberate calibration. The growth acceleration in FY2028 reflects Rhode lapping its acquisition contribution and becoming fully organic, combined with margin expansion and operating leverage. They have taken on more debt in relation to their new Rhode acquisition and I am estimating their effective interest rate to be around 5.5% to 6%, this is why I am not projecting as much of an improvement in eps as in their EBITDA.

Balance Sheet and Capital Allocation

ELF ended FY2026 with $290M cash and $842M gross debt — net debt of approximately $552M, or just under 1.7x adj. EBITDA. Debt consists primarily of the $600M term loan drawn to finance the Rhode acquisition, maturing March 2030, with mandatory amortization of $30M per year.

The balance sheet is manageable. Interest expense runs approximately $47–48M annually at a blended rate of ~5.5–6.0% — well within the company’s cash generation capacity. Free cash flow after capex was approximately $215M in FY2026 and should grow to $250–280M in FY2027 as EBITDA expands.

Capital priorities per management: brand growth investment first, technology/AI second, opportunistic buybacks third. A $400M buyback authorization remains outstanding. At $54 the stock is materially below management’s own assessment of intrinsic value — they repurchased $50M in FY2026 and have signaled continued opportunistic buying.

6. Valuation

Why Adjusted Metrics

GAAP net income in FY2026 was just $26M — not because the business is struggling, but because of $57.6M in non-cash Rhode earnout charges, $79M of acquisition-related D&A, and $87M of stock-based compensation. None of these reflect underlying cash earnings power. Adjusted metrics are the right basis — and conveniently, management guides to them explicitly every quarter.

Methodology 1 — EV/NTM Adj. EBITDA (Weight: 40%)

My primary metric. The most widely used by institutional investors covering ELF and the most reliable anchor given management guides to it directly.

Lower end: 13x. Slow-growth peers like Ulta and Coty trade at 8–11x growing revenue at 3–4%. Even in a scenario where organic growth stays stuck at low single digits, ELF is still growing total revenue at double digits with 20%+ EBITDA margins. 13x prices in significant pessimism while acknowledging this is a fundamentally better business than its slow-growth peers.

Higher end: 18x. Inter Parfums, which is the closest growth comparable, currently trades at 14–17x growing revenue at ~6%. ELF at my base case grows revenue at 13–14% with higher margins and a larger international runway. A modest premium to Inter Parfums is well-justified by the growth differential.

This gives me a FY2027 price range of $76–$108 and a FY2028 single estimate of $110 using a 15x multiple.

Methodology 2 — P/NTM Adj. EPS (Weight: 30%)

Management guides adj. EPS directly at $3.27–3.32 for FY2027. My estimate of $3.29 sits at the midpoint.

Lower end: 23x. A growth company that has recently disappointed on organic delivery deserves a discounted multiple. 23x reflects scepticism without pricing in collapse, which is consistent with what the market assigns to companies growing EPS at 15-20% that have recently missed expectations.

Higher end: 30x. ELF is compounding adjusted EPS at approximately 20%+ through my projection period. A 30x multiple on 20%+ EPS growth implies a PEG ratio below 1.5x, and is undemanding for a brand-driven consumer compounder with genuine international optionality. It is also under what ELF have historically traded at. Inter Parfums trades at 22–26x at half ELF’s projected EPS growth rate. A modest premium for double the growth is entirely reasonable.

This gives me a FY2027 price range of $76–$99 and a FY2028 single estimate of $107 at 25x.

Methodology 3 — EV/NTM Revenue (Weight: 20%)

An independent cross-check. Useful here because near-term earnings are suppressed by deliberate investment spending, the revenue multiple captures the top-line growth story without being distorted by margin timing.

Lower end: 2.8x. This is the multiple you assign if you believe nothing improves from here. No international expansion premium, no Rhode optionality, no margin pathway. Even then it implies meaningful upside.

Higher end: 3.5x. A business growing revenue at 13–14% with expanding margins and a brand in less than 20% of its largest retail partner’s global store count deserves a modest premium to the peer group average of 2.5–3x. 3.5x is conservative recognition of that optionality.

This gives me a FY2027 price range of $78–$99 and a FY2028 single estimate of $103 at 3.2x.

Methodology 4 - DCF (Weight: 10%)

WACC of 8.5%, terminal growth rate of 2.5%, projecting through FY2030 on base case unlevered free cash flows.

DCF implied intrinsic value: ~$95/share.

I weight this at only 10% given the sensitivity to terminal assumptions, a 1% move in WACC shifts the output by almost $20 per share. It is most useful as directional confirmation. And it confirms the same message as every other methodology.

Summary

Blending the four methodologies by weight, my fair value range across every methodology lands between $76 and $115, with a blended base case of $78 on the lower end and $106 on the higher end.

FY2027 blended target: $78–$90, implying +47–70% upside from $52.90

FY2028 blended target: $106, implying +100% upside from $52.90

Every methodology points to the same conclusion using multiples that remain well below what this business has historically traded at and what its growth profile deserves.

This does not require heroic assumptions. It requires the business to keep executing on the trajectory it has maintained for seven consecutive years, and the market to stop pricing in permanent impairment from a temporary, self-inflicted stumble.

I am long ELF at $52.90 with a 12–24 month price target of $78–107, implying 47–100% upside from current levels.

7. Risks

I do not want to paper over the risks. Here is where I could be wrong.

Risk 1 — Organic Growth Does Not Recover

This is the central risk. If the ELF brand’s unit volume decline is structural rather than price-induced. If the brand is simply maturing faster than management believes, then the 4–5% organic growth guided for FY2027 could prove optimistic, and the re-acceleration I am projecting for FY2028+ may not materialize.

My view: The price-cut test data (38% unit lift on a single SKU) is the clearest evidence I have that demand is intact. But one data point is not a trend. The Q1 FY2027 earnings report (due August 2026) will be the first real test. If organic consumption does not show sequential improvement, my thesis is weakened.

Risk 2 — Rhode Execution Risk

Rhode’s business model is heavily dependent on Hailey Bieber. The 10-K explicitly flags this. Her departure or reduced involvement would be a material risk to the brand. Additionally the Sephora EU launch, while exciting, is a complex multi-country simultaneous rollout. Execution risk is real.

My view: Hailey Bieber is a co-founder and significant equity holder through the earnout structure. Her financial incentives are aligned with performance through at least 2028. The EU launch risk is real but Sephora has gone all-in based on North American results, suggesting their own confidence in the brand.

Risk 3 — Tariff Re-escalation

ELF manufactures approximately 55% of products in China as of FY2026 (down from 99% three years ago). If tariffs re-escalate toward 100%+ levels seen in early 2025, gross margin would face significant pressure. The Supreme Court ruling invalidating IEEPA tariffs introduced some relief but the tariff environment remains uncertain.

My view: ELF has demonstrated they can diversify supply chain, from 1% to 45% ex-China production in three years. They also have $58.5M of IEEPA tariff refunds potentially incoming which management would deploy into consumer value and marketing. The risk is real but the company has shown they can adapt.

Risk 4 — Marketing Efficiency Deteriorating

Advertising has gone from 10.5% to 22.1% of net sales over four years with no sign of leverage. If ELF needs to keep spending more to maintain relevance (the classic consumer brand treadmill), then margin expansion will not materialize.

My view: Management guided marketing at 23–25% for FY2027, which is essentially flat to FY2026’s 24%. The argument for leverage over time is that brand awareness at 45% nationally should eventually require less spending to maintain than it did to build. But this has not shown up in the data yet. I model modest advertising leverage from FY2028 onward — if this does not materialize, my margin projections are too optimistic.

Risk 5 — Balance Sheet Constraint

$842M of gross debt with $30M mandatory annual amortization means ELF is not capital-light. If revenue growth disappoints and EBITDA misses, leverage metrics deteriorate quickly. At <2x net debt/EBITDA today the balance sheet is manageable — but there is limited room for error.

My view: FCF generation of $250M+ annually more than covers mandatory debt service. The risk is if a new large acquisition is pursued before Rhode is fully integrated and the debt is partially repaid.

8. Conclusion

e.l.f. Beauty is not a broken business. It is a great business going through a self-inflicted difficult period that the market is treating as permanent impairment.

The facts: 29 consecutive quarters of growth. 920 basis points of market share gained. Four brands each above $200M in retail sales. The number one brand at Sephora North America, is only in less than 20% of global Sephora doors. A stock down 65% from its highs trading at the same multiple as companies growing at a fraction of its pace.

The fix is not complicated: restore value to consumers through selective price reductions, execute on fall innovation, let Rhode’s international expansion do its work. Management has done this before. They have the track record.

My base case on FY2028 numbers implies a stock price of approximately $101–111 depending on the methodology, representing 90-110% upside from $52.90.

I initiated my position at $54. I am prepared to be patient. I think the next 12–18 months will determine whether organic growth recovers. If it does, the multiple will rerate and the upside is substantial. If it does not, I believe the downside is limited at current prices.

The risk/reward at $52.9 is one of the most compelling I have found in consumer equities in some time.

I am long $ELF.

This is not financial advice. Do your own research. I hold a long position in ELF at the time of writing. All financial data sourced from ELF SEC filings, earnings releases and earnings call transcripts.

SpaceX filed for its IPO at the worst possible time for new investors. The company's revenue crawled 15% to $4.69 billion in Q1 while capital expenditures exploded to $10.1 billion, which is more than double the prior year. Most of that capex went toward AI projects that aren't working. This is textbook peak-valuation timing.

The Numbers

Revenue growth decelerated from 33% in 2025 to just 15% in the first quarter of 2026. Meanwhile, SpaceX burned through cash at an unprecedented rate, posting a $4.28 billion net loss in Q1 alone after losing $4.94 billion for all of 2025.

