If you think about it, one of the best way any investor finds an edge is to identify a problem before the whole world is talking about it. After doing that, finding a company that is working towards solving this problem is actually the easy part.

I’ll give you an example.

Nvidia is perhaps, one of the most important companies in the world right now. Whether or not you think AI Capex trends seem like a fast inflating bubble situation, you cannot deny that the company is exceptionally well-positioned as the dominant foundational backbone of the AI boom. It controls an estimated 70–80% (or more) of the AI chip market through its high-performance GPUs, CUDA software ecosystem, and full data-center networking clusters.

As the generative AI phenomenon took the world by storm, in late 2022, with the release of ChatGPT, Nvidia was perhaps the quickest among the already established tech big names to capitalize on this transition.

If you go back and watch this 2011 video of Jensen Huang, over a decade before anyone outside a research lab cared about GPU compute, and you'll hear him describing, almost matter of factly, how Stanford researchers had bought consumer graphics cards from Fry's and discovered they could run molecular dynamics simulations twenty times faster than a CPU. He also talks about putting "ten years worth of Moore's Law" into the hands of scientists immediately.

Although there is no direct mention of artificial intelligence, the conditions he describes, general purpose parallel computing, a programming language built for it, floating point precision good enough for real scientific work, are exactly the substrate AI would eventually need a decade later.

This means Nvidia was very obviously a structural readiness play with an under-appreciated substrate in place. All it needed was a shift that suddenly made the world need the solution it had built and already anticipated a decade in advance.

A structural readiness play refers to a company that has built a capability years before the market has any pressing reason to pay for it, so throughout the years it would have to be absorbing the cost and the skepticism of staying the course through a long stretch where the bet looks unnecessary (or even mistaken).

Of course, the tricky part with structural readiness plays is that such companies only make sense in hindsight. The stock market is full of companies sinking billions of dollars in R&D towards problems that never arrive.

Better Place, the Israeli battery-swap EV infrastructure company, is a good example of this. This company spent close to a decade and nearly $1 billion building, what appeared to be a genuinely clever solution at the time, which were swap stations that would replace a car's depleted battery with a charged one in under five minutes. This looked like an obviously coming problem, which was that EV charging was too slow.

It raised money from real sophisticated investors on exactly this readiness-gap logic.

The company ended up going bankrupt in 2013.

As it turned out, the world's actual answer to slow charging turned out to be faster charging technology and better battery density, not swappable batteries. So Better Place had correctly identified the gap but built the wrong bridge across it.

So clearly, a company can be completely right about the problem and still lose everything on the answer, if the specific technical or strategic path it commits to turns out not to be the one the world eventually settles on. As such, if we would like our bets on structural readiness plays to actually pay off, we would also have to guage how sound the underlying bridge is.

In today's piece, I will be introducing a company that I believe fits this pattern very cleanly, and in a much less obvious corner of the market. Increasingly it seems that this gap is no longer up for debate and the bridge being this company is building seems to be the right one.

The Problem: The Blindspot in the Ocean

On the night of September 26, 2022, something detonated roughly seventy meters below the surface of the Baltic Sea. Soon after that, reports begain claiming that three of the four Nord Stream pipeline strands connecting Russia to Germany were destroyed, and columns of methane a kilometer wide were bubbling up to the surface. This was soon confirmed by satellite visuals. It remains, to this day, one of the boldest acts of infrastructure sabotage in recent years.

The most unsettling part about this attack was how little it took to carry out.

According to official findings, only a handful of people were involved, armed with a few hundred thousand dollars in explosives, and a rented pleasure yacht. The explosives were allegedly assembled and deployed by a six-person crew. It took a budget of around $300,000 to sever critical energy infrastructure that took years and untold billions to build, despite operating along the seafloor of one of the most heavily monitored bodies of water in the world.

Similarly, on to more recent times, the Strait of Hormuz (now a waterway that almost everyone in the world knows about at this point), has become a live flashpoint of global significance. About 20% of the world’s oil supply passes through this narrow checkpoint, making its smooth functioning so critical for the global economy.

As it turns out, disrupting this doesn’t actually require sophisticated attack submarines or billions of dollars worth of advanced capabilities. A naval mine can be built for as cheaply as $1500, and can be dropped from almost any vessel, left there to sit on the seafloor doing nothing until the moment a tanker sails over it.

For a disruptor, you don’t actually need to sink an entire fleet to shut down the strait. You just need to make every shipping company’s insurers nervous enough to reroute, or every captain cautious enough to slow down and wait for clearance.

The threat doesn’t even have to be executed to work. It just has to be plausible.

This is pretty much the crux of the problem out there. The ocean floor has never been treated as a defensive front line, and therefore is among one of the least defended parts of modern critical infrastructure. This is despite the fact that underwater plumbing, pipelines, mooring points, desalination infrastructure, and all the invisible wiring underneath a globalized economy is perpetually at risk.

Around 95% of intercontinental data traffic, financial transactions, defense communications, ordinary internet browsing, all of it, runs through undersea fiber cables that are, for the most part, unguarded and unmonitored outside of a narrow radius around major landing stations.

In the years since Nord Stream, there have been a string of incidents in the Baltic and North Sea involving damaged cables and dragged anchors. NATO members have begun openly discussing dedicated seabed patrol and surveillance efforts.

Identifying that the seafloor is undefended is the easy part, every defense analyst on earth could tell you that today. The harder question, especially for us investors, is who has spent the years before Nord Stream working towards the capability to address this.

Kraken Robotics: Building The Infrastructure For An Undeniable Gap

The obvious option in most people's minds would be to go long on underwater vehicle companies including Kongsberg, Saab AB, and Teledyne Marine, which build advanced autonomous systems for defense, ocean research, and industrial inspection.

However, betting on the vehicle supplier itself means betting on a specific platform winning a specific contract in a specific country, against competitors who are all chasing the same handful of high-profile navy programs.

These are not dynamics I recommend getting into because there is too much risk to make a meaningful gain.

Instead, it would be far more sensible to seek out the layer underneath all of them, the power source and the sensing payload that sits inside the vehicle regardless of which platform wins which contract, in which country, against which competitor.

This is where Kraken Robotics comes in.

Kraken Robotics is a $1.4 billion Canadian marine tech company that’s spent the better part of a decade building out subsea power, sonar, and navigation systems. Its customer base is split between defense navies and commercial offshore energy operators.

On paper, this company looks like a small industrial supplier of batteries for underwater vehicles, towed sonar systems, and services like seabed imaging. However, in the context of the critical problem we laid out above, Kraken is building both the eyes and power needed to defend ocean floors.

The most efficient subsea batteries let unmanned underwater vehicles stay in dive mode longer and range further. They also enable synthetic aperture sonar, a sensing technology effective enough to detect, classify, and identify something the size of a mine sitting on the ocean floor.

It also builds and sells towed sonar systems (the kind of hardware a navy drags behind a ship to sweep a channel before a tanker sails through it).

What most people don’t realize is that energy density and manufacturing yield in subsea batteries is a harder problem than it looks, and most UUV OEMs have historically underestimated it. On paper, picking high-density lithium cells looks straightforward. In practice, volumetric and system-level energy density drop dramatically due to harsh marine physics. I’m not going to get too technical here but this research paper offers excellent insights into this domain in depth (no pun intended).

Kraken is one of the very few players actually working on the hard engineering to address these very limitations. The company had already spent years developing higher energy density batteries, refining dual-frequency sonar payloads, and becoming one of only a handful of merchant suppliers in the world that other companies’ unmanned vehicles rely on for sonar.

As with the example of Better Place, the failed EV battery swapper, we know that identifying the right problem and building the right bridge are two different skills. Unlike Better Place, however, we don’t see Kraken betting on the wrong sensing modality, or the wrong power architecture, and getting quietly out-engineered by someone building a better bridge across the same gap. In fact, Kraken is actually not even betting on one bridge, but is pretty much building the substrate underneath several possible ones.

Given its positioning, it doesn’t need to guess which underwater vehicle program wins a given navy contract, or which specific mine countermeasure doctrine a country ultimately adopts. Whichever of these outcomes plays out, the company will supply the power and sensing layer that sits inside the vehicle regardless of who wins that contract. The company just needs unmanned underwater vehicles, in general, to keep proliferating and keep needing power and sensing.

In addition to this very strong positioning, there are a few other structural strengths worth pointing out:

1. A deliberate bet on capacity before demand.

Kraken made the conscious decision to build inventory and manufacturing capacity ahead of confirmed order flow, rather than after.

To most companies at that stage, that decision would look like a cost during the quiet years. Now that mine countermeasure demand has turned urgent, Kraken is able to deliver faster than larger, slower-moving competitors. This is a genuine edge in a market where lead time increasingly decides who wins the order.

2. Vertical integration as a moat.

The recently announced acquisition of Covelya Group, expected to close in the current quarter, combines Kraken’s batteries and sonar with Covelya’s navigation, positioning and communications technology.

Management has been emphasizing that this is not simply about adding to the revenue total. Instead, what we see is that this is about giving customers a single trusted supplier instead of several. It cuts the calibration and integration friction that comes with sourcing subsystems from multiple vendors. I see this move towards a deeper vendor relationship as a very durable structural advantage.

Financial Performance:

In terms of its performance, Kraken Robotics is a small company. It delivered just $102M in annual revenue for FY2025, a fraction of the size of the larger defense primes and offshore energy service providers it competes and partners with.

While adjusted EBITDA was positive at $25M for the year (a 24% margin), the company is not yet turning in a consistent profit, and is in fact in the loss zone, on a trailing twelve month basis. It has been running elevated CapEx ($30M in FY2025) to build out manufacturing capacity ahead of demand, funded partly through a recent equity raise that brought cash to just over $120M.

So Kraken is a pre-scale, capital-intensive small-cap story leaning heavily on a step-change in growth (65% revenue growth guided for 2026) and a pending acquisition (Covelya) to reach a size and profitability profile it hasn’t yet demonstrated on a standalone basis.

• FY2025 revenue: $102M, up from $91M in FY2024 (12% growth)

  • SeaPower battery revenue grew approximately 30% year over year

  • SAS (Synthetic Aperture Sonar) revenue grew approximately 60% year over year, with units sold in 2025 nearly matching the combined total of the prior two years

  • Subsea services revenue grew over 60% year over year, driven by the 3D at Depth acquisition and higher fleet utilization

  • KATFISH (towfish) revenue was the one weak spot, declining due to navy bid timing delays and near-completion of the Canadian Navy RMDS project

• Q4 2025 revenue: $28M, flat year over year

• Gross profit FY2025: $63M, up 42% year over year

• Gross margin FY2025: 62%, up from 49% in FY2024

• Q4 2025 gross margin: 70%, up from 48% in Q4 2024 (boosted by lower than expected project costs, a one-time benefit)

• Adjusted EBITDA FY2025: $25M, up 21% year over year

• Adjusted EBITDA margin FY2025: 24%, up from 23% in FY2024

• Q4 2025 adjusted EBITDA: $9.5M, up 36% year over year, implying a 33.5% margin versus 25% in Q4 2024

• CapEx FY2025: $30M, elevated due to the new Nova Scotia battery facility build

• Cash position at year-end: over $120M, following a successful equity financing

• Total assets: $313M at year-end, up from $162M in the prior year

• Working capital: $171M

Valuation Assessment

Because of its heavy capex and pre-net income dynamics, the company is trading at an EV/EBITDA ratio of ~90. This, on its own, is obviously a multiple many would consider too high to go long on. However, because of the high growth dynamics, analysts expect strong growth ahead. EV/EBITDA drops to ~25 on a forward looking basis.

This seems reasonable to me.

Risk Assessment

One of Kraken’s core strengths has been its clean, debt-light balance sheet. Standalone, Kraken carried $120M cash, minimal debt, and full flexibility to fund growth and acquisitions from strength.

That seems to be changingnow with the C$615M Covelya deal, funded partly through new credit facilities. This has pushed total liabilities to roughly $443M against $236M equity.

This is a materially more leveraged entity than before. Management targets cost synergies within 24 months to normalize leverage; delays would mean servicing acquisition debt without offsetting savings.

Going forward, I would watch net debt trends and synergy realization closely, this is a new thesis risk that didn’t exist standalone, prior to the acquisition move.

Investment Strategy & Milestone Tracker (for Premium Subscribers)

This section is normally reserved for paid subscribers only, but today we're unlocking it for everyone as a gift, so make sure to make the most of it

KRKNF is a strong stock, as highlighted by the geopolitically driven demand across its battery technology, the margin expansion story, and the Covelya combination that could turn Kraken into a genuine systems-level subsea supplier.

That said, I would not recommend a lump sum investment approach here.

Two binary, time-bound uncertainties (whether Covelya actually closes on schedule, and whether the historically back-half-weighted revenue pattern holds) make a staged, catalyst-gated entry a more precise way to build a position than committing capital all at once.

Here’s a much safer way to play this stock:

INITIAL POSITION (Tranche 1)

• Size: roughly 25 to 30 percent of intended total position

• Timing: now, or ahead of the Q1 2026 print

• Rationale: this establishes exposure to the base case, the organic Kraken business pre-Covelya, without betting the full position on a deal that hasn’t closed. The current $87M to $88M in announced product orders plus reiterated guidance gives some floor confidence, but deal-closing risk and back-half revenue weighting argue against going in heavier here.

