The growth investor sees a business whose revenue has now been approximately flat for three consecutive years, losing share to two of the largest companies on earth, carrying $2.7 billion of term debt raised largely to fund buybacks and retire maturing convertibles, with a negative book value and a “Sell” consensus from the eight analysts still covering it.

The value investor sees a business that converts roughly 42% of revenue into cash before interest, has reduced its diluted share count from 419 million to a guided 222–227 million, just posted its third consecutive quarter of paying-user growth after years of erosion, and trades at roughly ten times what I’d call its honest owner earnings.

Both descriptions are accurate. That is what makes it worth the page count.

This piece does not tell you which one to believe. It lays out the arithmetic on both sides, flags where management’s preferred metrics diverge from the cash a shareholder can actually claim, and identifies the specific, dated events that will resolve the disagreement. Where I’ve done my own math rather than quoted the company, I say so.

Looking for unbiased, fact-based news? Join 1440 today

Tired of getting your news from opinionated, biased news sources? Skip the opinions and get just the facts in a single, daily briefing. Find topics on politics, business, science, sports, culture, and more. Join 1440 now!

The snapshot

• Price (Aug 11, 2026, intraday): ~$34.70

• Shares outstanding (Aug 3, 2026): 217.4M, excluding unvested Co-Founder RSAs

• Market capitalization: ~$7.5B

• Net debt including finance leases: ~$2.5B

• Enterprise value: ~$9.8B on financial debt only, ~$10.4B including all lease obligations

• TTM revenue: $2.53B

• FY2026 revenue guidance: $2.513–2.523B

• FY2026 unlevered FCF guidance: at least $1.070B

• 52-week range: roughly $21.70–$35.72

• 52-week price change: +28.7%

• Short interest: 28.6M shares, or 13.2% of shares out and 21.5% of float

• Analyst consensus: Sell, average target $30.67, 8 analysts

Note those last two lines together. The stock trades roughly 13% above the average published price target, near a 52-week high, with more than a fifth of its float sold short. Whatever else this is, it is not a consensus long.

What the business actually is

Dropbox sells subscriptions to a content collaboration platform — file sync and share at its core, plus DocSend, Dropbox Sign, and, since 2024, Dash. It reports as a single segment. It serves roughly 700 million registered users, of whom 18.19 million pay. That is a 2.6% conversion rate, and it is the number the bulls and bears fight over hardest: the bear reads it as a nine-figure free-rider base that has already been fully harvested, the bull reads it as a distribution funnel no competitor can replicate at zero customer acquisition cost.

Business plans run $18 per user per month (Standard) and $30 (Advanced), with Enterprise custom. The company counts roughly 575,000 paying teams. Revenue splits 56% United States, 44% international, which matters more than usual right now because FX is doing real work in the reported growth rate.

The company cites third-party data putting it at 20.9% of an $11.6 billion content-sharing and collaboration applications market, second to Microsoft at 29.4%, with Google at 16.4% and Box and Apple in the 8–9% range. Treat that framing with the skepticism any company-selected market definition deserves — it is a slide, not a filing — but the ordinal ranking is not controversial.

One operational detail from the 10-Q that rarely makes the summaries: a single distribution partner accounted for 51% of trade and other receivables as of June 30, up from 47% at year-end. That is a concentration worth watching in a business that otherwise describes itself as self-serve.

Q2: read the growth line three ways

Reported revenue was $631.5 million, up 0.9% year over year from $625.7 million. That is the headline.

Management prefers the number excluding FormSwift, the product being wound down: up 1.7%.

The company also disclosed the constant-currency figure excluding FormSwift: up 0.1%.

All three are true. The gap between the second and third is the entire currency tailwind, and it is not small — the company guided to roughly $6 million of FX benefit in Q3 alone. An investor who believes the business has inflected should be underwriting the third number, not the second, because currency is not a durable operating improvement.

That said, the operating metrics underneath moved in the right direction, and after three years of the opposite that deserves weight:

• Paying users: 18.19 million, up roughly 96,000 sequentially — the third consecutive quarter of growth and the largest quarterly addition in roughly three years.

• Teams license growth turned positive for the first time since 2024.

• ARR $2.566 billion, up 1% (up 1.7% excluding FormSwift).

• ARPU $139.68 versus $138.32 a year ago.

Management guided ARPU to decline sequentially through the balance of 2026 on FX and a mix shift toward monthly billing and lower-priced “Simple” plans. So the user count and the revenue per user are now pulling in opposite directions, which is the honest way to describe a business trading down-market to defend its subscriber base. Whether that is stabilization or a slower form of erosion is genuinely unresolved by three quarters of data.

Guidance: Q3 revenue of $627–630 million, roughly flat at the midpoint ex-FormSwift. Full year raised to $2.513–2.523 billion — which, at the midpoint, is still marginally below the $2.521 billion FY2025 result, which was itself below FY2024’s $2.548 billion.

Three years. No revenue growth. That is the fact the bull case has to survive, not explain away.

The cash: what’s real, and what management’s metric leaves out

This is where I think most write-ups on Dropbox get sloppy, in both directions.

The company’s headline cash metric is unlevered free cash flow, guided to at least $1.070 billion for 2026, up from $1.016 billion in 2025. Q2 unlevered FCF was $283.5 million. Management notes a 13% free cash flow yield against the market cap. The definition, per the company’s own footnote, is net cash provided by operating activities less capital expenditures, excluding the impact of interest payments.

Two adjustments are required before that number describes anything an equity holder can spend.

First, interest is real. The term loan is real, the coupon is floating, and the balance went up. Interest expense, net was $50.0 million in Q2 alone, versus $18.6 million a year earlier — for the half, $86.7 million versus $33.2 million. That is already net of $9.6 million of interest income in the quarter, so gross interest ran closer to $60 million. The company drew $1.2 billion on the term loan in the first half. Holding the Q2 run rate, I get roughly $190 million of net interest for the full year, and rising if rates hold and the balance doesn’t fall.

Second, capex is not $9 million a year. Reported first-half capital expenditures were $4.5 million. That number is nearly meaningless, because Dropbox funds its datacenter infrastructure through finance leases, which show up in financing activities, not investing. First-half principal payments on finance lease obligations were $78.7 million — call it $155 million annualized. The cross-check works: depreciation and amortization ran $75.5 million in the half, or roughly $151 million annualized, almost exactly matching the finance-lease outflow. That is what maintenance capital intensity actually looks like at this company.

So, my math, not the company’s:

• FY2026 unlevered FCF, as guided: ~$1,070M

• Less estimated cash interest: ~$190M

• Less finance-lease principal: ~$155M

• Levered owner earnings: ~$725M

• Per share on ~224M weighted average shares: ~$3.24

• Multiple at $34.70: ~10.7x

• Yield on market cap: ~9.6%

Nine-and-a-half percent, not thirteen. Still a genuinely low multiple for a business with 80% gross margins and negligible customer concentration on the revenue line. But the roughly 340 basis points between the marketed yield and the adjusted one is exactly the sort of thing that should be corrected before comparing this to peers.

One more line worth keeping in view: stock-based compensation was $77.3 million in Q2, roughly 12% of revenue and about $300 million annualized. It is excluded from the 40% non-GAAP operating margin and included in the 26.1% GAAP margin. Since the company retires more stock than it issues, the dilution is offset in the share count — but the cash used to offset it is buyback cash that would otherwise be shrinking the float faster. Reasonable people treat this differently. It should not be treated as zero.

The capital structure is the thesis

Here is the fact pattern, laid out plainly.

Debt. Term loan principal of $2,674.8 million as of June 30. Zero-coupon convertible notes due 2028 of $693.3 million. Finance lease obligations of roughly $286 million. Gross debt including finance leases is roughly $3.65 billion; net of $1.114 billion in cash and short-term investments, roughly $2.5 billion. The company reports that as approximately 3.2x gross and 2.2x net adjusted EBITDA. A new $400 million revolver was put in place June 1 and sat undrawn at quarter-end. During the half, Dropbox drew $1.2 billion on the term loan and repaid $695.8 million of convertible notes at maturity. Total stockholders’ deficit is $2.19 billion.