The capital expenditure breakdown reveals where the money's going: $7.7 billion of the $10.1 billion quarterly capex went to AI initiatives. That's 76% of total capex flowing into xAI, the artificial intelligence unit Musk merged with SpaceX in February at a $1.25 trillion valuation.

By April, Musk admitted xAI "was not built right first time around" and needed to be "rebuilt from the foundations up." Grok, xAI's answer to ChatGPT, remains a niche player that drew lawsuits for enabling deepfake explicit content of women and children.

Why Now Makes Sense for SpaceX

The timing isn't entirely reckless from SpaceX's perspective. Public markets remain hungry for high-profile tech offerings, and retail investors still associate the SpaceX brand with success. The company can raise massive capital while Musk's reputation stays relatively intact.

Going public also provides an exit for early employees and investors who've waited over two decades for liquidity. SpaceX stock has been locked up in private markets, and many shareholders want to cash out regardless of optimal timing.

The IPO gives SpaceX ammunition to compete in the AI arms race. OpenAI, Anthropic, and Google are spending billions monthly on compute and talent. SpaceX needs public market capital to keep pace, even if that dilutes the core space business.

The Downside Risks

Public investors will subsidize AI experiments while getting minimal voting control. Musk's 85% voting power through dual-class shares means shareholders fund his vision without meaningful input.

The company faces regulatory scrutiny it's never experienced. Public SpaceX will report quarterly results, face SEC oversight, and answer to analysts who might question why a rocket company is burning billions on chatbots.

Most critically, SpaceX trades its flexibility for quarterly earnings pressure. Private SpaceX could take risks and make long-term bets. Public SpaceX will face questions every 90 days about AI losses while revenue growth slows.

Market Implications Run Deep

A successful SpaceX IPO validates the entire space economy and will lift every public space stock. Defense contractors, satellite operators, and space infrastructure companies should see immediate gains as institutional investors add space exposure to portfolios.

Companies like Redwire Corp ($RDW) could benefit significantly. Redwire provides space infrastructure and manufacturing services that become more valuable as the overall space economy expands. A massive SpaceX public offering legitimizes space investments and drives capital toward smaller players with actual growth stories.

If SpaceX trades poorly post-IPO, it'll crater confidence in space stocks across the board. Investors will question whether any space company deserves premium valuations if SpaceX can't maintain them.

Classic Pre-Peak Indicators

Companies go public when insiders want to cash out at maximum valuations, not when the business fundamentals look strongest. SpaceX checks every box:

Slowing core business growth. The space launch and Starlink satellite internet operations that built SpaceX's reputation grew just 15% year-over-year.

Massive capital allocation to unproven ventures. Three-quarters of capex flows into AI projects that Musk himself acknowledges aren't working.

Desperate expansion moves. SpaceX agreed to potentially pay $60 billion for coding startup Cursor or face a $10 billion breakup fee. For context, that's more than half of SpaceX's entire 2025 revenue of $18.67 billion.

Complex deal structures that benefit insiders. Musk controls 85% of voting power through his dual-class shareholding structure, ensuring new public investors get minimal say in company direction.

SpaceX built its reputation launching rockets and operating satellites. Now it's pivoting resources toward competing with OpenAI, Anthropic, and Google in artificial intelligence. This is markets where it has little to no competitive advantage.

The Anthropic deal illustrates the problem. SpaceX will collect $1.25 billion monthly through May 2029 from Anthropic for compute capacity at the Colossus 1 data center in Memphis. But this makes SpaceX a glorified cloud provider, not an AI innovator. The company's essentially renting out server space while burning billions trying to build Grok into a ChatGPT competitor.

Meanwhile, OpenAI appears twice in SpaceX's filing as a "key competitor" in AI. SpaceX is entering a market where established players already have massive head starts and proven products.

Valuation Timing

The February merger with xAI created a combined entity valued at $1.25 trillion. Goldman Sachs leads the underwriting syndicate, followed by Morgan Stanley, Bank of America, Citigroup, and JPMorgan Chase. This level of Wall Street muscle signals expectations for a massive offering that extracts maximum proceeds from public investors.

SpaceX plans to list on Nasdaq under ticker $SPCX, with roadshow marketing beginning June 8. The timing coincides with slowing growth in the core business and accelerating losses from AI investments.

SpaceX's prospectus describes Starbase, Texas as the "gateway to Mars" where "humanity's future in space unfolds in plain view." This marketing language obscures the financial reality: a company burning cash on AI projects while core space operations decelerate.

The company employs roughly 500 people in Starbase, which became an official company town in 2025 with SpaceX employee Bobby Peden as mayor. OSHA is currently investigating a worker death at the facility from May 15.

President Gwynne Shotwell received $85.8 million in 2025 compensation, mostly from options awards. Her base salary was $1.08 million. These compensation levels reflect a company preparing for a liquidity event, not one focused on operational efficiency.

What New Investors Actually Buy

Public investors will purchase shares in a rocket company that's morphing into an AI venture with unproven technology and massive cash burn. They'll get minimal voting rights while Musk maintains 85% control through dual-class shares.

The core SpaceX business — launching rockets and operating Starlink — generated solid returns. But that predictable revenue stream now subsidizes speculative AI investments that even Musk admits require complete rebuilds.

SpaceX will likely price its IPO at peak valuations while growth slows and costs explode. New investors will fund AI experiments rather than participate in the successful space business they think they're buying.

The IPO represents a classic bait-and-switch: market a proven aerospace company while actually selling shares in an AI startup that's burning billions with little to show for it. Goldman Sachs and the other underwriters will collect their fees regardless of how public investors fare afterward.

This is not financial advice. Do your own research before making any investment decisions. These are analytical frameworks and observations and not buy or sell recommendations.

That kind of disconnect between a business and its stock price is either a trap or an opportunity. I spent several weeks figuring out which one. I bought at $89.

Here’s everything I found

What ServiceNow Actually Is

Most people have a rough idea of what ServiceNow does. IT ticketing. Workflow software. That kind of thing. That description is technically accurate, but practically useless for understanding the business.

Let me give you the more honest version.

A Fortune 500 company runs on thousands of interconnected processes spanning every department. When an IT system breaks, someone has to route that issue to the right team, track it through resolution, update every affected system, manage the compliance trail, and document it for audit. When a new employee joins, HR has to onboard them, IT has to provision access across thirty different systems, legal has to get signatures, finance has to set up payroll. When a security vulnerability surfaces, the security team has to find every affected asset across the entire technology estate, prioritize it by business risk, route the remediation to the right people, verify it got fixed, and report it to the board.

All of that flows through ServiceNow. It sits at the operational center of the enterprise and coordinates work across every department, every system, every geography, every process. It’s the connective tissue. When McDermott says ServiceNow is the ERP for IT, he’s underselling it — it’s closer to the operating system for the entire business.

The numbers that prove this aren’t the revenue figures. They’re the retention numbers. 97% renewal rate. A 2011 customer cohort that has grown their initial contract value by 228 times over fourteen years. 85% of the Fortune 500 as customers. Once ServiceNow is embedded in how a company runs, removing it is not a software decision — it’s a multi-year organizational transformation that would cost more than the software ever did.

McDermott put a number on this in a podcast a month before Q1 earnings. He said it costs 10 times more to replicate even a simple ServiceNow application using a language model when you factor in the GPU infrastructure, human capital costs, and token fees. That’s not a marketing claim. That’s arithmetic. And it’s why the SaaS apocalypse thesis, while correct for parts of the software industry, is pointed directly at the wrong target here.

The Quarter That Confused Everyone

ServiceNow reported Q1 2026 on April 22nd. By every objective measure it was an excellent quarter. Subscription revenue grew 19% in constant currency, beating the high end of guidance. cRPO — the forward revenue indicator that matters most to institutional investors — grew 21% in constant currency, a full 100 basis points above guidance. Non-GAAP operating margin came in at 32%, 50 basis points above guidance. Free cash flow margin was 44%. They beat every single metric they guided to.

The stock dropped 12% after hours…

The reason is worth understanding precisely, because it explains both why the market is skeptical and why I think that skepticism is somewhat misapplied.

When they raised the full-year guide, the increase came entirely from the Armis acquisition contribution, approximately 125 basis points of subscription revenue growth. The organic full-year guide didn’t move. Investors looking for evidence of underlying AI-driven acceleration were looking at a business that beat Q1 but didn’t take that beat and translate it into a higher organic guide.

I understand why it disappointed, but I think the interpretation, that the core business is decelerating, misses what the leading indicators are telling you.

cRPO of $12.64 billion growing at 21% constant currency represents contracted future revenue. It covers 82% of their entire full-year revenue guide before they’ve sold a single new deal this year. The backlog of $27.7 billion is nearly two times annual revenue. McDermott said in a podcast the week before earnings that only 11% of enterprises he’d spoken with had moved past the pilot phase, this pipeline of customers still in AI experimentation represents a demand wave that has barely begun to hit the revenue line.

When the 89% who are still experimenting start deploying at scale, they’ll need exactly what ServiceNow provides. The control tower. The governance layer. The execution infrastructure. That’s not a thesis about future innovation. It’s a thesis about present-day backlog converting into future revenue.

The AI Story Is Just Getting Started

Here’s where the investment thesis gets interesting, and where I think the market is materially underestimating what’s coming.

ServiceNow has a product suite called Now Assist. This is AI capabilities built natively into every commercial tier of every product, priced at an incremental rate on top of the base subscription. Think of it as an AI layer sitting on top of everything they already sell. When an employee submits a ticket, Now Assist automatically summarizes it, suggests the resolution, and in many cases resolves it before a human ever touches it. When a developer needs to build a workflow, Now Assist generates the code. When a security analyst needs to understand a vulnerability, Now Assist maps the business impact across the CMDB.