TRANCHE 2: Covelya close confirmation

• Trigger: press release confirming the Covelya acquisition has closed, ideally on or near the stated Q2 2026 timeline

• Size: another 25 to 30 percent

• Rationale: this is the Critical milestone in the framework. Deal-closing risk (regulatory delay, repricing, collapse) is the single largest binary risk in the thesis right now. Waiting for confirmation removes that risk from your cost basis entirely, at the cost of paying whatever premium the market attaches to the closing news itself.

TRANCHE 3: Guidance reaffirmation with Covelya baked in

• Trigger: the updated, combined-entity guidance management said it will issue upon closing, plus confirmation that H1 2026 revenue is tracking toward plan despite the expected back-half weighting

• Size: another 20 to 25 percent

• Rationale: this is where you get to see whether the combined revenue and EBITDA numbers actually exceed the standalone $165M to $175M guide, and whether integration is showing up as execution rather than just press-release synergy language.

TRANCHE 4: Catalyst confirmation

• Trigger: any one of the following converting from pipeline to signed: a KATFISH program of record converting to an order, the large unnamed U.S. battery customer being named or quantified, or the TSX listing being completed

• Size: remaining 15 to 20 percent

• Rationale: these are the upside optionality items in the roadmap rather than base-case guidance items. Adding here means you’re paying for confirmed catalyst realization rather than speculation on any one of them landing.

RESERVE

Keep some portion, even 10 percent of total intended capital, uncommitted through the first full year of combined reporting.

The stated CUT triggers (deal cancellation, guidance cut, a second consecutive KATFISH disappointment without a credible external cause, sustained gross margin compression below 50 percent) are all things that would only become visible after at least one, and more likely two, quarterly prints post-close.

A staged entry that leaves room to simply stop adding, rather than needing to sell, is the lower-friction way to respect those trip wires.

Milestone Tracker

For anyone interested, this tranche-based investment strategy is based on 24 distinct milestones linked to Kraken Robotics. These are as follows:

Want a custom research report and investment strategy for a stock you are considering? Comission a custom report now.

Don’t just buy a stock because you like it, be strategic. Liking a company isn’t a thesis, and hoping it works out isn’t a plan.

If there’s a name on your watchlist, we’ll build it the same report and structured scorecard you see in this newsletter: the specific milestones management needs to hit, a clear Thesis Anchor to track on every call, and ADD, HOLD, and CUT triggers built around your position, not generic rules borrowed from someone else’s.

Commission a custom report today

Investor Takeaway

This is a stock for patient, thesis-driven investors, not for anyone needing quarterly consistency. It rewards a multi-year horizon, tolerance for defense procurement lumpiness, and conviction to hold through Covelya integration risk before the readiness gap catalyst fully plays out.

Size it as a satellite position, not a core holding, and let the next two to three years do the work.

Not sure if anyone can relate to this, but, in the past, I’ve actually spent a lot of time thinking about switching to a different internet browser. I typically use Google Chrome, and every so often I’ll hear about some faster, sleeker, and more private alternative and think, maybe it’s time. So I download it, poke around for ten minutes, and then, every single time, I end up right back on Chrome.

(This analysis has nothing to do with Google, or it’s very popular internet browser)

As may be the case with some of you, I eventually come to realize that all my saved passwords, emails, years of bookmarks I’ve organized, and so many of my workflows are somehow tangled into this one piece of app I’m pretty sure I never signed a contract for.

It’s more about why some businesses never need a contract to keep you, simply because unwinding yourself from them is so costly that it’s almost painful.

This is a very powerful moat.

Any company with customers that would face real operational risk, retraining cost, or some other pain point, has built something far stickier than a subscription agreement or a multi-year contract ever could.

This is why I absolutely love companies that have built a subtle dependency with years of workflows, training, and infrastructure. These companies are generally quite immune to competitors, even if they boast better spec sheets and lower prices.

Companies like this can be found across industries you wouldn't necessarily group together at first glance, such as enterprise software, medical devices, industrial equipment, and so many others.

In this issue of Earnings Intel, we’re actually going to be looking at public safety communication hardware company (quite a mouthful). This company is in an industry most investors never think to look at twice, but the company does check every box for this kind of moat. It achieves the pull we just discussed, but not quite in the way Google does with its consumers.

It’s very appealing because it has multibagger potential at a market cap under $300 million, soaring topline, EBITDA, and free cash flow, as well as structural margin expansion.

The company is BK Technologies Corporation (NYSE: BKTI).

BK Technologies (BKTI)

BK Technologies is a $290 million manufacturer of land mobile radios for public safety agencies, federal, state, and local first responders who depend on its hardware for mission-critical field communication. Specifically, it makes land mobile radio (LMR) equipment and communications solutions built specifically for public safety — fire, police, EMS, and other first responder and government agencies in the U.S. It’s headquartered in West Melbourne, Florida.

The company is doing nearly $90 million in annual revenue, and brings in about $14.2 million in net income, which points to a healthy margin in the hardware space.

Products

Its two main product families are the legacy KNG Series and the newer BKR Series, both operating in P25 digital and analog modes. For those who are unaware, P25 refers to the standard protocol U.S. public safety agencies use for interoperable radio communications.

Alongside hardware, BK also offers InteropONE, a push-to-talk-over-cellular SaaS service that lets emergency incident commanders set up group push-to-talk sessions directly between smartphone users. So it has a promising software/platform layer that sits behind its core offerings.

Customers

BKTI’s target market is primarily government, public safety, military, and state and municipal agencies. Additionally, it also serves some industrial and commercial enterprises that require the specialized hardware it deals in.

In its filings, the company discloses that it relies heavily on sales to U.S. government agencies and therefore must comply with the contracts, laws, and regulations tied to those sales.

This is why FCC approval timelines matter so much to this investment case.

Operating Model

The company has a master service agreement with East West Manufacturing for LMR radio production. This use of contract manufacturers gives it an asset light operating model. The result is the BKTI isn't carrying heavy owned manufacturing capex, which helps explain how a company this size can generate the free cash flow conversions we will discuss below.

Looking at the Numbers

From a pure numbers perspective, BKTI has been on quite the roll, lately.

Revenue growth has been very apparently accelerating in recent years. FY2025 closed out around $86.1 million, up about 12% year-over-year, with EPS climbing roughly 57% to $3.69. That momentum carried straight into 2026, with Q1 revenue rising by about 12% to $21.3 million.

Additionally, performance on the profitability side has been extremely impressive as well. Its gross margin expanded over 480 basis points to 51.8%, and net income climbed to roughly $2.8 million. Adjusted EBITDA margin increased 180 basis points to 18.7%, with diluted EPS of $0.69 versus $0.55 a year earlier.

Free cash flow surged by 44% in its most recent quarters.

These numbers show that earnings and cash flow are compounding multiple times faster than revenue, which suggests that the business model itself is doing work, not just the market.

Similarly, the margin expansion is entirely structural, and not the result of price hikes or one-time cost cuts. This is a healthier, more repeatable source of margin gains than most small-caps can point to.

The company’s fortress balance sheet is also worth mentioning here. It holds zero debt, and a record $29 million in cash, up from $22.8 million just one quarter earlier. The company can fund R&D and product development for the next several years without touching capital markets or diluting shareholders.

How BKTI Demonstrates The Upgrade Ladder Moat

Public safety radios aren’t a one-and-done purchase.

The way it usually works is that agencies buy a fleet, then upgrade it every several years as the technology improves, so that critical safety is up-to-date, without compromise.

In this case, this would mean moving from single-band radios (which only work on one frequency) to multiband radios (which can hop between frequencies, letting an agency talk to its own team on one channel and coordinate with a neighboring county or state system on another).

BKTI has sold the older generation (BKR 5000), the current generation (BKR 9000), and is now ramping up on the newer one (BKR 9500).

This is the crux of the upgrade ladder moat we’ve discussed in the beginning of this newsletter. There are over 95,000 BKR 5000 radios already sitting in the field function as a captive base for BKR 9000 and now BKR 9500 upgrades. As such, the company isn’t reliant on expensive new-customer acquisition to grow, because it’s largely selling to people who already trust and use its equipment.

So existing customers are converting to the new multiband version at a much faster clip than new customers are. Brand-new agencies are still in a slow “test and evaluate” phase, cautiously trying the product out before committing. But agencies already inside the BK ecosystem barely hesitate, as they know the brand, and their staff is already trained on similar equipment. Similarly, their dealer relationship is already in place, and the new radio lets them stay compatible with systems they already operate.

Switching to a competitor at that point would mean retraining crews, renegotiating dealer contracts, and risking interoperability with radio systems other agencies and departments depend on in an emergency.

For critical services like the fire department or police force, that’s not a decision made lightly, or often.

There are actually concrete examples of this in the company’s recent quarterly performance. Minnesota’s Department of Natural Resources placed an order for 500 of BK’s newer multiband radios. The agency already owned BK’s older radios and simply upgraded within the same platform, rather than shopping the market and running a competitive bid process. I believe this is a very strong proof point that more customers will be doing this moving forward, and this pattern is likely to persist.

This is also why the company’s earnings and free cash flow are climbing much higher than the topline growth we’re seeing. This past quarter, BKTI’s revenue grew about 12% year-over-year, but earnings (adjusted EBITDA) grew 24%, and free cash flow jumped 44%.

So clearly, BKTI isn’t fighting for every sale on price, which would impact its margins. Instead, it’s actually harvesting a base that’s already locked in.

Interestingly enough, these are only the gains coming from the BKR 5000 to the BKR 9500 transition. The company will be launching the advanced BKR 9500, sometime in the next year, pending FCC approval in H2 2026 and shipping in H1 2027.

This transition would be in-vehicle to on-person broadband. This is best read as the platform ambition sitting behind BK ONE solutions, moving the company from selling individual radios toward broadband-connected, wearable/on-person communications for first responders.

According to management, this next model is expected to be adopted even faster than the previous upgrade was. If that does play out, it will further reinforce the company’s structural advantage.

The company positions this as its Vision 2030 (not to be confused with the

On the other hand, if adoption disappoints, that would be the first real sign that the moat isn’t as strong as the story suggests.

For investors, this kind of switching-cost advantage is valuable precisely because it doesn’t rely on BK having to out-innovate competitors every single year to keep customers. It relies on the pain points of leaving staying higher than the cost of staying.

As long as public safety keeps needing radios that work seamlessly across departments and jurisdictions, that math will favor whichever company already has the fleet in the field. Right now, that’s BK Technologies.

Valuation Assessment

With its $292 million market cap, and trailing twelve month earnings of $14.2 million net income, BKTI is trading at 20.6x its earnings.

This is attractive, considering that the industry average in this space is closer to the 56.5 mark, while the median is 36.

To add, the DCF calculation suggests an undervaluation exceeding 60%:

Overall, considering the inherent structural strengths of the company, the most valuable of which is its built-in, low-cost-of-acquisition growth funnel, it seems highly valuable to me at its current price point. It has a concrete, near-term catalyst on the calendar (BKR 9500 FCC approval) expected in the second half of this year, with shipments in the first half of next. This gives investors an actual date to watch rather than a vague narrative.

I also like the fact that the same FCC certification process that’s a near-term catalyst for BKTI is also a real barrier to entry for would-be competitors, reinforcing the incumbency advantage once cleared.

Risks & Milestones to Track

Of course, this thesis I presented is not one without risks. For instance, there is timing uncertainty with regards to the FCC approval on the BKR 9500. Although management frames this as “on pace,” it is still is a regulatory dependency entirely outside the company’s control. Any slippage past H2 2026 pushes the H1 2027 shipment start and delays the next growth leg.

This is something I will be watching closely over the next few quarters.

Overall, I will also be paying attention to 25 milestones in particular over time:

Subscribe if you are interested in following this thesis along with me!

The way I believe we should go about with this investment is through a structured tranche system over time, based on the ADD/CUT/HOLD signals shared above:

• TRANCHE 1 — CORE STARTER POSITION - (allocate roughly 30-35% of intended full position)

  • Timing: Now / current levels, post-Q1 print.

  • Rationale: The thesis already cleared its first test this quarter — revenue growth, margin expansion, and cash generation all landed in line with or ahead of the story. Milestones 1, 2, 9, 12, and 15 are all tracking. This is a “management has credibility, buy the base case” tranche, not a conviction-max tranche, because the two Critical/binary catalysts (FCC approval, new-customer conversion) haven’t been tested yet.

  • Sizing logic: small enough that you’re not overexposed if the FCC approval milestone slips, large enough to have real skin in the thesis before the next catalyst.

• TRANCHE 2 — Q2 2026 EARNINGS CONFIRMATION (add roughly 25-30%)

  • Trigger to add: At least 4 of these 6 confirmed with no Critical misses — revenue tracking toward $90M run-rate, gross margin still at/above 50%, FCF growth still outpacing revenue growth, broad-based (not single-agency) demand language repeated, BKR 9500 pipeline/order commentary still positive, no walk-back of the H2 2026 FCC timeline.

  • Trigger to hold off instead (don’t add, don’t cut): SG&A lumpiness compresses margin for one quarter but management reaffirms full-year guidance — this is milestone 18 playing out exactly as flagged, not a thesis break.

  • Trigger to skip this tranche and reassess: Any softening of “on pace for FCC approval in H2 2026” language, even without an explicit delay announcement. This is the single most important sentence in the entire thesis (the anchor) — treat any hedging of it as reason to pause new capital, not just note it.