Buybacks. In the first half of 2026 the company retired 26.9 million shares, carried in the equity statement at $688.0 million, against $697.1 million of cash paid for repurchases in the period — the gap is settlement timing and excise tax. Q2 alone accounted for 12.6 million shares at roughly $315 million. Share count fell from 242.9 million at December 31 to 219.7 million at June 30, down 9.6% in six months. The board authorized an additional $900 million on June 1, leaving approximately $1.385 billion available at quarter-end, or about 18% of the current market capitalization.

The multi-year record. Diluted weighted average shares: 419M (2020), 396M, 363M, 346M, 323M, 273M (2025) — a 35% reduction. Guidance implies 222–227 million for 2026, which would be a 46% reduction from 2020. Annual repurchases over that stretch — $398M, $1,059M, $795M, $545M, $1,249M, and $1,714M — total roughly $5.8 billion from 2020 through 2025. Note the last year: $1.714 billion of buybacks against $1.016 billion of unlevered free cash flow.

Note that last comparison. Buybacks exceeded cash generation. The difference was debt. That is the structure in one sentence: Dropbox is using leverage to shrink the equity of a business with no revenue growth.

Whether that is intelligent or reckless is not a question of tone, it’s a question of one variable — the durability of the revenue base.

• If revenue is genuinely flat and the 40% margin holds, this is a well-executed LBO conducted in public. Cash flow per share compounds at roughly 8–10% annually from share count alone, with no help from the business, and the debt amortizes against a stable subscription stream.

• If revenue resumes a 3–5% annual decline, the leverage works in reverse. Fixed interest against a shrinking numerator means levered FCF falls faster than revenue, the buyback has to slow, and the compounding engine stalls precisely when the multiple would be compressing.

There is no third outcome that matters much. Everything else is detail.

Two structural items to keep on the calendar: the 2028 convertibles carry an initial conversion price of $35.35, which the stock is now sitting just beneath, with associated note hedges and warrants struck at $46.36; holders may generally convert beginning December 1, 2027. And the interest expense line will keep stepping up until either rates fall or the balance does.

Dash, and the AI question in both directions

Dash is Dropbox’s universal search and knowledge-management layer — search across connected applications, AI chat over the results, project “Stacks,” integrations spanning Google Drive, OneDrive, Gmail, Notion, Slack, Confluence, and Microsoft 365. The company sizes the search and knowledge-discovery software market at $6.5 billion in 2023 growing to $21.6 billion by 2028.

The case that AI helps. Management disclosed that more than 150,000 users are connecting to ChatGPT and Claude integrations through Dropbox, and framed the company as the trusted, permissioned content layer that AI systems need to reason over enterprise files. Native Dash intelligence rolls out to the Teams base in the second half of 2026. Critically, no meaningful AI monetization is embedded in 2026 guidance — the strategy is to improve retention and conversion first and monetize through ARPU later. If that works, it is upside not currently in any model, including the sell-side’s.

The case that AI hurts. Dropbox’s core product is file storage and sharing, which Microsoft and Google bundle into ecosystems that also now bundle AI assistants. A standalone storage subscription competing against a bundle has structurally weaker pricing power, and the 2026 shift toward cheaper “Simple” plans and monthly billing is at least consistent with that pressure. Dash competes directly with Microsoft Copilot and Glean in a category where distribution — not technology — decides outcomes.

The cost is already visible. Gross margin declined roughly 60 basis points year over year, which management attributed specifically to increased compute costs from rolling AI capabilities out to the Teams base. So AI is currently a margin headwind with an unquantified revenue benefit. That is not a criticism; it is the normal shape of an investment cycle. But it means the 2026 margin guidance of 40.0–40.5% is being delivered despite AI spend, via a Q2 workforce reduction tied to unifying the product organization and a go-to-market rebalancing — cost actions, not operating leverage.

Management, incentives, and the transition

On May 21, 2026, the board appointed Ashraf Alkarmi — previously General Manager, Core — as co-CEO and a director, effective May 26. Co-founder Drew Houston remains co-CEO for a transition period, after which he is expected to move to Executive Chairman with Alkarmi as sole CEO. Ross Tennenbaum is CFO. The company reincorporated in Nevada in March 2025.

Two observations, offered without a verdict.

The first is that founder transitions at flat-revenue companies cut both ways. A new CEO with a product background may be more willing to make the aggressive portfolio decisions a founder finds emotionally expensive. A new CEO also typically resets expectations, and the natural moment to do that is the first guidance cycle he owns outright.

The second concerns control. The dual-class structure remains: roughly 150.6 million Class A shares and 75.1 million Class B as of August 3. Insider ownership sits near 40% per third-party data. The capital allocation policy — lever up, retire stock, prioritize FCF per share — is therefore not subject to activist redirection. If you like the policy, that’s durability. If you don’t, there’s no mechanism to change it.

What the price appears to embed

A useful discipline is to work backward from the quote rather than forward from an opinion.

At roughly $34.70 and 217 million shares, the equity is about $7.5 billion. Against my ~$725 million of levered owner earnings, a no-growth perpetuity at a 10% required return produces $7.25 billion — essentially the current market capitalization. Equivalently, a 9% discount rate paired with a 1% perpetual decline gives the same answer.

So the market is roughly pricing: flat-to-slightly-declining cash flows forever, discounted at about ten percent. Not a value trap being priced for collapse; not a franchise being priced for growth. Something close to a bond with equity risk.

Against that baseline, three illustrative paths. These are scenarios, not forecasts, and small changes to the inputs move them a lot.

Path A — decline resumes (revenue down 3% a year). Levered FCF drifts toward $600 million by 2029 as interest stays fixed. The buyback necessarily slows to preserve coverage. An 8–9x multiple on a declining stream implies roughly $22–28 per share. Worth noting: the stock traded near $21.70 within the last twelve months, so this is not a tail.

Path B — flat holds. Revenue flat, margins 40%, buyback continues at the guided pace. FCF per share compounds roughly 8–9% annually on share count alone. At an unchanged 10–12x, that’s $38–46 over two to three years, before any multiple change.

Path C — Dash converts. Revenue growth of 3–5% arrives in 2027–2028 via retention, conversion, and eventual ARPU capture. Both the numerator and the multiple move. A 14–15x multiple on $850 million-plus of levered FCF against a shrinking share count gets you to $55 and above.

The distribution is unusually wide for a company this boring. That is a function of the leverage, not the operations.

The ledger

What a buyer is underwriting

• ~9.6% levered FCF yield on my adjusted numbers, ~13% on the company’s

• Share count down 35% since 2020, guided down 46%; $1.385B authorization remaining

• Three consecutive quarters of paying-user growth; largest add in roughly three years

• Teams license growth positive for the first time since 2024

• 40.0–40.5% non-GAAP operating margin guidance, raised twice this year

• 700M registered users as a zero-cost acquisition funnel; owned co-location infrastructure holding 5,100-plus petabytes of customer content (company-reported, 2025)

• Dash and AI monetization explicitly excluded from guidance — a free option, if it works

• 13.2% short interest and a “Sell” consensus, i.e., low expectations

• New $400M undrawn revolver; convertibles at a 0% coupon

What a buyer is accepting

• Three straight years without revenue growth; FY26 guidance still below FY24 actuals

• Constant-currency, ex-FormSwift growth of 0.1% — the improvement is largely FX

• $2.5B net debt against a business with no growth; buybacks funded partly by borrowing

• Interest expense up 169% year over year; net income down despite higher revenue

• Negative book value; $2.19B stockholders’ deficit

• ARPU guided to decline sequentially on mix shift to cheaper plans

• Competing against Microsoft and Google bundles that now include AI at no incremental price

• Roughly $300M of annual stock compensation excluded from the headline margin

• 51% receivables concentration in one distribution partner

• CEO transition mid-strategy

• Stock is up 28.7% in twelve months and trades above the average published price target — you are not early

What would settle this

Rather than a verdict, here are the specific, checkable items I’d hold this to. Three or four moving the same direction should change a reasonable person’s mind.