The adoption trajectory is remarkable. At the end of FY2024, Now Assist ACV was approximately $100 million, which was early commercial stage, mostly pilots. By Q4 FY2025 it crossed $600 million. By Q1 2026 it was tracking above $750 million! On the Q1 earnings call, McDermott raised the full-year FY2026 target from $1 billion to $1.5 billion. At Financial Analyst Day on May 4th, he confirmed $30 billion+ in subscription revenue by 2030, with 30% of total ACV coming from AI at that point.

The mechanism that gets them there is what McDermott calls the hockey stick. Customers buy Now Assist entitlements — a pack of AI tokens. They start using them. They run out. They buy more. That reload cycle is usage-based revenue sitting on top of the fixed subscription model, and it’s just beginning to show up in the numbers. One example from the Q4 2025 call: a leading American fast food chain renewed with 13 times more assist tokens than their original purchase because their agents were consuming them at scale. That’s not a one-off customer story. That’s the shape of what the consumption model looks like as adoption matures.

50% of ServiceNow’s net new business in Q1 2026 was already from non-seat pricing. Tokens, connectors, infrastructure. That number was close to zero twelve months earlier. You’re watching a pricing model transformation happen in real time, and the revenue implications compound significantly as usage scales.

Why AI Makes ServiceNow More Valuable, Not Less

I want to spend real time on this because it’s the core of why I think the market is wrong, and it requires engaging with the bear case seriously rather than dismissing it.

The disruption argument goes like this: large language models are getting so capable that enterprises will spin up custom AI agents to handle the work that workflow platforms currently manage. Why pay ServiceNow’s subscription fees when a well-prompted Claude or GPT-4 can route tickets, draft responses, and close cases autonomously?

This argument is correct as a description of what AI can do. It’s wrong about what enterprises will actually do with it.

Here’s the thing the disruption thesis misses: AI agents need governance infrastructure to operate safely in an enterprise. They need to know what they’re allowed to access. They need audit trails. They need compliance controls. They need rollback capabilities when something goes wrong. They need someone (or something) to be accountable when an agent makes a decision that costs the company money or exposes it to regulatory risk.

That governance infrastructure is exactly what ServiceNow provides. Not as a product feature bolted onto the side, but as the foundational operating model of the platform.

Jensen Huang made this point at ServiceNow’s Knowledge conference on May 5th. I’ll paraphrase rather than put words in his mouth, but the core of what he said was this: every AI agent needs the same identity management, access control, compliance tracking, and governance that human workers need. The difference is there will be far more agents than humans. The computation required for agentic AI versus simple generative AI has increased by a factor of ten in two years. More agents, more compute, more coordination complexity — all of it flows through ServiceNow’s control tower.

This is not a company being disintermediated by AI. It’s a company that becomes more essential as AI proliferates. Every agent that gets deployed inside an enterprise is a new managed entity that needs to be governed. Every autonomous workflow is a process that needs to be audited. Every AI model that touches sensitive data needs permission management that’s updated in real time as roles and contexts change.

ServiceNow is not losing to AI. ServiceNow is the infrastructure that makes enterprise AI possible at scale.

The Acquisitions Make Sense When You See the Whole Picture

Three acquisitions in rapid succession — Moveworks, Veza, Armis — triggered real concern from institutional investors. It looked like panic buying. Or worse, revenue plugging. Both interpretations are wrong.

Moveworks is the AI conversational layer. Employees type in plain English — “I can’t access Slack, fix it” — and the system figures out what they need and resolves it without a human touching it. ServiceNow had the workflow execution engine but no natural language front door. Moveworks gave them that. They integrated it into the platform in 20 days. The business it became — EmployeeWorks — closed more deals in Q1 2026 alone than Moveworks did in all of FY2025. That’s not integration risk. That’s distribution leverage working immediately.

Veza solves the identity governance problem for the agentic enterprise. Traditional identity management tools were built for human accounts. They weren’t designed to handle hundreds of AI agents making real-time decisions across dozens of systems, each needing different permissions depending on context, role, and the specific task being executed. Veza maps access relationships across humans, machines, and AI agents using a patented access graph, updated continuously. As enterprises scale AI agent deployments, the identity governance problem becomes the bottleneck. Veza removes it.

Armis sees everything that Veza governs. It discovers every connected asset across the enterprise in real time — not just servers and laptops, but IoT devices, operational technology on factory floors, medical equipment in hospitals, industrial controllers, building management systems, and shadow IT that procurement never approved. It does this without installing agents on the devices, which is the critical differentiator for environments where you can’t put software on every piece of equipment. Nine out of ten Fortune 10 companies already use Armis for this reason.

When you put all three together — Armis sees everything, Veza controls who can access it, ServiceNow executes the governance and remediation — you have the only complete AI governance stack available in enterprise software. No competitor has all three layers. Most competitors don’t have any of them.

McDermott’s description: Armis is our Instagram. The growth hasn’t started yet. The cross-sell motion into the existing ServiceNow base — 8,800+ customers, 85%+ of the Fortune 500 — has barely begun. Armis already has 40% Fortune 100 penetration independently. ServiceNow gives them a global enterprise sales force. That’s the unlock.

The Financial Model

Let me give you the actual numbers behind my thesis rather than vague projections.

My base case starts with FY2026 non-GAAP revenue of $15.9 billion. That’s slightly above management’s guide of $15.74-15.78 billion, reflecting a modest FX tailwind and my expectation that they’ll beat the high end of guidance as they’ve done consistently for years. Revenue growth runs at approximately 20% in FY2026, accelerating to 20.4% in FY2027, 21.4% in FY2028, and 21.9% in FY2029 as Now Assist consumption scales and the Armis cross-sell ramps. By FY2029 I have revenue at $28.3 billion, on track toward the $30 billion+ FY2030 commitment.

Non-GAAP gross margins recover from the current compressed level — 79.5% in FY2026 reflecting inference cost pressures — to 80.7% by FY2029 as NowLLM substitutes expensive frontier API calls for routine tasks and hyperscaler costs normalize.

Non-GAAP EBIT margins expand from 31.5% in FY2026 — matching management guidance exactly — to 33.3%, 35.3%, and 36.8% through FY2029. The driver is AI-driven internal efficiency. ServiceNow deployed their own AI agents internally and resolved 90% of employee IT requests autonomously. They’re growing revenue above 20% annually while holding headcount roughly flat. That’s the operating leverage, and it compounds.

EPS grows from $4.31 in FY2026 to $5.38, $6.83, and $8.58 through FY2029.

Valuation and Price Targets

Here’s where I land on value, and I’ll be explicit about the multiple logic rather than just throwing out a number.

ServiceNow at $95 is trading at approximately 22x non-GAAP NTM earnings. That’s near the lowest multiple this company has traded at in years. For context: Salesforce trades at 27x despite growing at 9% annually. Workday trades at 25x despite growing at 15%. ServiceNow is growing subscription revenue at 20%+ at $15+ billion in scale — a combination almost no enterprise software company has ever achieved — and it’s trading at a discount to slower-growing peers.

That’s the mispricing in one sentence.

My P/E multiples on the base case start at 40x for FY2026 — a recovery toward fair value from today’s depressed levels — declining to 35x, 28x, and 23x through FY2029 as the earnings base grows and growth naturally moderates. Even with that compression, the implied price targets are $172 at the low end of FY2026 P/E and $258 at the high end of FY2029 P/E.

On EV/EBIT the picture is similarly compelling. My FY2026 EBIT of approximately $5B at 35-45x gives you an EV of $175-225 billion. Add the net cash position of approximately $2.8 billion and divide by roughly 1,000 million diluted shares — you’re looking at implied prices of $178 to $228 in the base case for FY2026 alone, rising substantially by FY2029 as EBIT approaches $10 billion.

My fair value today — discounting the FY2027 base case at a rate that reflects the remaining uncertainty — is $145 to $155. The stock is at $95. That’s roughly 55-65% upside to what I think this business is worth right now, before any AI acceleration above the base case.

In two to three years, I believe this stock is worth at least $200. In the bull case — where Now Assist tracks toward $2.5 billion rather than $1.5 billion in FY2026 and the Armis cross-sell materializes faster than I’m modeling — $250+ by 2028 is not an unreasonable destination.

The SBC Problem Is Real and You Deserve to Know About It

I want to be genuinely transparent about something the non-GAAP numbers obscure, because hiding it would be intellectually dishonest.

ServiceNow pays approximately 15% of revenue in stock-based compensation annually. That’s roughly $2.3 billion in FY2026. It’s real money going to employees in the form of equity rather than cash, and it dilutes shareholders. The company issues shares to employees and then spends billions buying those shares back in the open market just to keep the diluted share count from rising materially. In Q1 2026 they spent $2.225 billion on buybacks while generating $1.665 billion in free cash flow. They spent more on anti-dilution buybacks than they generated in free cash flow.

On a true economic earnings basis — stripping SBC back out the way the company strips it out in non-GAAP — the EPS in FY2026 is closer to $2.60 than $4.31. The stock on that basis is trading at roughly 37x economic earnings rather than 22x non-GAAP earnings. That’s more expensive.

I’m not dismissing this. I’m acknowledging it and explaining why I still own the stock.

First, management has committed to getting SBC to single digits as a percentage of revenue. My model has it declining from 13.5% in FY2026 to 9.6% by FY2029 — slightly more aggressive than the base case but not wildly so given the trajectory. As that happens, the gap between non-GAAP EPS and economic EPS narrows dramatically. By FY2029 economic EPS reaches $6.41 on my numbers, converging toward the $8.58 non-GAAP figure. That convergence is itself an earnings growth story that doesn’t require any revenue acceleration — it’s pure multiple expansion from paying employees less in equity relative to the size of the business.