• TRANCHE 3 — FCC APPROVAL CONFIRMED (add roughly 25-30%)

  • Trigger to add: BKR 9500 receives FCC approval within the stated H2 2026 window (milestone 6). This is the binary regulatory gate — once cleared, the 9500 revenue leg becomes a matter of execution and demand rather than a matter of regulatory risk. This is the highest-conviction add point in the entire framework because it converts a Critical/CUT-risk milestone into a confirmed asset.

  • Secondary confirming signal to size up within this tranche: early dealer/contract onboarding commentary and any named early orders for the 9500 (echoing the Southwest customer pattern from Q1) alongside the approval.

  • Trigger to reduce sizing on this tranche instead of skipping it entirely: approval lands but is accompanied by commentary suggesting the H1 2027 shipment timeline is now at risk — approval without a credible shipment runway is a partial win, not a full one.

• TRANCHE 4 — POST-SHIPMENT PROOF (final 10-15%, reserve tranche)

  • Trigger to add: H1 2027 BKR 9500 shipments actually begin and management provides even directional volume or order commentary suggesting adoption is proceeding faster than the 9000 cycle did (per milestone 8’s soft commitment). This is the tranche that converts the thesis from “well-run hardware replacement story” to “second growth leg is real,” and it’s worth holding capital back specifically for this moment rather than front-loading the whole position on approval alone.

  • Trigger to withhold this tranche entirely: new-customer conversion (milestone 23) remains stalled in “test and evaluation” language through multiple quarters even after the 9500 launches — that would suggest the funnel is narrower than the existing-customer upgrade cycle management has been describing.

• CUT DISCIPLINE ACROSS ALL TRANCHES

  • Sell trigger, not just a pause: BKR 9500 FCC approval pushed beyond H2 2026, or full-year revenue/margin guidance is formally cut, or margin compression gets attributed to price/cost rather than mix. Any one of these alone should trigger trimming back to the Tranche 1 size or below, regardless of how much has been added in later tranches — these are the trip wires that were defined as thesis-breaking, not just disappointing.

  • Two soft misses together (e.g., new-customer conversion stays stalled AND a quarter shows narrow/single-agency demand with no offsetting named order) should be treated with the same weight as one Critical miss — don’t wait for a single dramatic headline when the erosion is happening gradually across multiple medium-weight milestones.

WHY THIS SEQUENCING SPECIFICALLY

The plan is built so that the largest single addition (Tranche 3) sits right after the one milestone that is both Critical-weighted and genuinely binary (FCC approval). That’s the point where uncertainty resolves most cleanly in either direction. Tranche 1 is small on purpose because two of the three biggest unknowns (regulatory timing, new-customer conversion) are still open. Tranche 4 is deliberately held in reserve rather than folded into Tranche 3, because approval de-risks the regulatory question but says nothing yet about actual adoption velocity — that’s a separate bet that deserves its own confirmation point.

Thanks for reading up to this point, it really means a lot. Subscribe for weekly deep dives on other overlooked opportunities.

Since day one, I’ve tried to build this newsletter on one principle, which is that the only reason to buy or sell a stock, is its underlying business strengthening or worsening in a way that the long term trajectory shifts. Not price swings in the short term.

The way I decided to do this was by developing comprehensive scoring trackers on popular companies with over 20 milestones, and buy, sell or hold signals based on my in depth study of the company.

That discipline isn’t going anywhere, and will always remain a foundation of Earnings Intel.

Moving forward, I’m now going to be putting that same rigor toward answering a much bigger question: which companies have genuine 100-bagger potential over the next decade.

Why this is a step up, not a step away

Tracking earnings milestones tells us whether a company is achieving all its targets that would confirm whether or not the thesis (that management has framed) is actually playing out.

That’s very valuable, but if I’m going to be honest, it wasn’t directed.

I was applying the same scrutiny to any popular company that happened to report earnings that week (regardless of whether it was actually worth building a long-term position in)

That’s a reasonable way to build a tracking system, but it’s not the best way to build wealth.

So I’m taking that exact discipline and pointing it at a much narrower, much higher-value target. We will, henceforth, be looking at the small number of companies actually capable of compounding into 100x-type outcomes over a decade.

Over the past several months I’ve gone deep into the literature on actually being able to identify multibaggers, and I’m glad to report that I’ve turned that into a working, repeatable methodology.

I’ve done this over the last few years, and am glad to have identified a number of multi-baggers well before they were on most people’s radars.

My track record is public and I’m keeping it that way. My entire archive of analyses over the last two years can be seen here.

Two years ago I flagged UTI at $15, and it’s around $41 today, roughly 2.7x. I called Micron (MU) at $80 — even after the recent pullback in memory stocks, it’s sitting near $825, close to 10x. Several other calls in that window have at least doubled. My methodology has now sharpened and pointed at the names that actually matter.

What stays exactly the same

• Long time horizons. This was never about trading, and it still isn’t. Years, not weeks — patient capital, Buffett/Munger-style.

• No shrinking the universe by size. I’m not going to artificially restrict coverage to include only names below a certain market cap (e.g. $1 billion). If a $25 billion company still has a credible path to 10x, it earns coverage. The methodology decides which companies matter now.

Each issue walks a company through the same lens:

1. Is it compounding capital at a high rate?

2. Does it have a moat that can actually hold off competition?

3. Is management genuinely aligned with shareholders, with real skin in the game?

4. Most importantly, you will be getting rigorous earnings intel - a solid checklist by which to carry forward your multibagger into the long-term.

When a name doesn’t clear the bar, I’ll tell you, even if the story is exciting.

If you want a front-row seat to how the small number of truly extraordinary long-term compounders actually get identified, this is exactly the direction I’ve been building toward, and now it’s the whole focus.

More very soon.

Year-to-date, the S&P 500 climbed by about 9.4%. During the same time, the stock for Coca Cola (NYSE: KO) climbed by 25.3%.

In fact, Coca Cola’s trajectory, over the last 20 years has been quite impressive, delivering safe capital appreciation to shareholders, without an exposure to the AI capex arms race that’s been propping up most of the index’s gains.

Many retail investors in today’s climate might be put off by a “boring compounder”, that lacks exposure to GPUs or hyperscaler spending trends. Afterall, why put your capital in a company that would give you a gain in 2 years, that you could get in less than 90 days?

Of course, that would be appealing to most people, and only a few would choose the slower option. However, investment gurus like Buffett and Munger would tell you to do the opposite. Berkshire Hathaway, in fact, has held Coca-Cola since 1988 for a reason that has nothing to do with price climb excitement. Berkshire invested $1.3 billion on the company’s stock back then, and this “boring investment” has compounded into $31 billion in the present, without even taking into account the $816 yearly dividend payout the company has been receiving.

A slow compounder like Coca Cola may not give you exciting growth in the short term, but it preserves wealth really well. Most people seeking 20%+ gains per month will also need to stomach even higher losses on a regular basis, which would make long-horizon investing incredibly unsustainable to stomach.

On the other hand, a slow compounder rarely hands you the kind of gut-punch drawdown that makes you panic-sell at the bottom, right before the real gains show up.

That’s pretty much the actual edge these companies end up delivering. It’s that they avoid the violent 30-40% pullbacks that shake most investors out of a position long before compounding has time to work. This is a game for the patient, but it can help turn over a monster gain over time, just like Berkshire did with Coca-Cola.

Compare that to trying to catch something like Nvidia in the early 2000s, which is more like trying to catch lightning in a bottle. You need to be early, you need to be right about a technology cycle nobody else sees yet, but by far the most difficult, you need to survive the 80% drawdowns that almost every eventual 100-bagger puts you through along the way. Most people don’t have the timing, the conviction, or honestly the stomach for that. Slow compounders don’t ask you to have any of it.

This is basically my argument for opting for slow compounders over high velocity teck picks.

Now let’s move on to Coca Cola itself.

The Peak Coke Myth

One of the main problems that most investors have with Coca Cola is that they assume that the company, which is a 140-year old beverage company, and has a presence in over 200 countries has little room to grow at this point. So with that, how much growth can one truly expect with if they deploy their capital towards a company facing such dynamics, afterall?

This obviously does seem to make sense.

The physical and chemical territory here (which includes formulation, flavor science, distribution logistics) is thoroughly mapped, not just by Coke, but by its competitors too.

But given the dynamics we’re dealing with here, that’s the wrong place to look for growth. Just because the physical territory is mapped, it’s wrong to assume that the growth story must be too.

Let’s look at the numeric growth itself.

Apart from the slump during the Covid years, Coca Cola has never seen a material decline in its revenue totals, and the trajectory only seems to climb upward. This is massively impressive given the market position the company has captured over the last century.

The key behind this is that the company strategically follows throughs with a handful of specific, repeatable mechanisms, instead of one big lever.

For this newsletter, we’ll go over four:

1. Revenue Growth Management

This is is really just precision pricing at the package level.

Instead of raising prices uniformly, Coca-Cola engineers different pack sizes and price points for different occasions and income levels.

A single mini can at a convenience store hits the lowest possible entry price for a cash-strapped consumer. A multipack at a grocery store targets a completely different, more premium occasion.

This is how the company grows revenue among consumers who are, by management’s own admission, under real income pressure, without simply pricing them out of the category.

2. Getting “more from more markets and more from more brands”

Rather than relying on Trademark Coca-Cola alone to carry growth, the company activates its full portfolio, fairlife, Powerade, Sprite, smartwater, Simply, and dozens of others, in whichever combination fits a given market’s needs. \

Mr. Pibb, relaunched with a bolder flavor and higher caffeine content, grew volume more than 20% in North America this quarter.

This reveals the depth of Coca Cola’s portfolio finding new pockets of demand the core brand alone never could.

3. Geographic Unnevenness

Okay so this part sounds counterintuitive but somehow it works pretty strongly in Coca-Cola’s favor.

The company doesn’treally need every market growing at once. It needs the portfolio spread wide enough that strength in one region offsets softness in another. This quarter that meant Latin America and North America carrying momentum while Mexico and China stayed under pressure.

A single-market or single-brand company doesn’t get that kind of natural hedge. Coca-Cola’s scale and portfolio breadth gives it the hedge it needs to achieve this.

4. Investing in Under-Penetrated Markets

India and China are the clearest examples of this.

Coca-Cola is accepting near-term headwinds on its margins and market share just to build affordability infrastructure and brand equity in markets where per-capita consumption still has enormous room to run relative to developed economies.

So essentially, Coca-Cola is pulling forward category growth in geographies where the category itself is still expanding. This is very different to the U.S. or Western Europe where the category is largely mature.

Put together, this is why a company can dominate a mature category and still grow. They’ve engineered a way to grow out its revenue while working with variables such as price architecture, portfolio depth, geographic diversification, forward investment in underpenetrated markets, and now, increasingly, from data.

Balance Sheet Capacity

Coca Cola’s fortress balance sheet is one of its greatest strengths.

The company’s total cash holdings and liquid investments amount to over $16 billion, while its net debt leverage stands at 1.4x EBITDA, which well below management’s own stated target range of 2x to 2.5x. Management ties this to “increased flexibility and optionality to continue to both reinvest in our business and return capital to shareowners.”

As such, the company is clearly sitting on unused balance sheet room by its own admission.

Tie into this the fact that Coca Cola generated approximately $6.9 billion in free cash flow for the quarter alone, highlights this strength. Combined with the under-levered balance sheet, this is dry powder that hasn’t been deployed yet — which makes it one of the more interesting things to track going forward.

What this could fund, in rough order of likelihood based on how management typically talks about capital allocation:

• Accelerated share buybacks — the lowest-friction option, doesn’t require new deal work, and quietly supports EPS growth on top of the operating momentum already guided.

• Continued dividend growth — Coca-Cola’s dividend track record is part of the long-term shareholder base’s core reason for holding the stock, so this lever gets used almost by default, but the pace could tell you something about confidence level.

• Bolt-on M&A — fairlife was the precedent here (a bolt-on that became a genuine growth driver once scaled). Given management’s comfort with “ahead of the curve” investment language in India, China, and now digital/AI marketing capability, a similar bolt-on in adjacent categories wouldn’t be surprising.

• Funding further refranchising/divestiture complexity — CCBA is closing, but if there are follow-on portfolio simplification moves, having balance sheet slack makes those transactions easier to structure without straining the credit profile.

• Revenue climbed 24% year-over-year to $119.8 billion, comfortably ahead of the roughly $117 billion Wall Street was looking for.

• Cloud revenue jumped 82% to $24.8 billion, accelerating from an already-blistering 63% growth rate in Q1.

• Search grew 17%.

• The cloud backlog now sits at $514 billion.

By almost every operating measure, this is a company absolutely owning the thesis it’s laid out.

The company also reported an eye-popping $9.11 EPS number, which contained a roughly $98 billion in unrealized gains on Alphabet’s equity investment portfolio (its Anthropic and SpaceX stakes, per reporting). Even after stripping out, core operating EPS was closer to $2.85-2.90, which reflects a 26% YoY jump.

And yet the stock sold off. Down roughly 4-5% in the immediate reaction, continuing a slide that’s actually been building since Alphabet’s all-time high back in May.

So what’s going on, and does it change anything about the thesis we’ve been tracking so far?

FIRST THINGS FIRST: HOW DID IT SCORE ON THE CHEAT SHEET

Before we get into the selloff, let’s actually run the tape against the milestone framework we built earlier. I believe this is the part that matters more than the stock’s mood swings because it helps us track the thesis objectively.

I apologize for the fuzzy quality of this image, it is, unfortunately, a necessary limitation with Substack, to ensure smooth email delivery. For anyone that wants a high definition image of this scorecard, it can be viewed here.