1. Constant-currency ARR excluding FormSwift turning positive. Not reported ARR. Not ex-FormSwift. The clean number. It was +0.2% in Q2.

2. A fourth and fifth consecutive quarter of paying-user growth — Q3 results (typically early November) and Q4 (typically February). Three quarters is suggestive; five through a full annual renewal cycle is a trend.

3. ARPU stabilizing rather than continuing the guided sequential decline once FX normalizes.

4. Any disclosed Dash monetization — attach rate, incremental ARPU, or a named SKU. Management has said 2026 guidance excludes it; the first quarter it doesn’t is the signal.

5. Term loan balance falling. If the next two quarters show principal paydown rather than further draws, management is signaling it thinks the leverage is at its ceiling.

6. Gross margin recovering the 60 basis points lost to AI compute, which would indicate the rollout costs are one-time rather than structural.

7. The first guidance cycle owned solely by the new CEO — historically the moment expectations get reset in either direction.

The next scheduled data point is Q3 earnings, typically in the first half of November.

Sources

All financial data is drawn from primary filings where available:

• Form 10-Q, quarter ended June 30, 2026, filed August 7, 2026: available on EDGAR here. Source for all balance sheet, income statement, cash flow, share count, debt, lease, stock-compensation, and receivables-concentration figures.

• Form 8-K, executive transition, filed May 2026: available on EDGAR here. Source for the CEO succession terms and dates.

• Dropbox Investor Relations — Q2 2026 earnings release, investor presentation, and prepared remarks at investors.dropbox.com, with the full filing archive here. Source for guidance, unlevered free cash flow, ARR, paying users, ARPU, buyback authorization, and multi-year share count and repurchase history.

• Market data — price, shares outstanding, short interest, analyst consensus, and 52-week performance as of August 11, 2026, via stockanalysis.com.

Market share, TAM, and storage-volume figures are as presented by the company on its investor slides and reflect vendor-selected market definitions rather than audited disclosure. They are labeled as such in the text.

Where this piece departs from company-reported metrics — specifically the treatment of cash interest and finance-lease principal in calculating levered owner earnings — the adjustments and arithmetic are shown explicitly above so readers can disagree with the assumptions rather than the conclusion.

Disclaimer and important disclosures

This publication is for informational and educational purposes only. It is not investment advice, and nothing in it constitutes a recommendation, offer, or solicitation to buy or sell any security.

The author is not a registered investment adviser, broker-dealer, financial planner, or licensed securities professional, and is not registered with the U.S. Securities and Exchange Commission, FINRA, or any state securities regulator. No content in this publication is personalized to any reader’s financial situation, investment objectives, risk tolerance, tax position, or time horizon. No advisory, fiduciary, or client relationship is created by reading this publication, subscribing to it, or corresponding with the author.

Positions. The author may hold, and may at any time acquire or dispose of, long or short positions in any security discussed, including through derivatives, without notice and without updating this publication. [INSERT YOUR CURRENT POSITION IN DBX HERE, OR STATE THAT YOU HOLD NONE.]

Compensation. The author has not received, and will not receive, any compensation, consideration, or benefit from Dropbox, Inc. for the publication of this piece or for its content or conclusions.

Accuracy and sources. Information here is derived from public filings, company disclosures, earnings call transcripts, and third-party data providers believed to be reliable, but its accuracy, completeness, and timeliness are not guaranteed. Third-party data may contain errors. Readers should independently verify all figures against the company’s filings with the SEC before making any decision. Financial data becomes stale quickly; figures are as of the dates stated and are not updated thereafter.

Estimates and forward-looking statements. This publication contains estimates, scenarios, projections, and forward-looking statements that are inherently uncertain and depend on assumptions that may prove incorrect. Scenario analyses are illustrative constructions intended to frame a range of outcomes, not predictions, targets, or forecasts. Actual results will differ, potentially materially. The author undertakes no obligation to update any statement.

Risk. All investing involves risk, including the permanent and total loss of principal. Securities discussed may be volatile, illiquid, or unsuitable for any particular investor. Leverage, whether at the issuer level or the investor level, magnifies losses as well as gains. Past performance is not indicative of future results.

Your responsibility. Readers are solely responsible for their own investment decisions and should consult a licensed financial adviser, tax professional, and/or attorney regarding their specific circumstances before acting on any information contained here. To the maximum extent permitted by law, the author and this publication disclaim all liability for any direct, indirect, incidental, consequential, or other losses arising from use of or reliance on this content.

After thirty years of turning over rocks in micro-cap land, I’ve developed a reflex: when a stock doubles on the word “drones,” I don’t ask whether the story is exciting. I ask what the business earned before anyone cared, what it’s likely to earn after the excitement passes, and whether the people running it think like owners. Lantronix fails the first test, is ambiguous on the second, and fails the third — and yet I’m not prepared to dismiss it, because buried inside this perennially unprofitable roll-up is one genuinely well-positioned product line at exactly the right moment in a defense spending cycle. This memo is my attempt to separate the business from the story, and the story from the price.

The one-line summary: you are being asked to pay roughly 120 times owner earnings for a company that has never earned a GAAP profit, because a $12 million product line might become a $30 million product line next year.Sometimes that bet pays. Usually it doesn’t. Let me show you the work.

The Business, Without the Adjectives

Lantronix describes itself in its fiscal 2025 Form 10-K as a global provider of Edge AI and Industrial IoT solutions. Translated into English: it’s a fabless hardware company that puts other people’s chips onto small circuit boards and sells them, along with boxes of networking equipment, mostly through distributors, mostly on purchase orders, with no long-term contracts. Three segments:

Embedded IoT Solutions — system-on-modules built around Qualcomm (and newly MediaTek) silicon, sold to OEMs who need a computing brain inside their camera, industrial device, or drone. ASPs of $400–700 in the drone programs. This is the good business: it grew 22% year-over-year last quarter, it embeds Lantronix into a customer’s product for the life of that product (real switching costs once designed in — requalification is expensive), and it’s where the defense franchise lives.

IoT System Solutions — the legacy estate: out-of-band management appliances, switches, media converters, gateways. Competent products in mature, GDP-minus markets; the company’s own 10-K concedes that some of these products serve declining categories. This segment delivered the best year in company history in fiscal 2024 courtesy of one large European smart-grid contract, then shrank 34% in fiscal 2025 when the contract rolled off. That is not a business; that is a project.

Software & Services — a device-management SaaS platform (Percepxion) and engineering-for-hire, recently grown from ~5–6% to ~8–9% of revenue. Directionally encouraging, immaterial today.

Note the structure of the thing. Top five distributors are ~37% of revenue. Manufacturing is outsourced to Thailand, China, and Taiwan. Components include single- and sole-sourced parts. Sales cycles run 9–24 months. Every one of those facts individually is survivable; together they describe a business with limited control over its own demand, supply, and pricing. The 40–46% gross margin band — about 43% recently — is the honest signal here. Companies with real moats in this industry push higher; commodity board-stuffers sit lower. Lantronix’s margin says exactly what it is: differentiated engineering, no pricing power to speak of.

The Ten-Foot Hurdle: A Five-Year Financial Autopsy

I always start with the tape, because managements narrate and financial statements confess. The five-year record, drawn from the company’s income statements and cash flow statements, reads as follows. Revenue marched from $71.5 million in fiscal 2021 to $129.7 million, $131.2 million, and a record $160.3 million in fiscal 2024, then fell 23% to $122.9 million in fiscal 2025. The GAAP bottom line in those same five years: losses of $4.0 million, $5.4 million, $9.0 million, $4.5 million, and $11.4 million. Operating cash flow swung from +$4.3 million to –$9.4 million, +$0.2 million, +$18.6 million, and +$7.3 million. Stock-based compensation ran $3.6, $6.3, $6.2, $8.3, and $6.1 million. And the share count climbed from roughly 29 million to today’s 39.4 million.