Second, the SBC burden is a feature of the growth phase, not a permanent state. ServiceNow competes for AI talent against OpenAI, Anthropic, Google, and Microsoft. Of course equity compensation is elevated right now. As the business matures and the competitive talent market normalizes somewhat, SBC as a percentage of revenue will decline organically even without deliberate action.

Third, ’m not buying ServiceNow on economic earnings today. I’m buying it on the trajectory of economic earnings over three to five years. Even on the SBC-adjusted basis, EPS is growing at 35-40% annually as revenue scales and the SBC burden shrinks proportionally. That’s an extremely compelling compounding story if you’re willing to hold through the noise.

The Bear Case

I try to take the other side seriously. Here’s where the bear case has genuine teeth.

Gross margin compression could be worse than modeled. AI inference costs are growing with every Now Assist deployment, and Jensen Huang himself noted at Knowledge that agentic AI requires ten times more computation than simple generative AI. If Now Assist adoption accelerates faster than the cost deflation in inference — which historically tracks the semiconductor cost curve — gross margins stay compressed longer than I’m projecting. Every 100 basis points of gross margin below my model translates to roughly $160 million in lost EBIT at current scale.

The Armis integration could disappoint. Three acquisitions in quick succession is operationally challenging. Moveworks integrated cleanly in 20 days, which is impressive. Armis is a much larger and more complex business — a cybersecurity company with a fundamentally different customer relationship and sales motion than ServiceNow’s core platform. If the integration takes longer, if the cross-sell doesn’t materialize as quickly as the deal thesis assumes, the revenue contribution is delayed and the margin headwinds persist longer.

Microsoft is not standing still. Copilot is embedded across the Microsoft 365 suite that every enterprise already uses. As Copilot gets more capable at workflow orchestration — routing approvals, managing IT tasks, handling HR inquiries — the case for a separate ServiceNow deployment weakens at the margin, particularly for smaller enterprises that don’t need the full governance infrastructure. ServiceNow’s answer is that enterprise-grade governance and individual productivity are different categories, and that’s true today. How true it remains in 2027 is a legitimate open question.

And the AI consumption hockey stick might take longer than I’m modeling. The $1.5 billion Now Assist ACV target for FY2026 represents roughly 150% year-over-year growth. That’s achievable — the trajectory clearly supports it — but it requires the reload cycle to kick in meaningfully across a large base of customers who are still in early adoption. If customers run their trials and don’t reload at the rates management is projecting, the AI revenue story slips by twelve to eighteen months.

In a genuine bear scenario — slower Now Assist growth, gross margin compression to the 78% range, multiples that stay depressed at 22-25x — the stock probably stays in the $75-100 range for the next year before the underlying business drags it higher. That’s the downside I’m sizing. It’s not catastrophic. It’s roughly flat from where I bought, which is a very different risk profile from typical high-multiple growth investments where a multiple compression can cut the stock in half.

What I’m Watching Most Closely

Two numbers will tell me if my thesis is on track or needs revision.

The first is cRPO growth in Q2 and Q3 2026, in constant currency. Management guided 19.5% for Q2. If it comes in at 21%+ — which Q1’s 100 basis point beat suggests is possible — the re-rating begins as the market recognizes the forward demand picture is strengthening, not weakening. If it comes in at 18% or below, the deceleration narrative gets louder and I’ll have to decide whether to add or reassess.

The second is Now Assist ACV at the end of FY2026. Management guided $1.5 billion. If the actual number is $1.7 billion or above, it validates the hockey stick and raises the FY2030 trajectory. If it comes in at $1.1-1.2 billion, the AI monetization thesis is real but slower than I’m modeling, and I’ll need to revise the outer year assumptions.

Everything else — gross margins, SBC, Armis revenue — matters but doesn’t change the core thesis. The core thesis is that ServiceNow is the enterprise AI governance infrastructure layer, that demand for that layer will grow as AI agents proliferate, and that the market is currently pricing the stock as if ServiceNow is a standard SaaS company facing disruption rather than the platform that makes enterprise disruption governable.

The Setup

Let me be direct about why I think this is a buying opportunity right now, at $95, rather than six months ago at $120 or six months from now at whatever the market decides.

The multiple compression that happened this year was driven by fear, not fundamentals. The SaaS disruption narrative hit the entire sector. The Armis acquisition raised legitimate questions about capital allocation. The organic guide not moving after a Q1 beat disappointed investors who wanted confirmation of acceleration.

But the underlying business — 97% renewal rates, $27.7 billion in backlog, $750 million in AI ACV and growing, the only complete AI governance stack in enterprise software, 8,800 customers with expansion potential across Armis, Veza, and now EmployeeWorks — has never been in better shape.

At 22x non-GAAP NTM earnings, the stock is trading below Salesforce and Workday despite growing faster and compounding at higher margins. That gap closes one of two ways: ServiceNow slows down, or the market re-rates it. I’m betting on the latter.

My fair value is $145-155 today. My price target in two to three years is $200+. I bought at $89. I’m comfortable holding through the noise, watching the Q2 cRPO print, and adding if the thesis continues to play out.

This is not a short-term trade. It’s a position in a company I think is going to be one of the defining enterprise technology platforms of the next decade — the layer through which AI enters the enterprise safely, at scale, with accountability.

The market is mispricing that. For now.

Full disclosure: I own shares in ServiceNow ($NOW), purchased at approximately $89. This is my personal investment analysis and reflects my own views and assumptions. It is not financial advice. My projections could be wrong. The risks described are real. Please do your own research before making any investment decisions.

I think that framing is both right and lazy.

Right, because the threat is real for a lot of names. Lazy, because it treats every software company as equally exposed — and they’re not. The difference between a business that gets disrupted and one that gets stronger through this transition usually comes down to three things: what data they sit on, what infrastructure they own, and whether switching away from them is a rational decision for the customer.

When I run that filter, most SaaS doesn’t pass. But some does. These are the five names I keep coming back to — companies where I think AI is a tailwind, not a threat, and where I find the structural logic genuinely compelling rather than just narratively convenient.

1. ServiceNow

The standard pitch on ServiceNow is “IT workflow software.” That framing is accurate and almost completely useless for understanding why this company is interesting right now.

What ServiceNow actually has is the CMDB — a Configuration Management Database that maps how an enterprise’s entire IT infrastructure is built. Every server, every application, every integration, every dependency. It’s essentially a living blueprint of how a company’s technology actually works underneath the surface. Most enterprises have nothing else like it.

That matters enormously in an AI agent world. When companies start deploying agents at scale — and they will — someone has to govern what those agents can access, audit what they did, and enforce boundaries when they go somewhere they shouldn’t. That governance layer needs to understand the infrastructure it’s governing. ServiceNow already does. Building that from scratch, with the data fidelity they have, would take years.

They’re calling the strategy an AI control tower. I think that’s the right framing. The idea is that ServiceNow becomes the place where AI agent activity across the enterprise gets logged, reviewed, and controlled. It’s a high-value position and a defensible one — not because of the product features, but because of the underlying data.

The validation points are worth noting. Anthropic uses ServiceNow. NVIDIA uses ServiceNow. When AI companies are using your software to manage their own operations, that’s a signal worth paying attention to.

The risk I take seriously is category collapse. The borders that defined SaaS for twenty years are dissolving fast. Salesforce is pushing into ITSM, which is ServiceNow’s home turf. ServiceNow is pushing into CRM, which is Salesforce’s. Microsoft is moving into agents everywhere. Everyone is in everyone else’s business simultaneously. ServiceNow is well positioned but it’s not playing from safety. It’s competing in an increasingly open field where the rules are being rewritten in real time.

That said — if I had to pick one SaaS name structurally built for the agent era, this is probably it.

Listen to what NVIDIA’s CEO, Jensen, has to say about ServiceNow:

2. Sea Limited

Sea gets filed under “emerging market e-commerce” by most people and then ignored. That categorization misses what’s actually going on.

They run three distinct businesses: Shopee, which is e-commerce across seven Southeast Asian countries and Brazil; SeaMoney, which handles lending, payments, and e-wallets across those same markets; and Garena, which is gaming — led by Free Fire, one of the most popular mobile shooters in the world, plus a publishing arm.

Each business has its own logic. Together they form something harder to disrupt than any of them individually.

Start with Shopee. The AI threat to e-commerce is usually framed around agents — if your shopping journey begins inside an AI interface rather than opening an app, you get more price comparison, more optionality, less platform loyalty. That’s a real dynamic and I don’t want to dismiss it entirely. But the threat assumes there’s somewhere better to go. Shopee owns the logistics and fulfillment network in markets where logistics is genuinely hard. They have the seller relationships, the consumer payment data, and the infrastructure to get things to people’s doors. An AI agent helping someone shop in Southeast Asia is still, in most cases, going to end up transacting on Shopee. There’s no obvious alternative with the same geographic reach and physical infrastructure.

SeaMoney is the quieter story. AI improves credit risk pricing, which directly improves the quality of their lending book. It improves lead scoring, which improves conversion for financial product distribution. These aren’t headline-grabbing applications but they compound into real margin improvement over time.

And then there’s Garena. This is where I think Sea has an upside vector that’s genuinely underappreciated.

Gaming is one of the most natural applications for generative AI — AI opponents that are indistinguishable from human players, dynamically generated maps and worlds, faster content production cycles.

Free Fire has a massive installed base across markets where mobile gaming is dominant. The ability to keep that base engaged through AI-enhanced content is a meaningful tailwind for a business that lives and dies on engagement.

The geography is under-followed. The AI case is underappreciated. The infrastructure moat is real. It’s one of the more interesting setups I see right now.