Now let’s go over the important parts, one by one.

Critical Milestones:

• Cloud revenue growth above 50%. HIT, decisively. 82% YoY, up from 63% in Q1. This is accelerating off an already-difficult comp. Clean ADD.

• Backlog conversion tracking toward the 50%-in-24-months guide. HOLD for now. Backlog grew to $514B from $462B, real dollar growth, but the sequential percentage growth slowed from a near-doubling to roughly 11%. Not a miss, since this target was not laid out to be achieved immediately in the next quarter. Management did reaffirm tbat the conversion guidance remains unchanged. So the underlying commitment held even as the growth rate normalized.

• 2027 CapEx quantified: MISS, for the second consecutive quarter. This is the one that matters most and we called it out explicitly after Q1 as a discipline test. It’s now failed twice.

High Priority Milestones:

• TPU third-party hardware revenue on the stated 2026-into-2027 timeline: HIT. Q2 was the first quarter this revenue was actually recognized, described as small now and ramping toward the “vast majority in 2027” language used since Q1. Tracking exactly to plan.

• Cloud operating margin above 30% despite the Wiz drag: HIT, and better than hit. Margin expanded to 35.6% from 32.9% in Q1. Forward caveat: management flagged a new third-party capacity leasing strategy for Q3 that will pressure margin near-term. Not a miss yet, but a new item now on the watch list.

• Gemini Enterprise MAU growth near 40% QoQ: Directionally strong but the specific comparable metric wasn’t repeated this quarter. Nearly 90% of Fortune 100 now on Gemini Enterprise and 70 million cumulative Agent Development Kit downloads are good proxies, so we’re scoring this ADD on strength of penetration data, with a note that the original growth-rate metric may be getting slowly retired now that the raw numbers look better as a story.

• Search ad coverage expansion beyond the ~20% baseline: Still no hard number, second consecutive quarter this has gone unquantified. AI Max is out of beta with 500,000 advertisers and a 15% average conversion lift, which is real, but management continues to gesture at this rather than measure it.

Medium Priority Milestones:

• Gemini 3.5 Pro shipping on the guided June timeline: MISS, and the notable kind. Guided explicitly for June back on June 3, this call instead said “currently in testing” with zero acknowledgment the original date passed. No cause named, no revised date given.

• Customer commitment overage: HIT and accelerating. 50% overage this quarter versus 45% in Q1. This is expansion revenue within the existing base, not just new-logo growth, and it’s one of the cleaner underlying-health signals in the whole print.

New addition to the list: free cash flow. Not on our original checklist, but it needs to be now. Q2 printed negative FCF of $5.9B, the first negative quarter in this coverage arc, a direct and named consequence of the CapEx acceleration. Adding this as a Critical milestone going forward — watch whether it returns to positive within a defined window or keeps sliding as the buildout continues.

WHY IS GOOGLE FALLING?

Three things seem to be doing the damage.

First, and biggest is that CapEx guidance went up again.

Alphabet lifted its full-year 2026 CapEx range to $195-205 billion, up from $180-190 billion set just one quarter ago. That’s the second guidance raise in as many quarters, and management still won’t put a number on 2027 beyond “significantly increase.”

From the perspective of shareholders, this somewhat sounds like the management is saying, “trust us, the returns are there” twice in a row, but they just don’t seem to be getting to a ceiling to underwrite against. That is concerning. The ambiguity is a real cost to the stock, even if the underlying logic (chasing genuine demand) hasn’t changed.

Second, and somewhat connected is that free cash flow actually went negative this quarter, to the tune of $5.9 billion.

FCF margin reportedly compressed from around 21% to roughly 9%. This is obviously the mechanical consequence of the CapEx acceleration, and it’s the first quarter where the spend has visibly shown up as cash leaving the building rather than just a number in a guidance slide. Things are different this time, and it’s starting to concern the market.

What Actually Went Right

Strip away the CapEx noise and the results actually highlight that the underlying business is arguably stronger than last quarter, not weaker.

Cloud didn’t just grow, it accelerated, even after backing out the first TPU hardware revenue recognized this quarter.

Operating margin in Cloud expanded to 35.6% from 32.9% in Q1, despite already absorbing the Wiz integration drag.

Existing customers are now exceeding their contract commitments by 50%, up from 45% last quarter, meaning the acceleration isn’t just new logos, it’s expansion within the base.

Nearly 90% of the Fortune 100 are now using Gemini Enterprise.

And critically, management reaffirmed the thesis that Google is still supply constrained, still turning away revenue they could otherwise capture.

Search also just posted an all-time high in query volume during the World Cup, still growing at 17%, with no visible sign of AI cannibalizing the ad business.

The bear case that generative search eats Google’s core cash cow continues to not show up in the actual numbers.

The Tension, Plainly Stated

So the demand-exceeds-supply thesis is intact. The margin expansion story is intact.

If anything has changed, it seems to be the market’s patience for funding that story in cash rather than narrative.

We now have two consecutive CapEx raises without a 2027 anchor, plus a first quarter of negative free cash flow. So in a way it is fair for the market to start feeling a little impatient now and ask, at what point does “high-ROIC spend” need to start showing up as actual cash?

Verdict: Hold, Leaning Add On Weakness

Nothing according to our thesis tracking framing trips a Cut.

The core engine is firing on every cylinder we’ve been tracking, and the thesis anchor still holds.

We’ve mentioned what’s causing the sell off, and it is neither a demand problem, a competitive problem, or a margin problem, it’s more of a a funding-timeline problem, and funding-timeline problems are exactly the kind of thing that gets overcorrected in a single trading session by a market currently jumpy about anything with “AI CapEx” attached to it.

The stock was already down over 12% from its May highs heading into this print, on sentiment more than fundamentals. This report gave the bears a headline to hang that sentiment on, but it didn’t give them a broken thesis to point to.

At current levels, we're being offered the strongest Cloud growth print in this company's history, an unwavering compute-constrained demand backdrop, and continued Search resilience against the AI-cannibalization bear case, all at a price that's discounting execution risk on a spend plan that management has been explicit and consistent about for three straight quarters.

That's the setup we want: real operating strength, a temporary sentiment-driven markdown, and a clear tripwire list to know if we're wrong.

Is this kind of analysis is useful to you? Imagine having it built specifically for the stocks you actually hold.

Commissioned scorecards are available with the same framework, same depth, built around your portfolio. Long-term thesis trackers, milestones, BUY/CUT/HOLD signals

Know what you own, know what to watch for, and never sell a great investment because of pure volatility.

Get your exclusive report now

Disclaimer: This newsletter is for informational and educational purposes only and does not constitute financial, investment, or legal advice. Nothing in this report is a recommendation to buy, hold, or sell any security. All analysis reflects the author’s independent interpretation of publicly available information (earnings calls, transcripts, and company disclosures) and may contain errors or omissions. Investing involves risk, including the potential loss of principal, and past performance is not indicative of future results. Always conduct your own due diligence and consult a licensed financial advisor before making any investment decisions.

Google’s entire bull case has been resting on the idea that they are “compute-constrained, not demand-constrained” and every dollar they spend closing that gap is high-ROIC dollar. The argument is basically that they’re the only company that owns the entire stack from silicon to distribution.

Of course, whether or not this thesis actually plays out is going to be the key question this quarter (and probably the next few quarters)

Generally, if we see management continuing its narrative that “we could’ve sold more if we had it.” that means the thesis is still going strong, and they might keep their CapEx trend at a similar trajectory.

LEG ONE: DEMAND GENUINELY EXCEEDS SUPPLY

• Management has said versions of this on every call this year

• Backlog nearly doubled sequentially to over $460 billion

• They’ve admitted cloud revenue would have been higher last quarter if they’d had the capacity to fill it

• That’s either a great problem to have, or a warning sign, depending on how long it persists

LEG TWO: THE SPEND IS DISCIPLINED BECAUSE OF VERTICAL INTEGRATION

• This is what separates Google’s CapEx story from a generic “everyone’s building data centers” narrative

• They own the TPU silicon, the models through DeepMind, and the distribution through Search, YouTube, and Workspace

• Every CapEx number gets paired with an efficiency proof point, like the 78% reduction in Gemini serving costs

• The argument: this spend is structurally cheaper than a company renting GPU capacity from someone else

LEG THREE: AI IS EXPANDING THE BUSINESS, NOT EATING IT

• This is the answer to the oldest bear case on Google

• Search queries at all-time highs

• AI Overviews usage climbing into the billions

• Management wants AI seen as additive revenue stacked on top of the existing ad machine, not a slow leak in it.

As promised, here is the full milestone cheat sheet I will be analysing

Is this kind of analysis is useful to you? Imagine having it built specifically for the stocks you actually hold.

Commissioned scorecards are available with the same framework, same depth, built around your portfolio. Long-term thesis trackers, milestones, BUY/CUT/HOLD signals

Know what you own, know what to watch for, and never sell a great investment because of pure volatility.

Get An Exclusive Report Now

WHY THIS MATTERS FOR THE NEXT CALL

The entire thesis only holds together as long as management keeps being able to say demand exceeds supply. That would be doing more work than any individual metric. The moment that framing shifts, even subtly, toward demand normalizing or supply catching up, the whole capital-intensity story needs to be re-underwritten.

WHAT WE WANT TO SEE, IN ORDER OF IMPORTANCE

1. The compute-constrained language survives intact

   • Ideally with a fresh proof point attached

2. Cloud growth stays in the 50%+ range

   • Some deceleration off last quarter’s 63% is expected and fine

   • A drop into the 40s or lower without a clean, named explanation is the first real crack

3. 2027 CapEx finally gets a number

   • Management has said “significantly increase” twice now without quantifying it

   • A third quarter of vague language starts to look less like prudence and more like avoidance

4. Cloud margin holds above 30%

• Despite the Wiz integration drag and rising depreciation

• This is the actual proof the full-stack cost advantage is real money, not just a talking point

Earnings Drop on Wednesday. Make Sure You’re Subscribed

Google reports Wednesday. I’ll be going through the transcript line by line against all 24 milestones, and the full breakdown will be out by Thursday. If you want the scorecard before it’s buried under a hundred other takes, now’s the time to subscribe, not after the print drops. The newsletter is completely free.

The full ADD / HOLD / CUT tranche framework and trip wires, are going out on the newsletter.

Feel free to reply to this email if you have any thoughts to share! I respond to all comments and queries about this thesis I’m tracking!

Given this fall from its 52-week high, I’ve come across many on Substack, Reddit, and X, calling Microsoft a steal.

I mean, if you go by the numbers, that certainly seems hard to argue with. Yes it definitely is cheap relative to its own history, but of course, that does not mean that it’s cheap in absolute terms.

Normally, when we see a company trading at a forward PE of 19.53 and an EV/EBITDA of 15.76, especially with the structural strengths Microsoft holds, the conversation is pretty short.

But is the Microsoft story as simple as the numbers make it seem?

For context, Microsoft is in the middle of a $31.9 billion per quarter capital expenditure cycle. Free cash flow has compressed to $15.8 billion this quarter as a result, which is jarring compared to its $46.7 billion in operating cash flow.

The market seems to have a very specific concern right now, and I think it boils down to this.

Is Microsoft’s AI infrastructure spending going to pay back at the returns management is implying, and if so, when?

In today’s breakdown, I’m not going to try to answer that with a prediction. We’ll be going to do something more useful. We are going to lay out exactly what Microsoft’s core thesis has been, and build a framework for tracking whether the evidence is pointing toward yes or no, quarter by quarter.

That is what Earnings Intel is built around. Know the thesis. Track the evidence. Only act when the thesis moves.

Here is the Microsoft thesis, and here is what you need to watch.

Before getting into our breakdown, here’s Microsoft’s company cheat sheat for investors or those watching from the sideline. I’d suggest you download it if you plan on tracking the Microsoft thesis into its next earnings (which take place in the next few weeks).

While I planned on keeping the tranche decision framework part of this cheat sheet hidden behind a paywall, I’ve decided to keep the whole scorecard free for all subscribers for the foreseeable future.

If you’d like to discuss anything specifically about these milestones, feel free to reply to this email, or if you’re on Substack, simply comment on this post.

If this kind of analysis is useful to you?

Imagine having it built specifically for the stocks you actually hold.

Commissioned scorecards are available — same framework, same depth, built around your portfolio. Know what you own, know what to watch for, and never get caught flat-footed on earnings day.

COMMISSION A CUSTOM REPORT

THE CORE THESIS (& WHAT WOULD BREAK IT)

Here is how to think about what you actually own when you own MSFT at this point in the cycle.

The thesis has three load-bearing pillars.

1. Pillar one is that AI ARR keeps compounding at scale.

   
   Right now it is growing at 123% on a $37 billion base. The market is pricing in continued compounding. If that growth rate decelerates sharply (say, below 75% year-over-year) it signals one of two things.

   Either enterprise AI adoption is hitting friction faster than expected, or competitors are taking share. Either way, it changes the multiple you should be willing to pay.

2. Pillar two is that the seat-to-consumption transition actually happens.
   
   Microsoft is in the middle of a deliberate shift. and it’s moving customers from fixed Copilot seat pricing toward a hybrid model where they also pay for what they use.
   
   Nearly 60% of Dynamics service customers are already buying usage-based credits. GitHub Copilot just moved to usage-based pricing entirely. If this transition works, revenue per customer expands significantly without needing to add new logos.
   