Five observations a value investor cannot un-see:

First, the company has never earned money. Roughly $34 million of cumulative GAAP losses on $615 million of cumulative revenue. Not during the IoT boom, not during the record smart-grid year, not ever in this window. The 10-K says so plainly: a history of net losses, with no assurance of future profitability. When a company can’t earn a profit at the top of its own cycle (fiscal 2024), I ask what conditions, exactly, it is waiting for.

Second, the growth was bought, not earned. Revenue went from $71M to $160M across a string of acquisitions — Maestro, Intrinsyc, Transition Networks, Net2Edge, Uplogix, Netcomm — and the losses persisted the entire way. Serial acquisition without profit is the classic roll-up tell: each deal resets the “adjusted” numbers, adds amortization to exclude, and buys another year of top-line narrative. The fair counterpoint: the Intrinsyc deal (2020) brought the Qualcomm module capability that is now the entire bull case. Even a mediocre capital allocator occasionally buys a gem. But judge the record as a whole: tens of millions of acquired revenue, zero cumulative profit produced.

Third — and this is the crux — the “free cash flow” is a mirage manufactured by stock-based compensation.Cumulative five-year free cash flow: roughly $13.7 million. Cumulative five-year SBC: roughly $30.5 million. Read that again. The company paid its people more than twice its entire free cash flow in stock, and shareholders footed the bill through a 36% increase in share count. SBC runs ~5% of revenue and currently equals about three-quarters of trailing FCF ($6.5M of SBC against $8.5M of FCF). When I compute owner earnings the way I was taught — cash flow that could actually be distributed without impairing the franchise, treating stock comp as the very real expense it is — I get roughly $2 million a year, or about five cents per share. Against a $6.10 stock, that is ~120x owner earnings. The non-GAAP EPS the company promotes (+$0.04 last quarter versus a GAAP loss of –$0.03, per the fiscal Q3 2026 release) is precisely this shell game: add back the stock comp, add back the deal amortization, declare victory.

Fourth, the quality of the good cash-flow years is suspect. Fiscal 2024’s $18.6M of operating cash flow — the number that makes the five-year average look respectable — was flattered by a large inventory drawdown. Liquidating working capital is a one-time act of housekeeping, not earnings power.

Fifth, revenue is lumpy in a way that punishes extrapolation. A 23% top-line decline in fiscal 2025, driven by one contract ending, in a company this size, tells you the “run-rate” is a fiction. Whatever you pay for this business, do not pay a multiple of a single year.

The Balance Sheet: Genuinely Fine

Credit where due. As of March 31, 2026: $23.5M cash, $8.7M debt (paid down ~$1M in the quarter), net cash ~$14.8M, current ratio ~3, and $7.9M of operating cash flow through nine months of fiscal 2026 (see the Q3 call highlights and key statistics). No dividend, no buyback (none in FY2025). The debt is an SVB/First-Citizens facility secured by substantially all assets, with covenants and a requirement to park most U.S. cash at the bank — standard small-company terms, mildly restrictive, not dangerous at current leverage. A material weakness in IT controls surfaced in fiscal 2023 and was remediated as of June 2025; I note it and move on, though I never fully forget such things. Bottom line: this is not a balance-sheet risk story. The company can fund itself. The question is purely whether the business is worth owning.

Management: Competent Operators, Negligible Owners

The regime change matters. Saleel Awsare took over as CEO in November 2023 out of Synaptics (SVP/GM; before that, president of Conexant) and promptly rebuilt the executive bench with semiconductor people from his own orbit — the CRO from Synaptics/Conexant, the strategy chief from module-maker Telit. The CFO, Brent Stringham, is a 13-year insider and ex-E&Y auditor promoted in January 2025. Their fingerprints are visible and mostly good: costs cut, low-margin business de-emphasized, debt reduced, the drone opportunity identified early and pursued aggressively, and — a detail I genuinely respect — no executive cash bonuses paid for the lousy fiscal 2025.

Now the part I don’t respect. Per the September 2025 proxy statement, the CEO owns 1.1% of the company — about 448,000 shares including options — and every other officer and director owns well under 1%. Nearly all of it arrived via RSU grants; the recent Form 4 traffic (browsable on the SEC filings feed) is directors accruing vested RSUs, not writing personal checks. In thirty years I have learned few signals more reliable than this one: management teams that talk about “game-changing” opportunities while declining to buy a single share in the open market are telling you their honest probability estimate with their wallets. The real owners here are founder Bernhard Bruscha (13.8%) and Chain of Lakes Group (7.8%) — the founder’s continued large stake is a modest comfort; founders of thirty-five-year-old companies have watched many strategies come and go.

One more behavioral note. On the May 2026 earnings call, the CEO described the defense budget opportunity as “game-changing” money and a “logarithmic” change if it all comes through. I’ve sat through a thousand earnings calls. Operators under-promise; promoters annualize the dream. When the CEO of a company doing $12 million in drone revenue reaches for “logarithmic,” I tighten my grip on my wallet — not because he’s lying, but because promotional language is how managements pre-spend credibility they haven’t earned yet.

The Drone Thesis, Steelmanned

Fairness requires I make the bull case as strongly as its advocates would, because it is not frivolous.

Lantronix’s NDAA/TAA-compliant Qualcomm-based module was designed into Red Cat’s Teal Black Widow — a U.S. Army-approved drone — with shipments under the Army’s Short Range Reconnaissance program that began in mid-2025, per the company’s fiscal year 2025 results release. Drone OEM engagements jumped from 10 to 17 in a single quarter. Management has raised its fiscal 2026 drone guidance twice, to $10–14M (~$12M midpoint), and guided unmanned systems to 15–20% of revenue in fiscal 2027 — implying $20–30M, roughly a double. The structural tailwinds are real and rare: the FCC’s December 2025 action bars DJI and other covered Chinese makers from new-model approvals in the U.S., an explicit regulatory moat handed to compliant domestic suppliers; the proposed FY2027 defense budget carries a record ~$75 billion for unmanned and autonomous systems; a first international design win shipped to the U.K.’s Evolve Dynamics; and Ukrainian manufacturers are actively pivoting supply chains away from China toward exactly the certification profile Lantronix carries (details throughout the full Q3 transcript). Add partnerships with Unusual Machines, MediaTek, Melchioni, and Pairpoint, and a stated ambition to move up the stack from camera compute toward full autonomy and counter-drone systems, where content per drone rises.

The design-win economics are the best part of the thesis. Once a compute module is qualified into a defense program, it tends to stay for the program’s life — requalification is slow and expensive. If small drones become a consumable of modern warfare procured in the hundreds of thousands (the Ukraine lesson), the qualified American compute supplier at $400–700 a socket is holding a royalty on the category.

Now the acid. Twelve million dollars at those ASPs is perhaps 20–25 thousand modules — a foothold, not a franchise. The revenue is concentrated around one anchor customer whose own fortunes gate Lantronix’s. Federal procurement timing — which management concedes is currently slowing its core enterprise business — giveth and taketh away. The $75 billion is a proposed budget, not an appropriation. And if the market gets as big as the bulls hope, Qualcomm’s other module partners, or drone OEMs going straight to the silicon vendor, will contest every socket. Moats granted by regulation (NDAA compliance) are shared with everyone else who complies.

Valuation: Three Honest Scenarios

At ~$225M EV against $118.6M trailing revenue, you’re paying ~1.9x sales, ~27x stated FCF, and ~120x owner earnings after expensing the stock comp. Sell-side coverage is five analysts, unanimous-ish Strong Buy, $9.90 average target (visible on TipRanks and stockanalysis.com) — which, for a micro-cap, mostly measures the banks’ appetite for the company’s fees, not the security’s merits.