3. MercadoLibre

The structural thesis here is nearly identical to Sea, which is either reassuring or boring depending on how you look at it. Different geography, Latin America, but same fundamental logic.

The AI risk people raise is the same agent-driven price comparison argument — if consumers start their shopping journey inside an AI tool, they become more price-sensitive and less platform-loyal. I think that’s worth monitoring. It’s not something I’d ignore. But the answer is the same as it is for Sea: if you need something delivered in Latin America, the logistics infrastructure matters. MercadoPago is already embedded in wallets at scale. The switching cost isn’t the app or the interface. It’s everything underneath that’s been built over two decades and can’t be replicated quickly.

AI makes MercadoLibre better in the ways it makes most fintech better, by sharper credit risk assessment, better product recommendations, improved targeting. None of that is the core thesis, but it adds up.

Sea has Garena as an additional AI upside vector that MercadoLibre simply doesn’t have. Gaming is a natural fit for generative AI in a way that pure e-commerce and fintech aren’t. So I’d give Sea a slight edge on total AI upside. But on the downside protection side, I would argue that based on the infrastructure moat, the physical layer and the financial services embeddedness, that they’re essentially equivalent.

If you believe in the Sea thesis, it’s hard to not be interested in MercadoLibre. The logic is the same. Latin America is just the geography.

4. Constellation Software $CAD

This one is less intuitive than the others and I think that’s exactly why it’s worth paying attention to.

Constellation Software is a conglomerate of vertical market software businesses. Not large markets. Deliberately small ones. Think marina management software. Cemetery software. Specific permitting systems for municipal governments. Court management tools for regional jurisdictions. Each one is a near-monopoly in a niche that most serious competitors wouldn’t bother looking at twice.

The AI disruption thesis collapses here almost immediately when you work through the actual economics. For an AI company to threaten one of Constellation’s portfolio businesses, they’d first have to decide the addressable market is worth the effort. It usually isn’t, these are tiny markets by design. Then they’d have to build distribution into industries where finding and closing customers requires deep relationships and domain expertise that takes years to develop. Then they’d have to convince a customer to switch away from software that is mission-critical, if it goes down, they cannot collect revenue, and that represents maybe one to two percent of their total cost base. The risk-reward of switching is completely irrational. You save a little money and you risk your whole operation. Nobody who runs a real business makes that trade.

That switching calculus is the actual moat. It’s not brand loyalty or network effects. It’s asymmetric risk for the customer. And AI doesn’t change that math. If anything, AI makes it slightly cheaper and faster to build and maintain niche software, which could accelerate Constellation’s acquisition pipeline and widen the universe of businesses they can buy and improve.

Constellation acquires, optimizes, holds forever, and reinvests. That model has worked for thirty years across economic cycles. I don’t see a version of the AI transition that breaks it. It’s the least exciting name on this list, and probably the most structurally secure.

5. AMD (Not SaaS)

AMD isn’t SaaS. I’m including it because the AI case is too interesting to leave out and because it illustrates something important about how to think about the infrastructure layer of this whole transition.

The CPU story first. AMD spent a long time losing to Intel. Then TSMC became the world’s best chip foundry. AMD focused purely on design, stopped trying to compete in manufacturing, and eventually produced something better. Their EPYC data center CPU now outperforms Intel on the metrics that matter most to hyperscalers. More virtual machines per server, more containers, lower total cost of ownership. Meta and OpenAI have both struck direct deals with AMD. That’s not a minor data point. Every dollar flowing into AI compute infrastructure is a tailwind for a company making one of the best data center CPUs in the world.

But the more interesting angle is the GPU story, and specifically the software layer underneath it.

NVIDIA’s real moat isn’t the hardware. It’s CUDA. The software framework that developers use to write GPU code. A massive ecosystem of models, tooling, and developer workflows has been built on CUDA over fifteen years. Switching from NVIDIA to AMD’s Instinct GPUs means porting that code to AMD’s equivalent framework, ROCm. That porting process is friction. Real, painful friction that keeps developers locked into NVIDIA even when AMD’s hardware is competitive.

Here’s what I keep thinking about: AI is exceptionally good at code translation. As these tools get better, the friction of converting CUDA to ROCm decreases. Not to zero, there are real technical challenges, but meaningfully. If that switching cost erodes over time, AMD’s GPU business becomes a more serious competitive threat to NVIDIA than the current market pricing implies. It’s a slow-moving dynamic, not a near-term catalyst. But it’s the kind of second-order effect that tends to matter a lot once it becomes visible.

The risk on AMD (and on every semiconductor name right now) is cyclicality. The bull case requires believing that AI compute demand is so structurally large and so durable that the normal cycle doesn’t apply. Maybe. But every time in history that capital has flooded into a high-return industry, supply has eventually overshot demand. The players spending aggressively today are all doing so because they’re terrified of undershooting. When everyone makes that bet simultaneously, overshoot becomes almost mathematically inevitable at some point. The new trough could be significantly higher than the old one. But cycles don’t stop being cycles because the numbers got bigger. That’s worth keeping in your peripheral vision even when the demand environment looks strong.

This is not financial advice. Do your own research before making any investment decisions. These are analytical observations and a framework for thinking — not buy or sell recommendations.

If you missed Palantir at $8. Missed Nvidia before the AI wave. Missed Micron before the memory supercycle. You know the feeling of watching a name run 10x while you were waiting for a “better entry.” Well, there’s a stock sitting in the space economy right now that almost no one is talking about, and the setup looks strikingly familiar.

I’m going to lay out the full case. You can be the judge.

The company is Redwire. Ticker:

Redwire (NYSE: RDW) is a small-cap space infrastructure company that most investors have never heard of. It’s not a rocket company. It’s not a satellite operator. It builds the actual hardware that goes inside spacecraft, the components every space mission depends on but nobody talks about. It currently trades at a price that, in my view, hasn’t caught up to the business at all. That’s the opportunity.

What this company actually does

When a satellite goes to orbit, someone has to build the solar arrays that power it, the structures that unfold in space, the navigation systems that keep it pointed correctly, and the docking mechanisms that allow it to rendezvous with other craft. More often than most people realize, that someone is this company.

They have hardware flying on the International Space Station. Their imaging and sun sensor technology was aboard NASA’s Artemis 2 — the first crewed Artemis mission. And through a recent acquisition, they own one of the most widely deployed small drone platforms in the U.S. military, now flying on over 250 aircraft across the Marines and Navy.

“This is the only public company building the actual hardware that goes inside spacecraft across the entire industry. And it hasn't run 500% yet."

The catalyst that changes everything

SpaceX is targeting a public offering in the coming months at a valuation of up to $2–2.5 trillion. When that event lands — or even as the anticipation reaches a fever pitch — institutional and retail capital is going to flood into space-economy names. And here is the problem investors are about to run into: there are almost no pure-play public space infrastructure companies.

You’ve got a launch company. A satellite-to-phone company. An Earth observation company. And then you’ve got this one — the only public company building the actual components that go inside the spacecraft across all of those missions and many more. When that search begins, there aren’t many places for the money to go.

A moat that took decades to build

The numbers behind the story

Revenue grew 58% year-over-year. That alone would make for a compelling setup. But the margin story is what elevates this from interesting to genuinely significant: gross margin expanded from 14.7% to 26.6% — nearly 12 percentage points — in a single year. That’s not a one-time quirk. It reflects a business where demand is outpacing costs and pricing power is real.

The backlog tells the forward-looking story. At $498 million — the highest in company history — it jumped 21% in a single quarter. The book-to-bill ratio of 1.92 means that for every dollar of work delivered, they signed nearly $2 of new contracts. Most defense primes are pleased with anything above 1.0. Hitting nearly 2.0 is exceptional by any measure.

Is a 10x actually realistic?

A genuine 10-bagger requires two things working simultaneously: a strong underlying business and a rerating catalyst. Most stocks only ever get one. This one has a credible argument for both — at the same time.

The business case is built on 58% revenue growth, a near-doubling of gross margins, and $498 million of locked-in backlog in an industry where contracts are sticky and competition is structurally limited. That combination creates the kind of operating leverage that compounds hard over time.

The catalyst case is a defined, time-bound event — the SpaceX IPO — that is likely to trigger a broad rerating of space-economy equities. This company is the clearest pure-play on space infrastructure among publicly traded names, and it has not yet benefited from the sector enthusiasm the way peers have.

The risks are real. This is a small-cap that burns cash, carries acquisition-related complexity, and depends on government program cycles that can shift. The 10x outcome requires sustained execution and market recognition — neither guaranteed. But few setups combine record backlog momentum, margin inflection, a near-impossible moat, and an identifiable sector catalyst all at once.

At current prices, the gap between what this business is worth and what the market is paying for it looks wide. When that gap closes, it tends to close fast.

The thesis in one paragraph

This is editorial content for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any security. Do your own due diligence before making any investment decision. All figures sourced from company earnings releases and public filings.

Not complicated. Not uncertain. Terrible. The kind of terrible where you mention the stock at dinner and someone laughs. Where the comment section under every earnings article is full of people explaining, with complete confidence, why you’re wrong. Where the chart looks like a tombstone and the analysts have stopped covering it.

That moment. That exact moment. Is usually where the money is.

AMD went through it. So did Micron. So did Nvidia before that, and Palantir before that, and Tesla before that. Every single one of these stocks had a stretch where owning them was embarrassing. Where the thesis felt lonely. Where the patient investor sat in silence while everyone else laughed.

And then the numbers showed up. And the laughing stopped. And the same people who called it a trash stock started calling it a generational opportunity, right as the easy money was already gone.

This is not a coincidence. It’s a pattern. And patterns, once you understand them clearly enough, become repeatable.