   If it creates churn or confusion, the near-term numbers get messy.
   

3. Pillar three, arguably the most important, is that the data flywheel does not stop spinning.
   
   Every Copilot interaction feeds back into Work IQ, making future responses better, which drives more usage, which generates more data.
   
   Microsoft is framing this as the moat that is genuinely hard to replicate. But we need to keep in mind that it only matters if usage intensity keeps growing.
   
   During the last quarter, Copilot queries per user were up nearly 20% quarter-over-quarter. That number needs to keep moving.

WHAT TO WATCH ON THE NEXT CALL

I already shared the 25 metrics to track in the cheat sheet above. But we need to keep in mind that some are much more critical than the others:

The AI ARR number and its growth rate.

This is absolutely critical. Satya said $37 billion growing at 123% in the last quarter. If that rate compresses materially (and not gradually) that is your first serious warning sign. I would suggest investors to write down the number every quarter and track the trajectory.

M365 Copilot seat growth.

Currently at 20 million seats, 250% year-over-year. The next threshold to watch is 25 million. More importantly, watch whether the growth rate holds above 150% year-over-year as the base gets larger. (A deceleration below 100% without a corresponding pickup in consumption revenue would be a meaningful yellow flag.)

Copilot queries per user. So this is the usage intensity metric.

Remember that seats are the headline. Queries per user tell us whether those seats are actually being used daily or sitting dormant. Weekly engagement is now reportedly at the same level as Outlook, which is an extraordinary claim if it holds.

I’ll be watching for this language to either strengthen or quietly disappear.

GitHub’s usage-based pricing transition.

So as mentioned above, one major Microsoft development seen this year is that the company is moving GitHub Copilot to usage-based pricing. I feel that this is a structurally correct long-term move but creates short-term noise.

Watch for any language around enterprise subscriber churn or revenue recognition timing in the next one to two quarters. A smooth transition is a high-conviction signal. A messy one is not thesis-breaking but warrants attention.

Free cash flow recovery.

CapEx was $31.9 billion previous quarter. Free cash flow was $15.8 billion as a result, which is well below the operating cash flow of $46.7 billion.

Management is implicitly signaling that revenue scaling will restore FCF conversion over time. Based on this, I would watch for sequential FCF improvement as the clearest indicator that the investment cycle is beginning to pay back.

Operating margin.

Held at 46% this quarter despite the CapEx load. If it compresses below 43% without a clear management explanation, something is wrong with the cost structure. If it holds or expands, it confirms that AI revenue is being absorbed at healthy incremental margins.

WHEN TO ACTUALLY SELL

When it comes to Microsoft, we need to keep in mind that not every “bad quarter” is a sell signal. This company is a massive, diversified business and quarters can get noisy. Here is the framework for when the thesis is genuinely breaking down versus when it’s just turbulence.

Sell (or seriously reduce) if any two of the following occur simultaneously:

• AI ARR growth rate falls below 75% year-over-year without a credible recovery path articulated by management.

• M365 Copilot seat growth falls below 100% year-over-year with no offsetting consumption revenue stepping in.

• Operating margin compresses below 43% with no clear path back articulated within one quarter.

• GitHub enterprise subscribers show measurable net churn after the pricing model shift.

• Management stops disclosing AI ARR as a standalone metric — the single biggest red flag of all, because it would suggest the number no longer tells a flattering story.

Hold through the noise if the AI ARR growth rate is decelerating gradually but still above 80%, if seat growth slows but consumption revenue is picking up the slack, or if margins compress slightly while FCF is recovering. These are growing pains, not thesis breaks.

THE BOTTOM LINE

Microsoft had a very good quarter, the last time it reported its results. More importantly, I feel that it had a quarter that is consistent with the thesis.

The infrastructure is scaling. The products are getting used. The enterprise deals are getting bigger. The data flywheel is spinning faster.

None of that means you buy blindly or hold forever regardless of what happens. What it means is that right now, the burden of proof is on the bear case, not the bull case. And the bear case has not shown up in the numbers yet.

Going forward, it’s critical to watch the AI ARR number. Watch the queries per user. Watch the FCF recovery.

We should also watch the language coming out of management’s mouths, because a lot of times that reveals when the conversation is starting to change.

Until then, the thesis is intact.

WANT ME TO WATCH THIS FOR YOU?

If you want to keep following Microsoft through this cycle without having to do all of this yourself, subscribe to Earnings Intel.

Every quarter, when the transcript drops, I will run it through this exact framework. Same milestones. Same trip wires. Same thesis anchor. I will tell you what hit, what missed, what management quietly changed in their language, and what it means for the position. No noise. No hot takes written twenty minutes after the bell. Just the systematic read that tells you whether the thesis is still intact or whether something has actually changed.

You will not have to wonder whether that AI ARR number is still growing at the right rate. You will not have to parse Satya’s prepared remarks for the one sentence that matters. You will not have to figure out whether a bad quarter is turbulence or a real signal.

That is what I am here for.

In case you’ve missed it, I released a Pre-Earnings Cheat Sheet earlier this week, based on the overall thesis management has been trying so hard to convey to its shareholders.

Based on the extracted thesis, I identified 25 milestones we should be watching out for in the new Nike earnings drop. Given this performance, we should be able to determine whether the company’s business standing (and investability) has improved, or if any trip wires have been triggered, which would warrant a cut from investor’s portfolio.

Well since, the earnings have finally come, I’m sticking to my promise, and sharing the post-Earnings scorecard I’ve built:

If this kind of analysis is useful to you, imagine having it built specifically for the stocks you actually hold.

Commissioned scorecards are available — same framework, same depth, built around your portfolio. Know what you own, know what to watch for, and never get caught flat-footed on earnings day.

Check Out Now

So How Did Nike Actually Perform?

Honestly, better than management was preparing us for. It’s a smart move to set expectations low, and the over-deliver.

The company delivered well on the North American front, which it was clearly betting big on. It also reported back with positive progress on the Greater China front. There’s ALSO been positive updates on its cost and capital structure.

The things that matter most to the thesis seem to have held.

• The Running division hit five consecutive quarters of double-digit growth and added roughly a billion dollars to the business over that stretch.

• The World Cup execution was genuinely impressive

  • Kit sales were 2.5 times the 2022 pace.

  • The Mercurial became the fastest-selling 24-hour cleated footwear launch in NIKE Direct history, with 1.5 billion views in the first week.

• The gross margin inflection got pulled forward one quarter, which is a real positive surprise.

• Management reiterated the Spring 2027 Sport Offense ramp and the CY2026 Win Now sunset without any softening of language whatsoever.

These are positive developments.

That being said, however, I still wouldn’t say we have our clean bull case yet.

One big reason for this is that the macro environment showed up mid-quarter in a way that complicated the story.

After a strong March, retail sales decelerated in mid-April, and management attributed it to consumer pressure on discretionary spending and traffic. Of course, this was correlated with gas prices in North America specifically.

Given these pressures, revenue guidance for the forward period was revised from down low single digits to down low to mid-single digits.

This is not a good sign for the company, and its topline guidance being snipped does not play well into the recovery thesis. To make things worse, Sportswear is now explicitly expected to be negative for the entire FY2027 (with improvement only expected in the back half.) Keep in mind, Sportswear and Jordan Streetwear together represent approximately half of NIKE’s entire revenue.

So to sum up, management has now formally deferred any meaningful recovery in its main product categories, pushing it to the second half of 2027. That means the revenue recovery story is essentially a FY2028 story, not FY2027. The sport performance business is working. The brand is rebuilding. But the size of the Sportswear hole means top-line growth is not coming soon.

Then there is the CFO departure, which adds another risk that did not exist before. Matt Friend pretty much architected the financial framework of the entire comeback, so him leaving is definitely a setback. His replacement will be executing on a roadmap they did not design.

So the summary is that while the World Cup is bringing a bit of a boost, and the gross margin path is improving faster than expected, the macro environment delays the company’s comeback plans.

Valuation

At its current ~$41 price, Nike stock trades at 27.5 times its forward GAAP earnings. I find that way too high, given the circumstances the company is sailing through, AND the fact that the median PE ratio in the consumer discretionary sector is closer to 15.

Nike would probably defend that premium by pointing to structural strengths such as athlete relationships, innovation capability, brand equity, wholesale infrastructure, global scale, etc. I would argue, however, that, how valuable could these strengths really be if they cannot be leveraged to ensure a recovery within the projected timeline. Factor into this the macro weakness, and the new risk factor of the CFO’s departure, things seem to definitely have taken a negative turn for Nike.

Therefore, to me, the stock seems pretty clearly overvalued.

This scorecard is for informational purposes only and does not constitute financial advice. All forward-looking statements are derived from management commentary and public filings. Past performance is not indicative of future results. Consult a qualified financial advisor before making investment decisions.

For context, if you had invested $10,000 into Nike in late 2022, your investment would today be worth around $2,351.

A lot seems to have gone wrong with the company in recent years, and the worst part is that this seems to mostly have been self-inflicted.

1. Went all in direct to consumer, cut wholesale partners, destroyed those relationships

2. Flooded the market with AF1, AJ1, Dunks, and killed the heat on some of their best franchises

3. They chased lifestyle revenue, and pretty much abandoned the sport performance identity (major strategic blunder imo).

4. One of their worst moves was to build a fixed-cost infrastructure for a digital business that never ended up reaching the scale management projected.

5. Misread China’s channel dynamics and let inventory pile up across digital and wholesale simultaneously

6. Its EMEA markets showed structural mismanagement, so the management’s response was to paper over it with volume, rather than addressing structural causes of weakness.

7. Tried to fix all of it at once from a position of weakened brand equity and elevated inventory everywhere which just worsened things for itself.

Management now insists that they are working to correct these misteps. The main thesis they’re trying to push is that the company never lost its fundamental advantages (athlete relationships, innovation capability, brand equity, wholesale infrastructure, global scale, etc.) and all wrongs actually related to discipline.

In prior earning calls, they seem to be pretty explicit that the timeline is longer than they initially suggested and some geographies are messier than initially diagnosed. China and EMEA have structural weaknesses compared to North America, and require more time.

They have guided Q4 revenue to fall between 2% and 4% of last year Q4’s total of $11.1 billion. If the reported revenue is below $10.6 billion, that’s going to be really negative for the Nike. They’re betting on the North American market showing strength, and if that does not play out, their thesis is breaking down.

Personally, I think the stock is expensive right now at 28x forward earnings, especially since the best signal we could be given is that the fundamental decline is slowing, not reversing. I think that’s a pretty thin foundation for a 28x multiple.

What to watch now

As promised, here is the Cheat Sheet that contains the 25 Milestones I am watching closely, which would help us understand whether or not the management’s thesis is actually playing out. Keep in mind, even management does not expect extraordinary recovery to show up right now. The bar here is not recovery. It is whether the damage is being contained, and whether the right signals are starting to show up underneath the headline numbers:

Based on these milestones, I’ve developed a tranch system that will help us decide whether the stock should get a positive, negative, or neutral rating in the upcoming earnings:

TRANCHE DECISION FRAMEWORK

ADD
Any 5 of the following 7 conditions met, with no Critical misses: (1) Running sustains double-digit growth in Q4 FY2026; (2) Q4 revenue lands within guided -2% to -4% range with North America positive; (3) Spring 2027 Sport Offense product ramp executes on schedule; (4) North America Digital discounts continue declining sequentially; (5) Gross margin expansion confirmed in Q2 FY2027; (6) Football, training, and basketball return to growth within 2 quarters; (7) Investor Day held fall CY2026 with quantitative long-term targets.

HOLD
Win Now action timeline intact (CY2026 completion), Running growth sustained, North America sequential improvement continuing — but one or two High milestones slipping (e.g., football/training ramp delayed one quarter, EMEA inventory cleanup extending into Q1 FY2027, Digital discounts flattish rather than improving). No Critical commitments explicitly walked back.

CUT
Any 2 of the following: (1) CY2026 Win Now completion deadline pushed explicitly; (2) Spring 2027 Sport Offense product launch underwhelms on sell-through; (3) North America growth reverses — all-channel positive February signal does not hold in Q4; (4) Q2 FY2027 gross margin expansion fails to materialize; (5) Running growth decelerates below double digits for two consecutive quarters. OR: single occurrence of Hill or Friend explicitly walking back the CY2026 Win Now completion language.

ROADMAP ADD CALLS

• Spring 2027 Sport Offense product ramp: first clean look at the new product architecture hitting shelves. If sell-through is strong, this is the ADD moment before the broader revenue recovery is fully priced in.

• Q2 FY2027 earnings: the gross margin inflection quarter. If expansion materializes as guided, the transitory-costs narrative is confirmed and the earnings recovery arc becomes investable.

• Investor Day (fall CY2026): if long-term quantitative targets are disclosed and the 2027–2028 innovation pipeline is compelling, this is a potential re-rating catalyst.

• World Cup 2026 (June–July): if Mercurial launch, federation kit execution, and 5,000-door retail presence land well, the football category momentum becomes a visible second pillar alongside running.

Upcoming earnings are very critical because NIKE has stacked its entire credibility on a sequence of very specific, time-bound commitments. I feel that the next two to three quarters are when the market finds out if those commitments hold.

Management gave us a map. Upcoming earnings are where we will find out if the map matches the terrain.

Immediately after the earnings drop, I’ll be publishing a full post-earnings diagnostic, milestone by milestone, and what it means for the position.

Make sure you’re subscribed so you don’t miss out!