Here is how I’d frame intrinsic value:

Bear case (drones stall, legacy keeps leaking). Strip the story and you own a ~$105–110M revenue hardware distributor with no history of profits, worth perhaps 0.7–0.9x sales to a strategic acquirer for the customer list and the out-of-band assets. Call it $75–100M plus net cash: $2.30–2.90 per share. Down 55–65%. This is not a tail scenario; it is roughly what the stock traded at eighteen months ago, before the narrative.

Base case (drones hit guidance, nothing more). FY2027 revenue ~$135M with $25M of drones; mix improvement nudges gross margin up; the company sustains, say, $8–10M of FCF but still pays $6–7M in stock. Owner earnings of $3–5M. Even at a generous 25x owner earnings plus cash: $3.00–3.80 per share. Notice something uncomfortable: the stock is above its base-case value even if management hits its numbers. The current price already embeds success beyond fiscal 2027.

Bull case (the franchise thesis). Drones reach $50M+ by FY2028–29 as programs multiply and content per drone rises; software attach grows; operating leverage finally appears; SBC moderates to ~3% of revenue. Owner earnings of $15–20M on a defense-flavored growth multiple of 20–25x: $8.50–13.00 per share. Real, but it requires four or five things to go right in sequence, on a company whose five-year record contains zero instances of things going right on the income statement.

Weight those however you like; I can’t get an expected value meaningfully above the current price without assigning the bull case better-than-even odds — which the base rates for micro-cap defense story stocks, promotional CEOs, and 1%-ownership insiders do not support. Graham’s old line about the market being a voting machine in the short run and a weighing machine in the long run applies precisely: the vote has been cast; the weighing is scheduled for around August 20, 2026, when Q4 results and the first formal fiscal 2027 guidance arrive.

What Would Change My Mind

I keep a short list. Insiders buying stock with their own money — even $100K of open-market purchases from the CEO would out-signal every conference-call adjective. FY2027 drone guidance confirmed at or above $25M with a second named anchor customer, reducing the Red Cat concentration. The next 10-K’s customer-concentration footnote. SBC falling below 4% of revenue while FCF holds. Appropriation — not proposal — of the unmanned-systems budget. Software and services sustaining above 10% of revenue. Give me three of those six and the base case migrates toward the bull case, and I’d reconsider at almost any price below $5.

The Verdict

Lantronix is what I’d call a cigar butt with a lottery ticket stapled to it — except the market has already paid up for the lottery ticket, which defeats the purpose of both. The balance sheet is sound, the new management is operationally capable, and the drone positioning is legitimately scarce. But the price demands that a company which has destroyed capital for five consecutive years suddenly begin compounding it, that a $12M product line become the company, and that shareholders keep tolerating stock compensation that consumes the cash flow. Value investing, as I practice it, is the discipline of paying for what is and getting what might be for free. Here you are paying triple for what might be and getting what is — a break-even hardware distributor — as the consolation prize.

My action: pass at $6, watch closely, and keep it on the list. The August report is the tell. If the story cracks and the shares revisit $3 with the drone programs intact, this becomes one of the more interesting risk/rewards in micro-cap land. Patience costs nothing; this price costs plenty.

This memo is for informational purposes only and reflects one analytical framing, not investment advice or a recommendation. Primary sources: the FY2025 Form 10-K, the 2025 proxy statement, the company’s press releases, earnings call transcripts through fiscal Q3 2026, and market data as of early July 2026. Verify all figures independently; small-cap data goes stale quickly.

Read more

Through the Monday, June 22, 2026 close, Salesforce ($CRM) had fallen for 14 consecutive trading sessions — every session since June 1. According to data compiled by MarketWatch and Morningstar, that is the longest losing streak on record in the company’s history as a public company. Our own analysis of Salesforce’s full daily price record (more than 5,000 trading days back to 2006) confirms it: the prior record was eight straight down days, set only a handful of times. Fourteen stands alone. (If the stock declines again in any session after this close, the streak simply lengthens — the record is already broken either way.)

Over that streak the stock fell from a June 1 close of $209.60 to $150.12 — a decline of roughly 28% in three weeks. It now sits roughly 46% below its 52-week high of $276.80, near the low end of its 52-week range ($147.58). On an intraday basis it touched its lowest levels since early 2023.

A losing streak, by itself, tells you nothing about where the stock goes next. It is a measure of sentiment and positioning, not of value. What follows is the underlying business and balance-sheet picture — the things that actually determine whether the current price is an opportunity, a value trap, or simply fair.

What the business actually looks like

Strip away the tape, and this is not a broken company. It is a large, profitable, cash-generative software franchise growing in the high single digits.

Two things stand out:

1. Revenue growth has decelerated, but it is still growing. FY2026 revenue of $41.5B was up roughly 9.6% year-over-year, and the three-year revenue CAGR is about 9.8%. This is no longer a hypergrowth company — it is a mature-growth one. That deceleration is central to the bear case.

2. Profitability has transformed. Operating margin went from 3% in FY2023 to 20% in FY2026, and GAAP net income went from essentially breakeven ($208M) to $7.5B over the same span. The margin expansion story — driven by cost discipline after years of activist pressure — is real and is reflected in the financials.

Cash generation is the strongest part of the picture. In FY2026 Salesforce produced $15.0B in operating cash flow. Consensus estimates put FY2027 free cash flow near $14.9B.

Capital returns: aggressive, and now debt-funded

This is where the story gets more nuanced, and where reasonable investors disagree.

In FY2026, Salesforce spent $12.6B on share buybacks and $1.6B on dividends. In February 2026 the board authorized a new, record $50 billion repurchase program. In the most recent quarter the company launched a $25 billion accelerated repurchase — half the authorization — and, critically, it is funding that $25B portion with new debt.

The bull reading: management believes the stock is materially undervalued and is using a depressed price plus cheap-relative-to-equity debt to retire shares aggressively. The buybacks helped reduce diluted share count and support EPS.

The bear reading, which must be stated honestly: funding buybacks with debt is a material shift in financial policy. The rating agencies noticed. Moody’s downgraded Salesforce one notch to A2, and S&P revised its outlook to negative, both citing the increased leverage. A buyback only creates value if the shares are genuinely cheap and the core business is intact; if the business is deteriorating, debt-funded buybacks can mask the decline rather than reverse it. That is a legitimate risk, not a hypothetical.

A note on share count: over the FY2023–FY2026 span, diluted weighted-average shares fell only about 4% (from ~997M to ~956M), because stock-based compensation issuance partly offset repurchases. The more aggressive reduction is recent, tied to the FY2026 and accelerated programs.

What insiders are doing

Insider activity is one of the cleaner signals available, so it’s worth being precise. Over the last 18 months (Jan 2025 – May 2026), across roughly 1,176 filed transactions:

• Open-market purchases (insiders buying with their own cash): 8. The largest was director G. Mason Morfit, who bought 96,000 shares at ~$260 in December 2025 (~$25M). Other directors bought smaller amounts, including purchases near $194–195 in March 2026 (David Kirk, Laura Alber) and ~$257 in April 2025.

• Open-market sales: 551, totaling roughly $118M. The largest sellers were CEO Marc Benioff (~$49.2M) and President Parker Harris (~$35.7M).

How to read this fairly: heavy insider selling at a large-cap is largely routine — executives diversify, and much of it is pre-scheduled (10b5-1 plans) or tied to option exercises and vesting. It is weak evidence of a bearish view. The buying is more informative because open-market purchases are discretionary and rare. A handful of directors put real personal money in — some near $260, well above the current $150 (i.e., they are underwater), and some near $194 in March (also above it). There has been no disclosed open-market insider buying at current levels in this dataset. Net: modest, scattered insider conviction on the buy side, dwarfed in dollar terms by routine executive selling.

Earnings: a quietly improving track record

The most recent quarter was a clear beat. For Q1 FY2027 (reported May 27, 2026), Salesforce posted EPS of $3.88 versus a consensus near $3.12 (a beat of roughly $0.76) and revenue of $11.13B, ahead of the ~$11.05B estimate. Notably, the stock still fell ~0.8% the day after — a sign the market’s concern is about the future (AI disruption, growth) more than the present (current results).