Here’s how I think about finding the next one.

What “Hated” Actually Looks Like

The word gets thrown around loosely, so it’s worth being precise.

A hated stock is not just a stock that’s down. Plenty of stocks decline for legitimate reasons. Some examples being: deteriorating fundamentals, structural disruption, management that’s actively destroying value. Those stocks deserve to be down. Buying them because they’re cheap is how investors lose money for years while waiting for a recovery that never comes.

A genuinely hated stock, the kind that becomes a great opportunity, has a specific profile. The business is sound. The fundamentals are either stable or quietly improving. But the stock has been dead money long enough that the crowd has stopped paying attention, and the narrative has curdled from “disappointing” to “broken.”

AMD fit this profile exactly. For years, the chart went nowhere while Nvidia pulled away in the GPU race. Every quarter, someone wrote the same article: AMD is losing. Nvidia is eating their lunch. Why would you own this? The stock became a punchline. “Advanced Money Destroyer.” “Advanced Money Decimator.” The jokes were endless, and they all pointed in the same direction, away from the stock.

But underneath the narrative, something else was happening. The architecture was improving. The data center business was building. The margin profile was moving in the right direction. The earnings inflection that would eventually reprice the stock was already in motion, it just hadn’t shown up loudly enough yet for the crowd to notice.

That gap between what the fundamentals are doing and what the narrative is saying is where the opportunity lives.

The Three Questions I Ask About Any Hated Stock

Not every hated stock becomes a love stock. The ones that don’t usually fail one of three tests. I’ve found that working through these three questions honestly and without letting the prevailing narrative color the answers and filters out most of the traps.

1. Is the balance sheet strong enough to survive the wait?

Timing is the one variable nobody gets right consistently. You might be right about the thesis and wrong about when the market recognizes it. That gap could be six months. It could be two years. A company with a clean balance sheet, manageable debt, and enough cash runway can survive that wait. A company that needs to raise capital while the stock is depressed gets diluted, and the thesis gets a lot harder.

This is one reason Revolve ( ) stands out right now. Roughly $335 million in cash, essentially zero debt, market cap of $1.2 billion. They don’t need the market to cooperate anytime soon. They can operate, grow, and wait. That’s a very different risk profile than a hated stock burning through cash while praying for a re-rating.

2. Is the margin trajectory moving in the right direction?

This is the single most underrated indicator in a hated stock. The crowd focuses on revenue and if it is it growing fast enough, is it beating estimates, is the top line exciting. But margins tell you something more important: is the business getting more or less efficient over time?

With AMD, the margin story was the signal that most people missed. Net income margins were on a trajectory that, extrapolated forward, implied an earnings power that the stock price wasn’t remotely pricing in. The market was looking at where AMD had been. The margins were pointing to where AMD was going.

When you’re looking at a hated stock, pull up five years of gross margin, operating margin, and net income margin. If they’re expanding, even slowly, even inconsistently, then that’s the business getting better while the narrative says it’s getting worse. That divergence is what you’re looking for.

3. Is there a catalyst that doesn’t require a miracle?

The biggest trap in hated stocks is needing everything to go right simultaneously. A turnaround that depends on a new product launching perfectly, a new CEO executing flawlessly, a new market opening up exactly on schedule is not a thesis, that’s a lottery ticket.

The best hated stock opportunities have a catalyst that’s already in motion. With AMD, it was the data center revenue ramp, and it was already happening, already showing up in the numbers, just not yet in the multiple. With Micron, it was the memory cycle turning. A well-documented, historically reliable pattern that was already past its trough. With Nike ( ) , the catalyst is simpler: a wholesale re-engagement that’s already underway under new management, in a brand that never actually lost its customer.

You’re not looking for a miracle. You’re looking for something that’s already started, that the market hasn’t priced in yet because the narrative is still pointing backward.

Hated for Good Reason vs. Hated for Cyclical Reasons

This distinction is everything.

Some stocks are hated because something is broken. A business model that doesn’t work at scale, a product that’s losing relevance, a balance sheet that’s slowly becoming a catastrophe. These stocks deserve every bit of skepticism they get. The danger is that they look identical from the outside to the ones that are hated for cyclical reasons.

Here’s a rough way to separate them.

A stock hated for cyclical reasons tends to have: stable or improving unit economics, a customer base that still shows up, a competitive position that hasn’t structurally eroded, and a management team that’s playing defense competently while waiting for conditions to improve. The problem is external factors like sector rotation, macro headwinds, interest rate pressure and general risk-off sentiment.

A stock hated for good reasons tends to have: deteriorating unit economics, a customer base that’s actually shrinking, competitive share loss that’s structural rather than temporary, and a management team that’s either in denial or actively making things worse. The problem is internal and no amount of macro tailwind fixes an internally broken business.

AMD, hated for cyclical reasons: the GPU market temporarily favored Nvidia’s architecture, but the underlying semiconductor demand cycle was turning. The business was never broken. It was just out of favor during a period when the competition had a temporary advantage.

Contrast that with a retailer that’s losing customers to e-commerce, has a cost structure it can’t reduce fast enough, and is in a category that’s genuinely commoditizing. Down 60%, cheap on a trailing earnings basis, and a value trap for anyone who buys the discount without reading the trajectory.

Current Candidates Running the Playbook

The list of stocks currently sitting where AMD sat a year ago is longer than most people realize.

SoFi Technologies ( ) is building one of the more interesting financial platforms of the next decade. Member growth is compounding. The balance sheet is strengthening. But the stock is down nearly 40% year-to-date with the broader market at all-time highs, because fintech is out and semiconductors are in. When sentiment rotates, SoFi won’t move slowly. It went from $6 to $30 in months during its last run.

ServiceNow ( ) and Salesforce ( ) are two of the best enterprise software businesses ever built. High switching costs, expanding margins, deeply embedded in their customers’ operations. Both are down significantly from their peaks. The market is currently pricing them as if AI is a threat to their business model. The alternative reading, that AI turns their platforms into productivity multipliers that justify massive expansion, isn’t priced in at all.

ELF Beauty ( ) is doing everything right in cosmetics. Growing market share, strong brand loyalty, international expansion barely started. Down 70% from its high because consumer is the wrong sector to be in right now.

None of these are guaranteed. That’s not the point. The point is that the same analytical framework that identified AMD as an opportunity a year ago; strong balance sheet, improving margins, catalyst already in motion, hated for cyclical rather than structural reasons, all applies cleanly to all of them.

The Most Important Part Nobody Talks About

Everything I’ve described so far is analytical. Findable. Teachable.

The part that’s genuinely hard, is the part that separates the investors who actually execute this playbook from the ones who understand it intellectually but never pull the trigger and is sitting with the discomfort.

When you buy a hated stock, you are, by definition, early. The narrative is still negative. The chart still looks bad. The people around you are still skeptical. There will be quarters where the stock goes down further. There will be articles confirming every bear case point. There will be moments where you genuinely wonder if you missed something.

That discomfort is not a signal to sell. It’s the price of admission.

AMD shareholders sat with that discomfort for years. So did Micron shareholders. So did Palantir shareholders who bought at $6 and held through every skeptical article about government contracts and Peter Thiel and profitability timelines.

The market rewards patience with compound interest. It rewards conviction with outsized returns. But it only pays those rewards to the investors willing to be early, which means willing to be uncomfortable, willing to be wrong in the short term, and willing to hold a thesis that nobody else believes yet.

The crowd that laughed at AMD a year ago is the same crowd ignoring five consumer names right now.

The question worth asking yourself is which side of that trade you want to be on.

This is not financial advice. Do your own research before making any investment decisions. These are analytical frameworks and observations and not buy or sell recommendations.

And then there’s the version most investors are actually living in.

Nike at $42. ELF Beauty down 70% from its peak. SoFi sitting at half its all-time high while posting record member growth. ServiceNow, one of the best software businesses ever built, down 55% in a single year.

Here’s the thing about that second version: it’s not a sign that these companies are broken. It’s a sign that all the attention, all the money, all the momentum, and all the headlines are crowded into one corner of the market. And history is pretty clear on what happens next.

I want to walk through five consumer names I’m watching closely right now. Not because they’re momentum plays. Because the math is quietly becoming very compelling, and the crowd hasn’t noticed yet.

ELF Beauty (The Best Growth Story in Cosmetics )

All-time high: $210+. Today: roughly $55.

ELF has done something most consumer brands never pull off. It built a premium perception at a mass-market price point, then grew market share consistently against giants like L’Oréal and Estée Lauder. Their unit economics are strong. Their brand loyalty with Gen Z is real. Their international expansion has barely started.

None of that matters to the market right now, because the market wants chips.

Down more than 70% from its peak, ELF is trading as if the growth story is over. But if you look at the actual fundamentals you will se that nothing structurally changed. The revenue trajectory, margins and brand strength are all as strong as ever. What’s changed is the category. Consumer is out. AI is in. That’s a sentiment shift, not a business shift.

A year from now, if ELF is trading anywhere between $125 and $175, people will talk about what an obvious opportunity it was. It always works that way in hindsight.

Nike - A 23,000% All-Time Gainer in a 5-Year Downtrend

Nike’s all-time high is around $170+. It’s trading near $42.

Think about that for a second. One of the most recognizable brands on earth. A company that has compounded wealth for shareholders for decades. A business with global distribution, pricing power, and a customer base that spans every income level and every continent.

And it’s in the middle of one of the longest downtrends of its modern history.

The bear case on Nike is real. They over-indexed on DTC, pulled back from wholesale partners, and are now dealing with a messy rebuilding process under new leadership. Those are legitimate concerns. But Nike has navigated harder transitions than this.