Want this level of analysis done exclusively for a stock on your portfolio? Commission a custom Earnings Scorecard here.

Get My Custom Report

What a quarter this has been for Micron. Most of us were expecting a bang for Q3, but nowhere near what actually landed.

• Revenue came in at $41.5B against $33.5B guidance (an extraordinary $8B beat at the top end).

• Gross margin printed at 84.9% against 81% guidance.

• EPS was $25.11 against $19.15 guidance.

• Free cash flow hit $18.3B, a new quarterly record.

• Net cash surged from $6.5B to $24.4B in a single quarter.

If you’ve received the Earnings-Intel pre-earnings cheat sheet, you probably know that we were watching 25 targets for this earnings call. Of 25 milestones tracked, 23 hit, one pending (HBM4 16-high product update), one on track with no change. Zero misses.

Want this level of analysis done exclusively for a stock on your portfolio? Commission a custom Earnings Scorecard here.

Get My Custom Report

(Scroll to the end of the newsletter for a detailed milestone-by-milestone read)

Overall, what I was looking for in the earnings call was to determine whether or not the management’s thesis is still intact. Previously, management was optimistic that supply-demand conditions for both DRAM and NAND will remain tight well beyond calendar 2026. This quarter, they pushed this forward to calendar 2027, which is excellent news for Micron shareholders.

There were two more developments that I believe mattered the most beyond the financial beat.

First, Micron announced 16 Strategic Customer Agreements, up from 1 at the prior scorecard. These are take-or-pay, non-cancellable, 5-year agreements with price floors and ceilings, $100B in minimum cumulative RPO across 14 of the 16, and $22B in customer cash deposits. The business model transformation that was a thesis premise is now a contractual reality.

Second, Q4 guidance was set at $50B revenue and 86% gross margins — numbers that would have been considered implausible as annual targets one year ago.

Tranche signal: ADD. All six ADD conditions met. No Critical misses.

Risks to keep in mind

Despite the exceptional performance, I will still be keeping track of the following risks in the upcoming quarters and years.

1. Roughly 40% of revenue will be locked at or near current CQ2 market prices. This means that Micron cannot benefit from further price appreciation on covered volume. I understand the need to do this to secure maximum SCAs, but it does cap upside somewhat.

2. There is a cash deposit return obligation in the SCAs. Around $18B in customer cash must be returned over the SCA term, back-end weighted. We should watch closely whether this affects how excess cash should be interpreted for capital return purposes.

3. CXMT and YMTC could potentially be seen as growing competitive threats to Micron. While the management seems to be framing their output as predominantly China-domestic, we cannot ignore their improving capabilities.

4. At this stage, customer concentration in the SCAs is quite noticable. Four large customers drive the bulk of SCA-related revenue and RPO. Take-or-pay terms provide protection, but counterparty concentration remains a structural risk.

5. 86% gross margin is beyond exceptional. Some would wonder if this should be taken as a peak (interestingly enough, this was considered for the prior 81% margin too). Management declined to guide beyond Q4 and acknowledged that incremental price gains have diminishing margin impact at current levels.

MILESTONE-BY-MILESTONE SCORING

CATEGORY 1: REVENUE & MARGINS

• Milestone 1 — Q3 FY2026 revenue reaches $33.5B ±$750M
  Result: $41.5B — a beat of approximately $8B above the top end of guidance
  Status: HIT — Exceptional beat
  Context: The $17.6B sequential increase was the largest in company history, eclipsing the prior record set just one quarter earlier. Revenue was up 346% year-over-year. The beat was driven by pricing acceleration across both DRAM and NAND, exceeding what management had signaled at the JPMorgan conference in May. No management-attributed cause for the upside beyond continued supply-demand tightness and favorable mix.

• Milestone 2 — Q3 FY2026 gross margin reaches approximately 81%
  Result: 84.9% — approximately 390bps above guidance
  Status: HIT — Exceptional beat
  Context: Gross margin of 84.9% is a new company record, up 10 percentage points sequentially. Management attributed the beat to higher pricing and favorable mix, with continued strong cost execution. Q4 guidance is 86%, a further 110bps sequential expansion. The magnitude of this beat directly validates the thesis that memory has been structurally repriced in the AI era.

• Milestone 3 — Q3 FY2026 EPS reaches $19.15 ±$0.40
  Result: $25.11 — a beat of approximately $6 above the top end of guidance
  Status: HIT — Exceptional beat
  Context: EPS grew 106% sequentially, the largest sequential EPS expansion in the company’s history in dollar terms. The beat reflects the combination of revenue upside and margin expansion flowing through cleanly to the bottom line. Tax rate came in at 14.9%, slightly below the guided ~15.1%, a minor additional tailwind.

• Milestone 4 — Free cash flow substantially exceeds Q2’s $6.9B record
  Result: $18.3B — approximately double the “could roughly double sequentially” framing from Mark Murphy
  Status: HIT — Exceptional beat
  Context: Q3 FCF of $18.3B against $7.1B in CapEx on $25.4B of operating cash flow. Murphy had suggested FCF could “roughly double” from Q2’s $6.9B — the actual result more than doubled that framing. Q4 FCF is guided to “increase substantially again,” with CapEx rising to ~$10B against a $50B revenue guide, implying Q4 FCF could reach $25B–$30B. Net cash balance reached $24.4B, up from $6.5B at Q2 end.

• Milestone 5 — DRAM revenue continues at approximately 79% of total revenue, with pricing elevated
  Result: DRAM at 76% of revenue, prices up low-60s% sequentially
  Status: HIT (with minor mix note)
  Context: DRAM’s share of revenue declined slightly from 79% to 76% as NAND nearly doubled sequentially (up 99%), reflecting the exceptional strength in data center SSD. DRAM pricing remained firmly elevated. The modest share shift is not thesis-negative — it reflects NAND mix improvement toward higher-value products, not DRAM deterioration.

CATEGORY 2: HBM EXECUTION

• Milestone 6 — HBM4 ramp at twice the pace of HBM3E 12-high, yields improving faster
  Result: Confirmed and exceeded. Management stated HBM4 12-high volume ramp is tracking twice as fast as HBM3E 12-high. Over $1 billion in HBM4 revenue already shipped. Management stated it expects to reach mature yields “significantly faster than HBM3E 12-high.”
  Status: HIT — Strengthened language from prior quarter
  Context: The word “significantly” was added to the yield ramp claim this quarter versus “faster” from the prior scorecard. This is a notable language upgrade. The $1B+ HBM4 revenue milestone in what is still early in the ramp confirms volume is real, not just sampling.

• Milestone 7 — HBM4E on track for CY2027 volume ramp on 1-gamma with TSMC logic die
  Result: Confirmed. Sanjay stated next-generation DRAM and NAND nodes are on track to begin volume production in second half of CY2027. Mark Murphy added that the HBM TAM is now expected to cross $100 billion in CY2027, pulled forward from CY2028.
  Status: HIT — Demand context significantly upgraded
  Context: The HBM TAM acceleration from $100B in 2028 to $100B in 2027 materially improves the demand backdrop for HBM4E’s launch year. Sumit Sadana confirmed demand for HBM across 2027 and 2028 “is far in excess of our ability to support.”

• Milestone 8 — HBM4 16-high (48GB) development proceeds without delays
  Result: No specific update provided in this transcript. Management did not explicitly address the 16-high product timeline.
  Status: Pending — no update
  Context: Absence of mention is not a negative signal given the breadth of announcements on this call. Watch for a specific update on the 48GB product on the Q4 call.

• Milestone 9 — HBM market share maintained broadly in line with overall DRAM share
  Result: Confirmed. Sumit Sadana explicitly restated: “Our goal is to have our HBM share consistent over time with our DRAM share.” This is a strategic choice, not a supply constraint.
  Status: HIT — Explicitly reaffirmed
  Context: The framing is important — management is choosing to cap HBM share at DRAM share levels to protect non-HBM supply for diversified end markets. This is a deliberate portfolio allocation decision, not a competitive limitation. Watching for any deviation from this framing.

CATEGORY 3: TECHNOLOGY NODE EXECUTION

• Milestone 10 — 1-gamma crossover to DRAM majority mix by mid-CY2026
  Result: Confirmed on track. Management stated 1-gamma is “ramping well and on track to become the highest volume node in Micron’s history.”
  Status: HIT (on track confirmation)
  Context: Manish Bhatia at the JPMorgan conference had already confirmed this was on schedule; this call provides a further reaffirmation one quarter later with no language softening.

• Milestone 11 — G9 NAND crossover to majority bit mix by mid-CY2026
  Result: Confirmed on track. Management stated G9 is “ramping well and on track to become the highest volume node in Micron’s history.”
  Status: HIT (on track confirmation)
  Context: The data center SSD revenue of $5B+ in Q3, more than doubling sequentially, is the most direct evidence of G9 ramp execution bearing commercial fruit. The PCIe Gen6 platform based on G9 is the primary driver of Micron’s record SSD share gains.

• Milestone 12 — Multiyear EUV supply agreement with ASML concluded
  Result: Confirmed explicitly in prepared remarks. “We recently concluded a multiyear EUV supply agreement with ASML, supporting our increased adoption of EUV at the 1-delta node and future generations.”
  Status: HIT — Explicitly confirmed in Q3 prepared remarks
  Context: This is now a signed agreement rather than a “recently concluded” item from the prior conference. Locks in tool availability for the 1-delta node transition and beyond.

CATEGORY 4: SUPPLY CHAIN & MANUFACTURING CAPACITY

• Milestone 13 — FY2026 CapEx exceeds $25B, with Q3 CapEx ~$7B
  Result: Q3 CapEx was $7.1B (in line with guidance). Full year FY2026 CapEx now guided at ~$27B. Q4 CapEx guided at ~$10B.
  Status: HIT — Updated upward
  Context: The $27B full-year number is above the prior “above $25B” guidance. More significantly, Mark Murphy confirmed FY2027 CapEx will be above the mid-40s range that had been implied by prior guidance, and Micron is exiting FY2026 at a $10B quarterly run rate that will “step up from there.” This is a substantial capital commitment increase driven by demand confidence.

• Milestone 14 — Tongluo begins meaningful product shipments in FY2028; second cleanroom construction begins by end of FY2026
  Result: Pulled in. Management stated Tongluo will now support meaningful product shipments in mid-CY2027, approximately one quarter ahead of prior guidance. Second cleanroom construction has begun.
  Status: HIT — Ahead of schedule on both items
  Context: Management explicitly cited the timeline acceleration — “about a quarter earlier than our prior expectations.” The second cleanroom at Tongluo will support EUV equipment, which is an incremental detail not previously disclosed.

• Milestone 15 — Idaho Fab 1 wafer output by mid-CY2027; Idaho Fab 2 ground preparation underway targeting late CY2028
  Result: Confirmed. ID1 “on track for first wafer output in mid-calendar 2027.” ID2 “in late calendar 2028.”
  Status: HIT — On track, consistent with prior guidance
  Context: No timeline change here. Management confirmed both milestones are proceeding as planned with construction “well underway.”

• Milestone 16 — Singapore HBM packaging facility contributes meaningfully to HBM supply beginning CY2027
  Result: Upgraded. Management now stated this facility “will contribute meaningfully to Micron’s HBM packaging capacity beginning in the first half of calendar year 2027,” versus the prior “calendar year 2027” language.
  Status: HIT — Timeline tightened and front-half CY2027 specified
  Context: The “first half” specification is a meaningful upgrade from the vague CY2027 prior statement. Given HBM demand far exceeds supply, any acceleration in packaging capacity is directly thesis-additive.

• Milestone 17 — New Singapore NAND fab achieves initial wafer output in H2 CY2028
  Result: Management stated “construction activities and time lines are on track for our other facilities in Japan and Singapore.” No specific update on the NAND fab timeline.
  Status: Pending — on track, no change
  Context: No news is consistent with on-track status. This is a longer-dated milestone and is not a near-term signal.

CATEGORY 5: STRATEGIC CUSTOMER AGREEMENTS

• Milestone 18 — Additional SCAs beyond the first 5-year agreement
  Result: 16 SCAs announced, versus 1 at the time of the prior scorecard. This is the single most consequential scorecard update this quarter.
  Status: HIT — Massively exceeded
  Context: The scale, structure, and financial commitments disclosed on this call represent a fundamental business model transformation announcement, not an incremental update. Key terms: 5-year agreements from CY2026 through CY2030; take-or-pay with no cancellation provisions; price bands (floor and ceiling) with ceiling at current CQ2 market prices; $100B minimum cumulative RPO across 14 of 16 agreements; $22B in customer cash deposits and financial commitments ($18B in cash); coverage of approximately 20% of DRAM volume and 33% of NAND volume currently, with a target of approximately 50%+ of total company revenue when all planned SCAs are executed; four large customers and three medium customers across data center, consumer, and automotive.

• Milestone 19 — SCA discussions extend across multiple markets including NAND/SSD
  Result: Confirmed and exceeded. SCAs explicitly cover DRAM, HBM, and NAND. Sumit confirmed that for hyperscaler SCAs where HBM is required, it is included in the agreement. NAND is described as constrained and in demand, with customers motivated to secure supply across both product categories.
  Status: HIT — Explicitly confirmed across all major product categories
  Context: The breadth of the SCA framework across DRAM, HBM, and NAND simultaneously removes the prior uncertainty about whether NAND/SSD would be included. This is a thesis extension, not just a confirmation.