The prior several quarters show a mixed pattern on the headline EPS line (several apparent “misses” against consensus in FY2026), but those comparisons are muddied by GAAP-vs-adjusted definitional differences and one-time items. Revenue has been steadier, generally landing at or slightly above consensus. The takeaway: execution on the top line has been consistent; the debate is about durability, not about whether the company is hitting near-term numbers.

Valuation: be careful which “P/E” you use

This is the single most important thing to get right, because it’s where most takes go wrong.

• On a trailing GAAP basis, Salesforce trades at about 17.4x earnings (price $150.12 ÷ trailing GAAP EPS of $8.64). That is an earnings yield of roughly 5.8%.

• Forward estimates of “EPS” near $19 for FY2027 are non-GAAP/adjusted figures, which exclude stock-based compensation and amortization. Dividing today’s price by that adjusted number produces a ~7.9x “forward P/E” — a real figure analysts cite, but not comparable to the 17.4x trailing GAAP multiple. Mixing the two is how people accidentally claim the stock is at “8x earnings” when the GAAP reality is roughly double that.

Honest framing: on trailing GAAP earnings, ~17.4x is not extraordinarily cheap for a high-single-digit grower — it’s roughly a market multiple. On adjusted earnings and free cash flow, the stock looks cheaper, and that’s the lens the bulls (and most sell-side analysts) use. Both are valid; they are simply different yardsticks. The dividend yield is negligible (~0.01%, effectively a rounding error against the share price), so this is a buyback-and-growth story, not an income story.

What Wall Street thinks

Among the 28 analysts with current ratings, 20 are bullish, 7 neutral, 1 bearish — a consensus “Buy.” But the dispersion is enormous and is the real story:

• Average price target: ~$249 (median ~$245)

• High target: $400 (Needham)

• Low target: $173 (and one firm, Citigroup, sits at Neutral with a $187 target)

Even the lowest published target ($173) is above the current $150. That tells you the sell-side broadly views the stock as oversold relative to their models — but the spread from $173 to $400 also tells you there is no consensus on what this business is worth, which is exactly what you’d expect for a company whose terminal value hinges on an unresolved question.

The actual debate, in one paragraph each

The bull case (neutral statement of it): A profitable, $41.5B-revenue franchise generating ~$15B in annual cash flow, still growing ~9%, with expanding margins, trading ~46% off its high near multi-year lows. Management is so convinced it’s cheap that it authorized a $50B buyback and is taking on debt to accelerate it. It’s using the downturn to acquire AI capabilities — most notably a definitive agreement, signed June 15, 2026, to buy Fin (formerly Intercom) for ~$3.6 billion (expected to close around early 2027), alongside earlier-2026 deals for m3ter and Contentful — all to fold into its Agentforce agent platform. If Agentforce reaccelerates growth, today’s price looks like a gift.

The bear case (neutral statement of it): Growth has decelerated from hypergrowth to mature-growth and could slow further. The entire premise of seat-based enterprise software is under question if AI agents reduce the number of human “seats” companies need — and Salesforce’s own Agentforce push is an implicit acknowledgment of that threat. The debt-funded buyback drew a Moody’s downgrade and an S&P negative outlook, and could be masking deterioration rather than creating value (the IBM-in-the-2010s comparison). On trailing GAAP earnings (~17.4x) the stock is not screamingly cheap, and “cheap” only holds on adjusted metrics that exclude real costs like stock comp.

The bottom line

The facts are not in dispute: Salesforce just set a record losing streak, it is deeply out of favor, it is highly profitable and cash-generative, it is buying back stock aggressively with borrowed money, and its core business model faces a genuine, unresolved question about AI. Whether that adds up to a buy depends entirely on one judgment call you have to make yourself: do you believe AI is a tailwind that Salesforce can harness (Agentforce, the Fin acquisition), or a headwind that erodes its seat-based core faster than margins and buybacks can compensate?

The numbers can frame that question with precision. They cannot answer it for you.

Sources: company financial statements (FY2023–FY2026), consensus analyst data, Form 4 insider filings, and quoted market data as of the June 22, 2026 close. Acquisition, buyback, and credit-rating details corroborated by CNBC, MarketWatch, Morningstar, Reuters/Irish Times, and rating-agency commentary (Moody’s, S&P).

Disclaimer: This newsletter is for informational and educational purposes only and does not constitute investment, financial, legal, or tax advice, nor a recommendation, solicitation, or offer to buy or sell any security. The author is not a registered investment adviser or broker-dealer. All data is believed to be accurate as of the dates indicated but is provided “as is” without warranty; figures may be revised or restated, and markets move continuously. Past performance — including any losing or winning streak — is not indicative of future results. Forward estimates are consensus projections and may prove materially wrong. The author may hold positions in the securities mentioned. Investing involves risk, including the possible loss of principal. Do your own research and consult a licensed professional before making any investment decision.

What this company actually is

If the ticker CMRC threw you, you’re not alone. On August 1, 2025, BigCommerce Holdings rebranded to Commerce.com, Inc. and swapped its ticker from BIGC to CMRC. Same business, new wrapper. Commerce.com is an open SaaS commerce platform — think of it as the enterprise-friendly alternative to Shopify — and it’s now the parent of three brands: the core BigCommerce storefront engine, Feedonomics (product-feed and marketplace data management), and Makeswift (visual page building).

The pitch is “open SaaS”: rather than locking merchants into one walled garden, Commerce.com sells a composable, API-first stack that mid-market and enterprise retailers bolt onto existing systems. It’s a real, audited, Nasdaq-listed software company headquartered in Austin — not a shell, not a meme. And at a sub-$250M market cap, it’s squarely in our wheelhouse.

How Commerce.com actually makes money

To underwrite a name like this, you have to understand the revenue engine — not just the headline. Commerce.com runs a multi-tenant, subscription-led SaaS model with two broad customer tiers. Enterprise plans target large retailers doing $50M+ in annual online sales and are sold on negotiated, usage-and-feature-tiered contracts; this is where the company has deliberately pushed upmarket over the last several years to chase larger, stickier accounts. Essentials plans (Standard, Plus, Pro) serve mid-market and smaller merchants on more standardized monthly pricing.

Layered on top of the core storefront subscription are two growth levers that matter to the bull case. Feedonomics manages product data feeds and syndication to marketplaces like Amazon, Google, and Meta — a high-retention, mission-critical service that expands the company’s wallet share per merchant and is less commoditized than basic hosting. Makeswift adds visual, no-code page building that keeps marketing teams inside the platform. The strategic logic is straightforward: move from “we host your store” to “we run your entire commerce data and experience layer,” which raises switching costs and supports the 78.7% gross margin.

The reason the upmarket shift matters for our thesis is mix. Enterprise revenue tends to be larger-contract, multi-year, and higher-retention — which is exactly what you want when top-line growth is slowing, because it makes the remaining growth more durable and more predictable. The flip side, and the bears’ counter, is that enterprise sales cycles are long and lumpy, and a 2.8% growth year suggests the upmarket motion isn’t yet offsetting churn and slower new-logo adds at the lower end. Both things can be true at once.

The chart that tells the story

This is the part that makes value-minded readers lean in. The stock has fallen roughly 65% over two years, from about $8.38 in June 2024 to $2.91 today. The 52-week range alone runs from $2.41 to $5.54 — meaning you’re buying near the lower end of where it’s traded all year. On the day I pulled this data, shares were down 8.49%, so this is a name actively being thrown out, not one that’s already re-rated.

Here’s the senior-analyst framing I’d put on a one-pager: when a profitless-growth story gets repriced as a no-growth value story, the bar to surprise is low. CMRC doesn’t need to reaccelerate to 20% growth to work from here. It needs to keep doing what the financials below already show it’s doing.