What typically happens with stocks like Nike is that Wall Street writes the story off entirely, and then one quarter, the numbers start to turn. A small comp improvement. A margin recovery. A wholesale re-engagement that surprises to the upside. And suddenly the stock re-rates from “value trap” to “turnaround in progress,” and you miss the first 40% move while you’re waiting for confirmation.

I’m not waiting for confirmation. At $42, with the brand still intact and new management actively rebuilding the playbook, the risk/reward has shifted.

Celsius Holdings

Celsius peaked near $100. It’s around $30 today.

The energy drink market is a duopoly. Monster and Red Bull have owned it for two decades. Celsius is the only brand that has genuinely threatened that structure in recent memory, and they’ve done it by going after a different customer, with a different positioning, at a different shelf location.

Now layer in what’s happened in the last 12 months. The Alani Nu acquisition brings one of the fastest-growing women’s wellness brands into the portfolio. The Rockstar distribution partnership opens convenience store access that Celsius couldn’t reach alone. And international is still a fraction of what Monster does globally, which means the ceiling here is far higher than the stock price implies.

Down 70% from its high, you’re buying this at a point where the market has essentially written off the entire growth narrative. I think that’s wrong. The distribution infrastructure being built right now is the kind of thing that shows up in revenue two or three years from now, not this quarter.

SoFi Technologies - When It Moves, It Moves Fast

SoFi’s all-time high is around $32. It’s trading near $15.
Down 40% year-to-date, with the broader market at all-time highs.

SoFi is becoming a real financial institution. Not a fintech story, but an actual bank, with a growing balance sheet, expanding product set, and member growth that continues to compound quarter over quarter. The business they’re building over the next decade looks genuinely significant.

But none of that matters in a market that wants semiconductors.

Here’s what I keep coming back to with SoFi: when this stock moves, it moves fast. It went from $6 to $30 in what felt like a matter of months during its last run. Not because the business changed overnight, but because sentiment flipped and the buyers who had been sitting out suddenly rushed in.

It will happen again. I don’t know if the run starts in three months or twelve. But when it does, the move from $15 to $45 will happen quickly. When that happens everyone will be talking about SoFi again right as the easy money is gone.

I’d rather own it now, while it’s boring.

Revolve

Revolve has a market cap of roughly $1.2 billion.
They’re sitting on approximately $335 million in cash and cash equivalents. They carry essentially zero debt.

Do the math. You’re paying about $865 million for the operating business itself. For a growing, profitable fashion e-commerce brand with a loyal customer base, strong influencer distribution, and an international runway that’s barely been explored.

The stock is at $18. It gets almost no attention. When people talk about consumer discretionary ideas, Revolve rarely comes up.

That’s usually a feature, not a bug.

The balance sheet alone creates a floor that most people aren’t accounting for. And when consumer sentiment eventually rotates back, and when the same dollars that have been chasing Nvidia and AMD start looking for beaten-down names with real businesses underneath them, a company trading at a significant discount to the sum of its parts with a clean balance sheet tends to move sharply.

The market is currently telling you that consumer is dead, fintech is dead, and the only things worth owning are chips and AI infrastructure. That may be true for the next six months. It won’t be true for the next three years.

These five companies have real businesses, real customers, and in most cases, real earnings. They’re not hated because something is fundamentally broken. They’re hated because they’re in the wrong sector at the wrong moment.

That’s usually exactly when you want to be paying attention.

That probably sounds like arrogance. It’s not. It’s pattern recognition. And it’s one of the most underrated skills in equity research.

The first 10 minutes aren’t about building a model. They’re not about reading the 10-K cover to cover. They’re about asking the right questions in the right order so you don’t waste time on companies that will never make it into your portfolio.

Let me walk you through exactly what I do.

The first 2 minutes: What does this company actually do?

This sounds obvious. It isn’t.

I don’t read the investor relations description. I don’t read the CEO’s letter. Those are written by people paid to make the business sound exciting.

I go straight to the revenue breakdown. How does this company make money? Segment by segment, product by product. A business that can’t be summarized in two sentences of pure economics, with no marketing language, is either genuinely complex (rare) or deliberately obscuring something (common).

A company that says it’s a “platform” but generates 60% of revenue from one customer and 80% from one-time professional services is not a platform. That realization just saved you four hours.

Minute 2-4: Is the business getting better or worse?

Pull up five years of revenue, gross margin, and operating margin. That’s it. Don’t dive deeper yet.

I’m not running a DCF or any other valuations. I’m looking at the direction and consistency of the narrative.

A great business tends to show: revenue growth, expanding or stable gross margins, and operating leverage over time (operating margin improving faster than revenue grows). A business that can’t show any of these over five years needs an extraordinary explanation before it earns more of my attention.

Pay specific attention to gross margin trends. They tell you more than almost any other metric:

• Expanding gross margins = pricing power, mix shift toward better products, or scale benefits. All good.

• Contracting gross margins = competition, commoditization, or a business model that doesn’t scale. All bad.

• Volatile gross margins = often a sign of poor pricing discipline or a business exposed to input cost swings.

This step alone filters out probably half the companies I look at.

Minute 4-6: Where does the cash actually go?

The income statement is a story. The cash flow statement is the truth.

I pull up the cash flow statement and look for the operating cash flow and the capex, to get the firms free cash flow. I then see if this is consistent over the last three to five years.

What I’m asking is: “does this business generate real cash, or does it consume it?”

A company can show positive net income while burning cash through working capital changes, aggressive capitalization of costs, or enormous capex requirements. I’ve seen businesses with 15% net income margins and negative free cash flow for a decade. That’s not a business. That’s a capital destruction machine dressed up in GAAP clothes.

The ratio I care the most about in this phase is FCF as a percentage of net income. Above 100% consistently = high-quality earnings. Below 50% consistently = I want to know why before I go any further.

Here is a company I found using this method:

Min 6-8: What does the balance sheet tell me about risk?

I’m not building a credit model here. I’m looking for anything that could kill the investment thesis before it plays out.

Three things I check immediately:

1. Debt and liquidity. Gross debt relative to EBITDA or operating cash flow. Does this company have the cash runway to survive if growth slows? A business that needs the capital markets to function is a business that has surrendered control of its own destiny.

2. Goodwill and intangibles as a percentage of total assets. A balance sheet loaded with goodwill is a record of past acquisitions, and often, past overpayments. High goodwill relative to assets doesn’t automatically disqualify a company, but it raises questions. What was acquired? Did it work? Are there impairment risks sitting in the balance sheet?

3. Off-balance sheet obligations. I quickly scan for operating lease commitments and any contingent liabilities mentioned in the notes. Some businesses look asset-light on the balance sheet but are actually deeply committed to long-term lease obligations that don’t show up cleanly in the headline numbers.

Minute 8-10: Is the market pricing in reality or fantasy?

Now, and only now, do I look at the valuation.

I start with what I call the sanity check trinity:

• EV/EBITDA or EV/EBIT (for profitable businesses)

• EV/Revenue (current and forward)

• FCF yield (for mature businesses)

I compare these against the company’s own historical ranges and against obvious sector peers. I’m not trying to determine fair value yet. I’m trying to determine whether the market has already priced in a perfect outcome or whether there’s a plausible entry point.

A business trading at 25x revenue with declining growth and compressing margins is pricing in a turnaround that hasn’t happened yet. That’s a story I need extraordinary conviction to enter. A business trading at 8x EBITDA with expanding margins and a dominant market position in a growing category is a story I want to understand more deeply.

The valuation check also tells me what question the market is asking. If a company trades at a high multiple, the market is betting on growth, so the key question becomes if that growth is sustainable? If it trades at a low multiple despite good margins and stable cash flows, the market is worried about something, so the key question becomes: what does the market know that I don’t?

So What Do I Know After 10 Minutes

By now, I can answer six questions:

That last question is the point.

Most companies don’t make it past this filter. That’s by design. The best analysts I know aren’t the ones who build the most elaborate models, but rather the ones who are ruthless about where they spend their time.

The main lesson

Every great investment I’ve been a part of started with clarity on the basics. Every bad one involved convincing myself that the basics didn’t matter, that the story was compelling enough to override what the numbers were trying to tell me.

The first 10 minutes force you to look at the numbers before the story. That’s the whole point.

The story comes later. The numbers come first.

The stock is down roughly 20% so far this year. Not because demand slowed.

FX headwinds hit reported numbers in a weak quarter and the market reacted to the headline without looking at what the order book was saying. The book-to-bill in that same quarter was 1.23x. Demand was never the problem.

The business itself has 92% gross margins. Zero financial debt. Reenue have grown every year for nearly two decades except two pandemic years.

Management has committed on the record to 30% operating margins by 2028 (up from 22% from today), backed by a cost structure where every incremental unit of revenue drops almost entirely to profit.

Below I walk through the full thesis, revised valuation, 2028 price target, and exactly why I see this pullback as one of the better opportunities in this position since I first added it.

Read more

- “You are not buying a business, you are buying a probability.”

There’s something finance textbooks don’t spend enough time on: standard valuation tools were built for companies that make money.

They assume positive cash flows, a growing business, and the implicit belief that the firm will be around in ten years. But some of the most interesting situations in markets involve companies that aren’t doing any of those things… Companies burning cash, drowning in debt, where the stock “should” be zero but is still trading at $2.

Understanding how to think about distressed companies is a useful skill. Not because you’ll necessarily invest in them, but because it sharpens how you think about capital structure, risk, and what equity actually represents.

Let’s walk through it step by step by using Spirit Airlines as our real-world example throughout.

First - Understand Why the Company Is Distressed

Before touching a single number, you need to diagnose the problem. Not all distress is the same, and the cause determines almost everything about how you approach the valuation.