CATEGORY 6: DEMAND ENVIRONMENT

• Milestone 20 — Key customers continue receiving only 50%–67% of bit demand in medium term
  Result: Sumit Sadana stated that for some customers, “supply numbers are a fraction of what they want,” and that aggregate supply is “substantially below” aggregate demand. The 50%–67% framing was not repeated verbatim but the underlying condition is clearly intact or worsened.
  Status: HIT — Supply constraint thesis intact or deepened
  Context: Management shifted from a specific percentage range to qualitative language suggesting the gap may be wider in some cases (”a fraction of what they want”). This is not a softening — it reads as a strengthening of the supply constraint narrative. No CUT signal here.

• Milestone 21 — Data center DRAM and NAND bits exceed 50% of industry TAM in CY2026
  Result: Management stated data center DRAM and NAND bit shipments in CY2026 are “expected to more than double from two years ago.” The annualized data center revenue run rate is described as exceeding $100 billion. The 50%+ TAM thesis is implicitly reinforced.
  Status: HIT — Consistent with prior guidance
  Context: The annualized $100B+ data center revenue figure for Micron alone is the most powerful validation of the data center TAM inflection thesis.

• Milestone 22 — Supply-demand tightness language sustained as “well beyond calendar 2026”
  Result: Language explicitly upgraded. Prior: “well beyond calendar 2026.” New: “we expect supply-demand conditions for both DRAM and NAND to remain tight beyond calendar 2027.”
  Status: HIT — Thesis Anchor upgraded materially
  Context: This is the most important single data point in the entire scorecard update. The supply-demand tightness horizon has been extended by at least a full year. Management added that “even as we expect industry supply to improve gradually in 2028, we currently do not have line of sight as to when memory supply will be able to catch up with increasing demand.” This is a significant strengthening of the structural thesis.

• Milestone 23 — AEBU revenue sustains at record levels, automotive plus industrial exceeding $2B per quarter
  Result: AEBU revenue was a record $4.6B, up 71% sequentially. AEBU gross margins reached 79%, up 11 percentage points sequentially.
  Status: HIT — Significantly exceeded
  Context: $4.6B in a single quarter is more than double the prior $2B threshold. Management cited L2+ and above vehicle mix doubling in CY2026 to over 20% of total vehicles, with a trajectory toward 40%+ by 2030. Humanoid robot memory content commentary reinforces the long-duration thesis for AEBU.

• Milestone 24 — PC and smartphone unit declines contained to low-double-digit range, content per device growing
  Result: Management now says PC and smartphone industry revenue is expected to grow despite unit volume declines, “reflecting resilient demand for high-end devices at higher prices.” Unit decline assumption is broadly consistent with prior guidance; the framing has shifted more constructively to revenue growth despite unit pressure.
  Status: HIT — In line with guidance, framing upgraded
  Context: The shift from discussing unit declines as a risk to citing industry revenue growth as an offset suggests ASP mix is running better than feared. Agentic AI device platforms (OpenClaw, NemoClaw) are cited as the driver of content demand at the high end.

CATEGORY 7: CAPITAL ALLOCATION & BALANCE SHEET

• Milestone 25 — Net cash grows from Q2’s $6.5B record; FCF roughly doubles sequentially
  Result: Net cash reached $24.4B, versus $6.5B at Q2 end — a nearly 4x increase in a single quarter. All three major credit rating agencies upgraded Micron this year, including a BBB+ upgrade in Q3. Dividend confirmed. Share repurchase capacity expanding post December 2026 CHIPS Act anniversary, with commitment to return 100% of excess cash.
  Status: HIT — Exceeded materially
  Context: The $24.4B net cash position combined with projected Q4 FCF of approximately $25B–$30B means Micron could exit FY2026 with a net cash balance exceeding $40B. Mark Murphy confirmed plans to increase capital returns from December 9, 2026 onward. This balance sheet position is without precedent for a semiconductor company of this profile.

TRANCHE SIGNAL UPDATE

Based on this scorecard: ADD

The combination of zero Critical misses, multiple Critical milestone beats, the Thesis Anchor language strengthening from “beyond 2026” to “beyond 2027,” and the SCA announcement collectively satisfy the ADD rule as defined: 5 of 6 ADD conditions met with no Critical misses. All 6 ADD conditions are met.

The only monitoring items remaining are: HBM4 16-high product update, the pace of additional SCAs beyond the current 16, and whether Q4’s 86% gross margin guide holds or expands further when reported.

This scorecard is for informational and analytical purposes only. It does not constitute financial advice, a recommendation to buy or sell any security, or an investment prospectus. Always conduct independent due diligence and consult a licensed financial advisor before making investment decisions.

For more high-conviction earnings breakdowns and management accountability frameworks, follow along for future scorecards.

Subscribe now

Last quarter, the company posted the largest single-quarter revenue jump in its history, guided to 81% gross margin, and confirmed that key customers are still receiving only half to two-thirds of their bit demand.

The bull case is extraordinary here, but of course, so is the execution bar required to sustain it.

Here is the cheat sheet to watch when going through the earnings.

Please note that I have also included a tranch system of when I would add, trim or hold, but of course that’s my interpretation of things, and you should do your own research when making such decisions.

What makes these earnings so critical is that the win may already be priced in. This is why future guidance will be more important this time than the actual performance.

The entire valuation case rests on duration, not magnitude. If supply-demand tightness extends well into 2027 and beyond, then Q3's 81% margins may be the highest Micron would ever achieve.

If guidance language softens even modestly, the market will immediately start pricing in cycle mean-reversion, and the multiple compresses fast regardless of what the Q3 print says.

Immediately after the earnings drop, I'll be publishing a full post-earnings diagnostic, milestone by milestone, and what it means for the position.

Make sure you’re subscribed so you don’t miss out!

Subscribe now

However, one very interesting company not many people know about is Adeia Inc. (ADEA). This company is absolutely critical in the memory space, as it owns the patents that make next-generation chips possible, and it honestly seems that the industry has no choice but to license them.

Let me explain why.

Traditional chip scaling is hitting a physical wall. So, increasingly, we are seeing the industry moving towards stacking chips together using hybrid bonding, advanced packaging, and chiplet architectures.

All three of these (especially hybrid bonding) sit squarely inside Adeia’s patent portfolio.

The company has already signed royalty agreements with AMD, SanDisk and Kioxia, etc.

Secondly, AI is a massive accelerant. AI workloads demand more compute density, more memory bandwidth, and more thermal headroom than any prior computing cycle. Each of these constraints are pushing the industry deeper towards more sophisticated memory hardware.

The company seems cheap, trading at 22x forward earnings.

In this deep dive,

Welcome to my deep dive into ADEA

COMPANY OVERVIEW

• Company: Adeia Inc.

• Ticker: ADEA (NYSE)

• Hard Guidance on Record:

  • Full-year revenue $395M–$435M (reiterated);

  • adjusted EBITDA margin ~55%;

  • OpEx $184M–$192M;

  • interest expense $34M–$36M;

  • other income $5.5M–$6.5M;

  • CapEx ~$2M; recurring revenue ~$90M by Q4 2026

• Notable Event: CEO Paul Davis announced intent to step down later in 2026; successor search underway targeting Q4 appointment

MILESTONE SCORECARD

Image form of the cheat sheet can be found here for free download.

CATEGORY: Revenue & Guidance

• Milestone 1 — Full-year revenue lands within $395M–$435M guidance range
  Weight: Critical | Signal: ADD
  Context: Management reiterated this range after a Q1 beat driven partly by AMD retroactive catch-up. A full-year miss would call into question both the pipeline robustness and the quality of Q1 results. Hit means the new-deal engine is performing; miss raises questions about DISH/DIRECTV drag and pipeline conversion.

• Milestone 2 — Quarterly recurring revenue reaches approximately $90M by Q4 2026
  Weight: High | Signal: ADD
  Context: Q1 recurring was $66.3M due to SanDisk/Kioxia timing and subscriber declines. CFO guided recurring to grow sequentially each quarter toward $90M by year-end. This is the clearest near-term ramp signal; a shortfall in Q2 or Q3 would flag execution risk on the revenue recovery narrative.

• Milestone 3 — Non-Pay-TV recurring revenue growth sustains 20%+ year-over-year
  Weight: High | Signal: ADD
  Context: Q1 printed 28% YoY growth in non-Pay-TV recurring, continuing a 4–5 quarter trend. This metric is management’s proof point for diversification. Sustained 20%+ tells you the pivot away from Pay-TV dependency is structural, not transient.

• Milestone 4 — Full-year adjusted EBITDA margin sustains approximately 55%
  Weight: High | Signal: HOLD
  Context: Q1 margin was 60%, elevated partly by variable comp seasonality working in their favor. Full-year guidance of ~55% implies compression as litigation spending and variable comp normalize. Sustained margin well above 55% is upside; a dip below 50% would flag cost discipline concerns.

• Milestone 5 — Q2 2026 revenue is modestly lower than Q1’s $104.8M (as guided)
  Weight: Medium | Signal: HOLD
  Context: Management explicitly guided Q2 slightly below Q1 and H1/H2 roughly equal. Any material downside surprise beyond “modest” would suggest deal pipeline slippage or execution gaps beyond known seasonality.

CATEGORY: Semiconductor Licensing

• Milestone 6 — AMD becomes a recurring greater-than-10% customer in subsequent quarters
  Weight: Critical | Signal: ADD
  Context: CFO flagged AMD as likely a >10% customer going forward (net of retroactive component). CEO corrected this intra-call, suggesting the ongoing run-rate may fall below 10%. If AMD becomes a sustained >10% customer, it validates the semiconductor thesis at scale. If the CEO’s correction proves correct and AMD normalizes below 10%, it means the big Q1 number was mostly retroactive — structurally important but less recurring than it appeared.

• Milestone 7 — At least one additional logic or memory semiconductor licensee signed in 2026
  Weight: Critical | Signal: ADD
  Context: The AMD deal was framed as validating a broader semiconductor licensing opportunity in both logic and memory. If no additional semiconductor deals close in 2026, the thesis that AMD opens a category door — rather than being a one-off — remains unproven. This is the single most important new-deal milestone for the year.

• Milestone 8 — SanDisk and Kioxia revenue contribution resumes in Q2 and Q3 as guided
  Weight: High | Signal: ADD
  Context: Both contributed zero revenue in Q1 due to agreement structure timing. Management stated they will contribute “meaningful revenue in following quarters.” A resumption in Q2 is a clean, testable milestone. A miss would suggest agreement structures are more complex than guided or that there are undisclosed renegotiation dynamics.

• Milestone 9 — Hybrid bonding adoption expands into consumer device applications beyond data centers
  Weight: Medium | Signal: ADD
  Context: Management cited “early indication” of hybrid bonding expanding into consumer electronics — framed as a long-term, high-volume optionality event. This is a soft milestone; track for language escalation from “early indication” to “active pipeline” or “signed agreements” in future calls.

• Milestone 10 — RapidCool thermal management technology advances from 5W/mm² to a named commercialization milestone
  Weight: Medium | Signal: ADD
  Context: Management upgraded RapidCool cooling capability to ~5W/mm² (up from 3W) and cited growing partner interest. No commercial agreement is yet in place. Watch for a licensing agreement or named partner announcement as the upgrade trigger. Continued “partner interest” language with no signed deal is a neutral signal, not a hit.

CATEGORY: Dispute Resolution

• Milestone 11 — DISH Network dispute resolved with a signed license agreement
  Weight: High | Signal: ADD
  Context: DISH’s agreement expired at end of March and they are now operating unlicensed. Management cited the Disney and AMD precedents (resolved within 4 months) as the model for resolution speed. A DISH settlement removes a known revenue headwind and validates the litigation-to-license playbook. Prolonged litigation extending beyond 2026 would create revenue uncertainty and elevated litigation cost drag.

• Milestone 12 — DIRECTV dispute resolved with a signed license agreement
  Weight: High | Signal: ADD
  Context: Mentioned alongside DISH as an open resolution target. Given DIRECTV’s scale in Pay-TV, a successful renewal would be a material revenue contributor and affirm that the legacy Pay-TV base is not eroding faster than expected. Track whether management updates confidence language on this specifically.

• Milestone 13 — Litigation expense remains within $20M–$26M annualized range (implied by Q1 run-rate)
  Weight: Medium | Signal: HOLD
  Context: Q1 litigation was $6M. With DISH now active and DIRECTV open, litigation costs could escalate meaningfully. A significant step-up beyond this range, without corresponding resolution revenue, would compress margins and signal a drawn-out dispute cycle rather than the “efficient resolution” model management projects.

CATEGORY: Portfolio Development

• Milestone 14 — Patent portfolio grows from 13,750 to over 14,500 assets by year-end 2026
  Weight: Low | Signal: HOLD
  Context: Management flagged that portfolio growth will moderate from recent double-digit rates. This milestone tracks whether the deceleration is managed or abrupt. Portfolio depth is a lagging indicator of licensing cycles but matters for the long-term $500M revenue target.

• Milestone 15 — 85/15 internal-to-external innovation split maintained across full year 2026
  Weight: Low | Signal: HOLD
  Context: Management cited this as a deliberate ratio. A sustained shift toward external acquisitions would signal either internal R&D productivity concerns or an acceleration of inorganic strategy — both worth tracking for what they imply about portfolio quality and cost structure.

• Milestone 16 — E-commerce licensing pipeline converts to at least two new customer agreements in 2026
  Weight: Medium | Signal: ADD
  Context: L’Oréal was added as a new e-commerce customer but management acknowledged it remains a small revenue contributor. The qualifier “could be much more significant” signals optionality, not certainty. Two or more new e-commerce agreements in 2026 would validate this as a real revenue diversification vector.