The numbers: slow growth, but a real turnaround in profitability

Revenue reached $342.3M in FY2025, up 2.8% year over year — a clear deceleration from the 7.6% growth posted in FY2024. That’s the bear’s first exhibit, and it’s legitimate: this is not a fast grower anymore.

But look at the red line. Operating losses have collapsed from -$72.4M in FY2023 to -$41.7M in FY2024 to just -$16.2M in FY2025. Net loss followed the same path: -$64.7M → -$27.0M → -$19.3M. Management cut its way to the edge of GAAP profitability without gutting the product — R&D was still $73.0M in FY2025.

The metric that matters most for a stock this cheap is cash. Operating cash flow swung from -$24.2M in FY2023 to a positive $25.5M in FY2025. After capex, that’s roughly $16.9M of free cash flow — a 4.9% FCF margin. Gross margin is a healthy 78.7%, which is exactly the kind of software economics that makes operating leverage possible as the company scales.

The balance sheet backs the turnaround: $141.1M in cash and short-term investments against $39.4M of book equity, and no traditional total-debt figure reported. In plain English: this company is not going to be forced into a dilutive raise at a bad price. It can fund itself.

Valuation: this is the whole argument

At $2.91 and a $240M market cap on $342.3M of revenue, CMRC trades at roughly 0.70x sales. For context, healthy SaaS businesses with 78% gross margins routinely command 4–8x sales. Even Shopify, the 800-pound gorilla, trades at a massive premium to this. You are paying value-stock multiples for a software company that is finally generating cash.

On forward earnings, consensus has CMRC swinging to a profit: estimated EPS of $0.25 in FY2026 and $0.41 in FY2027, on revenue of about $355.2M next year and roughly $30.8M of free cash flow. That puts the stock at about 11.7x forward FY2026 earnings — cheap for any profitable software name. The trailing picture is still negative (TTM EPS of -$0.19), so this is explicitly a “the turn is happening” thesis, not a “it already turned” one.

What Wall Street thinks (and why they can’t agree)

This is one of the most divided analyst setups I follow. Across 4 covering analysts, it’s a dead split: 2 bullish, 2 bearish, for a “Hold” consensus. The average price target is $5.12 — about 76% above today’s price — but the dispersion is enormous.

Needham’s Scott Berg carries a $7.50 Buy (roughly 158% upside), while Barclays’ Raimo Lenschow sits at an Underweight $3.00 (basically flat from here, about 3% upside). Morgan Stanley downgraded to Underweight at $4 in January 2026. When targets range from $3.00 to $7.50 on the same stock, it tells you the outcome hinges almost entirely on one variable: does growth stabilize, or keep sliding?

The honest scorecard

✅ The bull case (pros)

• Dirt-cheap: ~0.70x sales for a 78.7% gross-margin software business is rare.

• Real profitability turn: operating loss cut from -$72.4M to -$16.2M in two years; now FCF-positive (~$16.9M).

• Fortress balance sheet: $141.1M cash + investments, minimal debt — no forced dilution.

• Consensus sees a profit: EPS estimated at $0.25 (FY26) and $0.41 (FY27), ~11.7x forward earnings.

• Optionality: Feedonomics + agentic/AI commerce tools are credible re-rating catalysts the bulls point to.

⚠️ The bear case (cons)

• Growth is decelerating: revenue grew just 2.8% in FY2025 vs 7.6% in FY2024. Low growth caps the multiple.

• Still GAAP-unprofitable: TTM EPS is -$0.19. The “turn” is a forecast, not yet a fact.

• Shopify problem: it competes against a far larger, better-capitalized rival with stronger momentum.

• SBC dilution: stock-based comp was $23.6M — about 6.9% of revenue — quietly diluting holders.

• Analyst skepticism is loud: Morgan Stanley and Barclays both rate it Underweight; the low target is just $3.00.

What I’m watching from here

For readers who want to track this alongside me, here are the specific signposts that would move my read in either direction. I’d rather give you a checklist than a hot take.

Bullish triggers (would push me toward outright constructive): (1) two consecutive quarters of revenue growth stabilizing or reaccelerating above the ~3% FY2025 pace; (2) the first clean GAAP-profitable quarter, validating the consensus FY2026 EPS of $0.25; (3) evidence that Feedonomics and any agentic/AI-commerce products are driving net revenue retention higher; (4) a continued march of free cash flow above the ~$16.9M FY2025 level toward the ~$30.8M the Street models for FY2026.

Bearish triggers (would make me step aside): (1) revenue actually declining year over year, which would confirm the bears’ “no-growth” framing; (2) a reacceleration in losses or a swing back to cash burn; (3) stock-based comp staying stubbornly near 7% of revenue, meaning per-share value leaks even as the business improves; (4) any sign Shopify or Adobe Commerce is taking enterprise logos directly from Commerce.com.

The beauty — and the danger — of a 0.70x-sales stock is that expectations are already on the floor. The market is paying for stagnation, and “merely stable” would likely be enough to re-rate the shares. But the growth ceiling and the Shopify shadow are real, and worth sizing around.

Putting it together

Here’s how I’d frame the setup without telling you what to do with it. This is a business decelerating to ~3% growth in a Shopify-dominated market — that’s a real ceiling, and the bears aren’t wrong to be cautious. At the same time, at 0.70x sales with $141M in net cash, a balance sheet that removes solvency risk, and a profitability inflection that’s already showing up in cash flow, the downside looks more protected than the chart implies, while the upside (if growth merely stabilizes and the AI-commerce optionality plays out) could be meaningful. The average analyst target of $5.12 implies ~76% upside; even the skeptics’ floor near $3.00 is roughly where it trades today.

This is a classic cheap-software setup: the market has priced in permanent stagnation, and the open question is simply whether that’s correct. Watch the next two quarterly prints for any sign of revenue stabilization — that’s the single variable that decides whether this re-rates toward $5 or stays stuck near $3. Where you land on it is your call.

⚖️ IMPORTANT DISCLAIMER — PLEASE READ

This newsletter is for informational and educational purposes only and represents the personal opinions of the author. It is not investment, financial, legal, or tax advice, and nothing here is a recommendation, solicitation, or offer to buy or sell any security. Micro- and small-cap stocks like $CMRC are highly volatile, can be thinly traded, and may result in the total loss of your investment. The author and/or affiliated parties may hold positions in securities mentioned and may buy or sell them at any time without notice. All data is sourced from third-party providers believed to be reliable as of the publication date but is not guaranteed to be accurate, complete, or current; figures may be revised. Past performance and analyst price targets are not indicative of future results. Always do your own research and consult a licensed financial advisor before making any investment decision. By reading this, you agree that the author and publisher bear no liability for any losses or damages arising from your use of this information.

The company is Magnite . And the answer is more complicated than “buy the other side of the trade,” because the thing catching the share has problems of its own, sits in a worse structural position than The Trade Desk ever did, and carries a lottery ticket stapled to it that could be worth more than the entire rest of the business.

Here is the setup that got my attention. In fiscal 2025 Magnite generated $714 million of revenue, $670 million of what it calls Contribution ex-TAC, and $232 million of adjusted EBITDA at a 34.7% margin. Every one of those was a record. The company turned GAAP-profitable for the second straight year, repaid its convertible notes, authorized a buyback, and watched its highest-margin, fastest-growing business cross 50% of the company for the first time. By almost any operating measure, 2025 was the best year in the company’s history.

The stock went from roughly $26 last August to roughly $14 today. The market cap is about $2 billion. The business got better and the stock got cheaper, and that gap is the entire reason this name is worth a week of reading.

I have spent that week with the FY2025 10-K filed in February 2026, the Q1 2026 release from early May, the last several years of annual reports and earnings transcripts back through the 2022 advertising recession, the trade coverage from AdExchanger and Digiday and PPC Land, and the public record of the Google antitrust case that hangs over the whole thing. What follows is my attempt to figure out whether the disconnect is an opportunity or a warning.