The most dangerous situation is when a company has both problems simultaneously. Broken business model and unsustainable leverage. That combination is where equity holders tend to get wiped out entirely.

Step 1: Forget EPS. The Only Question Is the Clock

The first mistake analysts make with distressed companies is reaching for familiar metrics. Earnings per share. P/E ratios. Revenue growth. These are essentially useless when a company is losing money.

There is only one question that matters in distress: how long until they run out of cash?

Monthly cash burn. How much cash is leaving the business each month? This sets the clock.

Cash on the balance sheet, divide by burn rate. That is your runway in months.

Debt maturity schedule. When are payments due? One missed interest payment triggers default.

Revolving credit facilities. Are these available? Covenants often restrict access exactly when the company needs liquidity most.

Asset sale potential. Can they sell something to buy time?

Revenue is not safety. A company can have billions in annual revenue and still go bankrupt if its cost structure is broken. Always start with cash.

Step 2: Value the Firm First. Not the Equity

Stop trying to value the equity directly. Instead, value the firm as a whole, and then figure out what, if anything, is left for equity holders after every other claim is satisfied.

Think of the firm’s total value as a single pool of assets. Multiple claims sit on top of those assets in a strict priority order. This is the capital structure waterfall.

Equity is always last. The question is simply whether any firm value remains after every debt claim above it is fully satisfied.

To estimate that firm value, you build a forward-looking view of the business. What can this company realistically generate in cash if it survives and restructures? That becomes your estimate of the underlying asset value. For Spirit, working through normalised revenues, cost assumptions, and a terminal value on the business produced a firm value estimate of approximately $1.22 billion.

Against $3.27 billion in total debt, that leaves equity deeply underwater on a pure liquidation basis. But (and this is the important part) that does not mean equity is automatically worth zero. Which brings us to the actual model.

The Model: Valuing Distressed Equity as a Call Option

Here is the most intellectually important concept in distressed valuation, and the one that is most consistently misunderstood.

When a firm’s assets are worth less than its total debt, conventional analysis says equity is worthless. But markets persistently price distressed equity above zero. The reason is optionality.

Equity holders have capped downside. In bankruptcy, they hand the keys to creditors and walk away, they cannot lose more than they invested. But they retain open-ended upside if the firm’s value somehow recovers above the debt. That asymmetry is the definition of a call option.

This insight, which was formalised by economists Fischer Black, Myron Scholes, and Robert Merton, tells us that equity in a distressed firm is mathematically equivalent to a call option on the firm’s assets. We can use the Black-Scholes option pricing model to put a number on it.

Here is how the mapping works:

Now let’s run through each step of actually building this.

Step 1: Find the Annualised Variance in Stock and Bond Prices

The first technical challenge is that we cannot directly observe the variance in the firm’s total asset value. It is not something that trades. So we construct it from the things that do trade: the stock price and the bond price.

You need at least three years of monthly price data for both the stock and the bonds. For each, you calculate the monthly percentage change, and then annualise the variance. The annualisation is straightforward, and that is by multiply the monthly variance by 12.

For Spirit, using three years of monthly data prior to bankruptcy:

A couple of things worth noting here. The annualised stock variance of 0.40 is very high, which reflects the extreme volatility Spirit shares experienced in the years leading up to bankruptcy. The bond variance of 0.26 is also elevated. And the correlation of 0.25 between stock and bond price movements is relatively low, which as we will see has an important effect on the firm variance calculation.

Subscribe now

Step 2: Find the Annualised Variance in Firm Value

Now we combine the stock and bond variances into a single estimate of how volatile the entire firm’s value is. We do this using the standard portfolio variance formula, weighted by each component’s share of the total firm value.

The formula is:

This is essentially the same logic as calculating portfolio variance in a two-asset portfolio, except here the two assets are equity and debt, and together they make up 100% of the firm.

For Spirit, the weights come from the capital structure at the time of valuation:

Plugging those into the formula:

• Equity variance term: (0.117)² × 0.4019 = 0.0055

• Debt variance term: (0.883)² × 0.2592 = 0.2021

• Covariance term: 2 × (0.117) × (0.883) × 0.25 × 0.6340 × 0.5091 = 0.0167 × 2 = 0.0167

Annualised firm variance (σ²firm) = 0.2118

Notice what this tells us. The firm’s variance is heavily dominated by the debt variance term, because debt makes up 88.3% of Spirit’s capital structure. This is a crucial insight: in a highly leveraged distressed company, the volatility of the debt matters almost as much as the volatility of the equity when estimating overall firm risk.

Step 3: Value the Equity as a Call Option

Now we have everything we need to run the Black-Scholes model. The five inputs are:

The Black-Scholes formula for a call option gives us two intermediate values; d1 and d2, before producing the final option value.

What d1 and d2 are actually telling you: N(d2), which came out at 0.1193, is approximately the risk-neutral probability that the firm’s value will exceed the debt at expiry. In other words, the model is saying there is roughly a 12% chance that Spirit’s value recovers above $3.27 billion within the five-year window. That is a small probability. But it is not zero, and that matters.

Plugging into the full Black-Scholes call formula:

Equity Value = S × N(d1) − X × e^(−rt) × N(d2)

= $1,222M × 0.4406 − $3,267M × e^(−0.06 × 5) × 0.1193

= $538.4M − $288.9M

= $249.4 million

Per share: $249.4M ÷ 109.5M shares = $2.28 per share

The implied debt recovery of 92 cents on the dollar is actually a very useful cross-check. It tells you what the model implies creditors can expect to recover, which you can then compare against where Spirit’s bonds were actually trading in the distressed debt market. If the bonds were trading at 40 cents on the dollar but the model implies 92 cent recovery, something in your assumptions needs revisiting. That kind of triangulation is how you test whether the model is telling you something real.

What the Model Is Really Saying

Three things drive the $2.28 equity value:

Probability of recovery. N(d2) of 0.1193 means the model assigns roughly 12% probability to equity recovering value. Small, but in a $3.27 billion capital structure, even a 12% chance on the residual produces meaningful option value.

Time. Spirit’s weighted average debt duration of 5 years is doing real work here. A company with the same financials but debt maturing in 6 months would have dramatically lower equity value, since there is simply not enough time for the swing. More time equals more option value, all else equal.

Volatility. The annualised firm variance of 0.2118 is elevated, and this is the counterintuitive part. Higher volatility actually helps equity in a distressed firm. When you are deeply out of the money, you need a large swing to get into the money. A more volatile firm has more chance of making that swing. This is why distressed stocks often become more volatile rather than less as the situation deteriorates, as the market is pricing optionality, not fundamentals.

How I lost Out on A lot of Money…

The Spirit Airlines valuation above was not built this week. It was built a couple of months ago, as a modelling exercise working through exactly this framework. At the time, Spirit was trading at approximately $0.27 per share on the OTC market under the ticker $FLYYQ. The model said $2.28. The market said $0.27. Interesting gap. Noted. Moved on.

This week, Spirit surged over 600%.

There is a particular kind of frustration that comes with building a model that identifies significant undervaluation, choosing not to act on it, and then watching the stock go up 600%…

But here is why the outcome is instructive rather than just painful.

What drove the move

Reports emerged that the Trump administration is negotiating a $500 million federal rescue package for Spirit. The structure: the government provides financing in exchange for warrants that could give the U.S. government up to 90% ownership of the airline post-bankruptcy. President Trump stated on CNBC: “Maybe the federal government should help that one out.” Spirit has a hard deadline of April 30, it needs access to new financing or $240 million in restricted cash to continue operations. A bankruptcy court hearing is set for the same date.

The stock did not surge because Spirit suddenly became a good airline. It surged because the recovery probability shifted. Recall that N(d2), the probability of equity recovering value, was sitting at roughly 12% when Spirit traded at $0.27. A credible government backstop moves that probability meaningfully higher. Higher recovery probability means higher call value means higher equity price. The model predicted exactly the mechanism of this move, even if it could not predict the catalyst.

Why this is still not straightforward for equity holders

The waterfall matters here more than ever. The government is reportedly receiving warrants for up to 90% of the reorganised company. Existing common shareholders could be left owning 10% of whatever value remains after all debt claims are satisfied. The company surviving and equity holders recovering meaningful value are two entirely different outcomes, and conflating them is one of the most expensive mistakes in distressed investing.

The model said $2.28. The stock is at $1.56. On pure Black-Scholes logic, with the rescue deal in play, there is a reasonable argument it has further to run, if the deal closes and if the dilution is less severe than reported. Those are two large ifs. The clock runs out April 30.

Mental Checklist

Spirit runs through all six cleanly. It is a $5 billion revenue business that went to zero for equity, produced a $2.28 theoretical valuation from the model, traded at $0.27, then surged 600% on a government rescue, and where the waterfall and dilution math mean existing shareholders may still end up with very little even if the airline survives.

The 600% move was rational. Whether $1.56 is the right price today depends entirely on deal terms, court outcomes, and dilution math that most people buying the stock this week have not worked through.

That is always the edge in distressed investing. Not the headline. The math underneath it. And at the end of the day, that math is still just a framework for placing a bet. You are not buying a business, you are buying a probability. A probability that the firm survives, that value flows down the waterfall far enough to reach you, and that the market is currently mispricing that probability. Get those three things right and distressed equity can be extraordinarily rewarding. Get any one of them wrong and you are last in line with nothing to show for it.

Disclaimer: This post is for educational purposes only and does not constitute financial or investment advice. Nothing written here should be taken as a recommendation to buy, sell, or hold any security. Distressed and bankrupt securities are among the highest-risk instruments in public markets and are unsuitable for most investors. Always do your own research and consult a professional before making any investment decisions. All figures are sourced from Spirit Airlines public filings and public reporting as of April 26, 2026.