CATEGORY: Capital Allocation

• Milestone 17 — Term loan balance reduced below $350M by year-end 2026
  Weight: Medium | Signal: HOLD
  Context: Q1 ended at $398.6M after $28.1M in payments. Management has targeted the $300M–$400M zone as a comfortable long-term carry. Continued deleveraging sustains the credit rating upgrade trajectory and improves optionality for debt refinancing ahead of the June 2028 maturity.

• Milestone 18 — Debt refinancing process initiated at least 12 months before June 2028 maturity (by June 2027)
  Weight: High | Signal: HOLD
  Context: CFO explicitly stated this is their target timeline. Failure to initiate by mid-2027 would leave them exposed to market conditions in a compressed window. Given that CFO noted current rates are “not optimal,” watch for any language around fixed-rate structures in H2 2026.

• Milestone 19 — Share repurchase program remains active with meaningful buyback volume in 2026
  Weight: Low | Signal: HOLD
  Context: $150M remaining under the current authorization. Q1 repurchased $10M. Continuity of buybacks signals management confidence in cash generation. Any suspension would be a sentiment flag, though it could also reflect capital reallocation toward larger acquisitions.

CATEGORY: CEO Succession

• Milestone 20 — Successor CEO identified and appointed by Q4 2026 as committed
  Weight: Critical | Signal: HOLD
  Context: Paul Davis announced his departure with a target Q4 handoff. Leadership transitions at IP licensing companies carry real execution risk — relationship continuity with licensees and pipeline management depend heavily on key executives. A prolonged search beyond Q4, or appointment of an external candidate without IP licensing background, warrants heightened scrutiny. The board has engaged a nationally recognized search firm, which is a process positive.

TRANCHE DECISION FRAMEWORK

ADD — Add to position
Condition: Any 4 of these 5 conditions met, with zero Critical misses:

1. Milestone #7 confirmed: at least one additional semiconductor (logic or memory) licensee signed in 2026 (this is the single most important new-deal test of the thesis)

2. Milestone #2 confirmed: recurring revenue on track to reach ~$90M in Q4 (validated at Q2 or Q3 check-in)

3. Milestone #11 or #12 confirmed: DISH or DIRECTV resolved within 2026

4. Milestone #3 confirmed: non-Pay-TV recurring revenue growth sustains above 20% YoY

5. Milestone #20 confirmed: successor CEO named with IP licensing or technology M&A background, no extended transition disruption

HOLD — Maintain current position
Condition: Thesis intact but at least one execution uncertainty unresolved:

1. AMD proves to be a one-time retroactive contributor rather than a recurring run-rate driver, but no Critical milestones are outright missed

2. DISH and DIRECTV disputes extend into 2027 without resolution, elevating litigation cost drag, but recurring revenue ramp remains on track

3. CEO succession extends beyond Q4 2026 but an interim plan is communicated without disruption to pipeline deals

4. No new semiconductor deal signed by Q2 end but management maintains confidence language and pipeline activity is visible

5. RapidCool interest from partners remains verbal with no signed agreement, but semiconductor licensing milestones are hitting

CUT — Reduce or exit position
Condition: Any 2 of the following trip wires, or any single starred item:
★ Full-year revenue guidance is revised down below the $395M floor — signals pipeline conversion failure or deal-timing deterioration beyond explained factors
★ Management explicitly walks back the $500M long-term licensing target or decouples it from any defined timeline
— No new semiconductor licensee signed by Q3 2026 AND management softens language from “other leading logic and memory companies are following similar paths” to conditional or uncertain framing
— Recurring revenue trajectory fails to recover toward $90M by Q3 — SanDisk/Kioxia contribution underwhelms and Pay-TV subscriber decline accelerates
— CEO departure triggers visible customer relationship disruption or replacement hire has no IP licensing background
— DISH and DIRECTV both remain unresolved entering 2027 AND litigation expense runs above $30M annualized without offsetting deal economics

This scorecard is for informational purposes only and does not constitute investment advice. All milestone assessments are based solely on publicly available management commentary. Consult a qualified financial advisor before making any investment decisions.

Follow for more high-conviction earnings breakdowns.

A Quick Summary of Results

Oracle closed fiscal 2026 with a quarter that, on the headline numbers, validated nearly everything management has been promising through this AI infrastructure supercycle.

• Q4 revenue came in at $19.2 billion, up 21% in US dollars, pushing full-year revenue past $67 billion for the first time in company history.

• Cloud infrastructure grew 93%

• Cloud applications held double-digit growth at 10%

• Most importantly, remaining performance obligations (RPO) exploded to $638 billion, up 363% year over year.

The quarter also marked a leadership transition in the finance function. Hilary Maxson, two months into the CFO seat, used her debut call to fully reconfirm the long-term targets set at the October Analyst Day: 31% revenue CAGR and 28% EPS CAGR through fiscal 2030.

A new CFO re-underwriting her predecessor’s most aggressive commitments is not a small signal.

FY27 guidance of +34% constant-currency revenue growth actually runs ahead of that five-year CAGR, with non-GAAP EPS guided to $8.05.

Beneath the headlines, we seee a lot more nuance. Gross margins are compressing as data centers ramp, growth rates across the hypergrowth segments are decelerating off extreme bases, and a few of last quarter’s most concrete promises seem to have gone unmentioned.

Don’t get me wrong, none of it actually breaks the thesis, but it is worth mentioning nonetheless.

Scorecard: Where They Delivered (and Missed)

Against the explicit commitments made on the Q3 call, Oracle cleared most of the hard hurdles:

• The promised FY27 capital expenditure guidance arrived on schedule: roughly $70 billion in net cash outlay, with reported CapEx running $20–25 billion higher due to customer prepayments.

• The commitment to raise no additional debt in calendar 2026 was explicitly reaffirmed, as was the pledge to preserve the investment-grade credit rating.

• The 30–40% AI infrastructure gross margin framework was maintained, now supplemented with a new disclosure: steady-state return on invested capital in the high 20s at the project level.

• Most impressively, the alternative funding model Oracle introduced just two quarters ago has scaled dramatically.

• “Bring-your-own-hardware” and prepaid contracts went from $29 billion signed in Q3 to $67 billion in new AI contracts signed in Q4 alone. This brought the cumulative total to $75 billion, reportedly with no margin degradation.

• Physical delivery also tracked well: 1.2 gigawatts delivered across FY26, key sites like Shackleford running a month ahead of schedule on power, and Q1 FY27 delivery alone approaching 1 gigawatt — nearly matching the prior four quarters combined.

Despite these strengths, the following areas, which seem less flashy, are also worth mentioning:

• Gross margin fell roughly 5 points for the year and is guided to step down again in FY27. Understandable given the infrastructure ramp-up.

• Cloud apps growth ticked down from 11% to 10%, and the Fusion product-line breakdown disclosed in Q3 disappeared. A one-point deceleration is not too serious, but this is something to watch over time.

• Multicloud database growth decelerated from 531% to a still-extraordinary 404%. Law of large numbers, and not really weakness.

• The TikTok equity income that was supposed to land in Q4 results also went unmentioned. Likely immaterial, and should be found in the 4k.

• Roughly 10% of reported full-year EPS came from one-time investment gains (Ampere, Bloom Energy). This could lead to a high comp to match in future earning periods.

Uncovering The Market Opportunity

I think the management has really made it clear as to where they see this market heading towards. This direction is worth taking seriously because Oracle sits at an unusual vantage point as both a major AI infrastructure provider and one of the largest enterprise consumers of AI tooling.

First of all, the market is bigger than the cloud market it succeeds.

Clay Magouyrk stated plainly that AI infrastructure “makes the existing cloud infrastructure market look small,” sizing it at trillions of dollars per year.

Several years into the buildout, demand still massively exceeds supply, and management expects more entrants — neoclouds, sovereign players, even SpaceX building data centers — without viewing that as a margin threat in the medium term.

Another, very clear theme seems to be that inference (and not training) is now the engine. The center of gravity really is shifting. Before the industry norm was a concentrated set of model-training contracts. Now it’s broadly distributed inference demand. Within this shift, agentic coding is the most visible and voracious use case.

Critically, management argued that inference latency is actually governed by accelerator architecture rather than data center geography. What this basically means is that hyperscalers can essentially keep siting capacity where power and land are cheap rather than near population centers.

This seems like a pretty clear structural cost advantage thesis.

The customer conversation, per Oracle’s CEO, is no longer “should we use AI” but “how quick can we deploy enterprise-grade agentic systems” and “how measurable is the ROI”.

The decisive raw material is private operational data — decades of it — and most of it has never left the database it was born in. The companies that control where that data lives control the on-ramp to enterprise AI.

Finally, there also seems to be this implication, from management that the SaaS-pocalypse thesis is dying. To be fair, they did acknowledge that fears of AI killing application software caused some delayed purchase decisions a couple of quarters ago, but they report that their customers have moved on, especially in mission-critical systems, where nobody is replacing a core banking platform or electronic health record with a “cobbling together of niche AI features.”

Why Oracle Believes the Structure Favors Them

The argument for Oracle’s position links each of those market dynamics to a structural asset.

The full stack is the moat. Oracle is the only vendor selling the AI data centers, the general-purpose cloud around them, the database holding the private data, the SaaS applications generating that data, and the agent development tooling on top — and roughly 10–20% of AI data center spend flows to higher-margin adjacent services that pure-play GPU landlords cannot capture.

Data gravity does the customer acquisition.

The most valuable inference target is decades of operational data already sitting in Oracle databases. Therefore, every AI initiative starts on Oracle’s home turf, and the multicloud partnerships with Microsoft, Google, and AWS extend that gravity into competitors’ clouds.

The contract architecture protects the economics. They have a strong structure with fixed pricing only where costs are locked, and floating mechanisms where they are not. This is why management says memory and component inflation has not dented margins, while bring-your-own-hardware and prepaid structures push capital intensity onto customers at equal or better margins.

In fact, this monetization model seems to be maturing in real time.

The “AI at no additional cost” refrain has evolved into a tiered structure of free embedded agents, paid token bundles (33 customers in the initial rollout), and outcome-based pricing tied to measurable results like candidates screened or patient throughput.

This shift from giveaway to monetization may matter a lot to the long-term apps story.

Interestingly enough, the bear case is also easy to spot

• Two consecutive years of gross margin compression

• EPS growth flattered by one-time gains

• Hypergrowth segments decelerating

• Renewal disclosure showing only 49% of customers with expiring GPU contracts renewed (albeit covering 92% of the GPUs, with 97.5% fleet utilization absorbing the rest).

Overall, the machine does seem to be working to me. But we need to watch if it becomes more expensive and more complicated than the management is suggesting.

What We’re Tracking Into FY27

These are the specific, checkable items we will score on upcoming calls, starting with Q1 FY27 results on September 10 and the Investor Day on October 28 in Las Vegas.

Make sure you’ve subscribed so you can read our follow-up scorecards on these dates.

Subscribe now

1. Q1 FY27 guidance vs. actuals

   Total revenue growth of 27–29% in USD, cloud revenue growth of 58–64%, and non-GAAP EPS of $1.72–$1.76.

   Management also guided to acceleration in the back half of FY27 as new megawatts come online; the H2 ramp is the claim to test.

2. Gross margin trajectory

   Management says infrastructure margins “improve rapidly” as data centers reach full contractual revenue. FY27 is guided down; the inflection point needs a date, so we will be paying attention to the management’s commentary about this going forward.

3. Site delivery milestones

   Abilene’s remaining ~58% of capacity (35% due within 90 days of the call), and first customer deliveries at Shackleford and Doña Ana in H1 calendar 2027, Saline and Port Washington in H2 2027.

4. RPO conversion

   12% of the $638 billion recognized within 12 months, 34% within 13–36 months, with both percentages promised to accelerate. Backlog is only as good as its burn rate.

5. The funding plan

   Roughly $40 billion in debt and equity in FY27 (including the $20 billion ATM), no new debt in calendar 2026, investment-grade rating intact, and the ~$70 billion net cash outlay holding.

6. GPU renewal and utilization metrics

   Whether the 49%/92% renewal split and 97.5% utilization hold as the customer base broadens, and whether smaller-customer churn becomes a pattern.

7. Token bundle and outcome-pricing adoption

   33 customers in the limited rollout; we should watch for the fleet-wide expansion and any disclosed revenue contribution.

In Conclusion,

Oracle delivered the quarter it needed and reconfirmed the five-year story under new financial leadership. The structural argument (full stack, data gravity, flexible capital models) seems to remain intact and arguably strengthened by $75 billion in customer-funded contracts.

But the scorecard discipline matters more now, not less: the gap between a $638 billion backlog and the margins it ultimately produces will be decided in the operational details management is sometimes selective about disclosing. We will keep score. Be sure to subscribe for future updates.

This newsletter is for informational and educational purposes only and does not constitute financial, investment, or legal advice.

What we do here is narrow by design: we track what management says on earnings calls and score it against what they later deliver. That is one useful lens — but it is only one.

Earnings calls are management’s version of the story; a full investment view requires the filings, the footnotes, the competitive landscape, valuation, and your own circumstances and risk tolerance. Nothing here is a recommendation to buy, sell, or hold any security. Do your own research and consult a licensed financial advisor before making investment decisions. We’re just keeping score, and sharing our insights.