Read more

Today the stock trades near $21. The market cap is under $10 billion. The chart looks like a building collapsed.

What follows is an attempt to answer the question every investor I know is asking about this name: is this a generational buying opportunity in a wounded but durable franchise, or the slow recognition that the moat was always thinner than the multiple suggested? I have spent the better part of a week reading the 10-Ks going back to 2018, the earnings call transcripts from 2022 through Q1 2026, the trade press from AdExchanger and Digiday, the securities class action complaint, and Jeff Green’s own LinkedIn posts about the Publicis dispute. I think the answer is more interesting than either side of the consensus.

Read more

First, Some History

Entravision was, for most of its life, an unremarkable Spanish-language broadcaster. It owned TV and radio stations serving U.S. Hispanic communities — decent cash flow, slow growth, nothing that would make a growth investor look twice.

Then in 2018, it quietly acquired a small mobile advertising technology company called Smadex, based in Barcelona. Financial terms were never disclosed. At the time it barely registered as news.

Smadex is a demand-side platform, or DSP — the software that sits between advertisers and the open internet, buying ad inventory programmatically in real time. Its specialty is mobile user acquisition: helping app developers and mobile game publishers find high-value users at scale, using machine learning to predict which users will actually spend money long after install.

For several years Smadex operated quietly in Entravision’s shadow. Then two things happened in quick succession.

In March 2024, Meta announced it was winding down its Authorized Sales Partner program — the arrangement that allowed Entravision to resell Facebook advertising, which at the time accounted for over 50% of total company revenue. The stock fell more than 60% in a single day.

And in January 2024, Smadex quietly hired its first account manager dedicated to the Greater China Region.

These two events, happening almost simultaneously, set up one of the more interesting stories in small-cap media today.

The Numbers Behind the 68% After-Hours Jump

On May 5, 2026, Entravision reported Q1 2026 results. Here is what caused the market’s reaction:

ATS segment (Smadex + Adwake):

• Q1 2026 revenue: $154.6 million — up 204% year-over-year

• Q1 2026 operating profit: $34.3 million — up 427% year-over-year

• Consolidated revenue: $197 million — up 114% year-over-year

• Net income swung from a $47.97 million loss to $12.36 million profit

To put that sequential acceleration in context, here is the full ATS revenue trajectory:

• FY2023: ~$101M

• FY2024: $143M (+42% YoY)

• FY2025: $271M (+90% YoY)

• Q4 2025: $89M (+123% YoY)

• Q1 2026: $155M (+204% YoY)

The growth is not decelerating. It is accelerating. Annualized off Q1 2026, the ATS segment is running at roughly $620 million in revenue with approximately $137 million in annualized operating profit.

Where the valuation stands today. With the stock trading around $8 and approximately 92 million shares outstanding, the market cap is roughly $740 million. Net debt — $163 million in credit facility debt minus $71 million in cash — sits at about $92 million. That puts enterprise value at approximately $830 million.

On an EV/annualized ATS revenue basis, the stock trades at around 1.3x. That is not a screaming bargain for a random business — but for one growing at 200% annually, with operating margins expanding and a business that only recently turned cash-flow positive, it is still meaningfully below where comparable high-growth adtech businesses trade. The market has noticed this company. It has not yet finished pricing it.

What Smadex Actually Does — And Why It’s Hard to Replicate

Read more

This is what a fallen growth stock looks like when the market has stopped believing the growth story — and continues to disbelieve it even when the company beats expectations.

I wrote up this thesis ten days ago at $50. Today, after a quarter that confirmed most of what the bull case argued and a corporate restructuring announcement that opened up a new sum-of-parts catalyst, the stock is cheaper. That deserves a refresh, not a rewrite.

Here’s what makes PYPL more interesting at $45.50 than it was at $50: even after assuming the growth thesis is dead, the math on capital returns alone is striking. The company is generating roughly $6 billion of free cash flow on a market cap of about $41 billion — a ~14.5% FCF yield — and returning essentially all of it to shareholders through buybacks and a newly initiated dividend.

If you bought at today’s price and the business simply held flat for five years, you’d own a meaningfully larger slice of a still-profitable cash machine. That’s the entire thesis in one sentence.

The question is whether “holds flat” is realistic. The Q1 print and the Venmo reorg both pushed the answer in the same direction: more realistic than the tape suggests.

How we got here: the five-year fall

To understand why PayPal trades where it does, you need the brief history.

PayPal was a pandemic darling. From the start of 2020 to mid-2021, the stock more than tripled to $310. The narrative was that COVID was permanently accelerating e-commerce, that PayPal would be the default checkout for a digitized world, and that Venmo would eventually be monetized into a second growth engine. Active accounts surged. Total payment volume surged. The company guided to 750 million active accounts by 2025, more than double its eventual peak.

Then everything broke at once.

E-commerce growth normalized as physical stores reopened. The 750 million target was quietly abandoned, then explicitly retracted. PayPal admitted it had been adding low-quality, incentivized accounts that didn’t transact. Take rates began compressing — and have continued to compress every year since, from 2.03% in 2021 to 1.85% in 2025 to 1.62% in Q1 2026 — because the fastest-growing piece of the business has been Braintree, the unbranded card processing operation that powers checkout for Uber, DoorDash, and others, and it operates at a fraction of the margin of branded checkout. Apple Pay went from a niche product to a default behavior on every iPhone. Shop Pay processed more volume each quarter than seemed possible. Stripe Link began offering one-click checkout that didn’t require the consumer to ever see PayPal.

Three CEOs have now tried to fix this.

Dan Schulman, who ran PayPal from the 2015 eBay spin-off through 2023, leaned into growth at any cost during the pandemic and then struggled to pivot. He was replaced in September 2023 by Alex Chriss, an Intuit veteran hired specifically to refocus the company on profitable growth. Chriss had a coherent plan: deprioritize low-margin Braintree, reinvigorate branded checkout, launch a one-click product called Fastlane to compete with Apple Pay, monetize Venmo via the debit card, and restart the buyback machine. He delivered on the capital returns immediately — $6 billion of buybacks in 2024, another $6 billion in 2025 — but the operational turnaround was slower than the board’s patience allowed.

In Q4 2025, branded checkout volume growth — the single most important metric in the entire company — decelerated to 1%, down from 7% a year earlier. The board fired Chriss on February 2, 2026. CFO Jamie Miller stepped in as interim. On March 1, 2026, Enrique Lores — the outgoing CEO of HP Inc. and a five-year PayPal board member — took the chair.

Two months later, in late April 2026, Lores announced his first major structural move: PayPal would reorganize into three operating segments — Checkout & PayPal; Consumer Financial Services & Venmo; Crypto & Payment Services — making Venmo a standalone unit for the first time. Within hours of the news, CNBC reported that Stripe and other potential acquirers were circling. The Venmo question — long a “someday” optionality argument for PYPL bulls — became a near-term catalyst.

This is where the current investment opportunity begins.

The new management team

If you’re going to underwrite a turnaround, you need to underwrite the people running it. The current bench is genuinely strong, and the comp structure tells you exactly what they’re being paid to do.

Enrique Lores, President and CEO

Lores spent more than 30 years at HP, the last six as CEO. His track record is built on three things: he architected the separation of HP and HPE in 2015 to improve agility, he simplified HP’s cost structure during a brutal period for PCs and printing, and he delivered six consecutive quarters of revenue growth in a business everyone had written off as a melting ice cube. He pushed HP into subscriptions, services, and AI-enabled hardware while protecting cash generation.

Importantly, he is not new to PayPal. He has been on the PayPal board for five years and served as Chair. He didn’t need a learning ramp on the strategic issues; he was in the room when the previous CEO presented the plans that didn’t work. That dual experience — sitting in board meetings hearing what wasn’t getting done, and now being the person who has to do it — is unusual and useful.

The Venmo reorg announcement, three weeks into his tenure, makes the Lores playbook explicit. He’s running the same play he ran at HP: simplify the structure, give each business its own P&L and accountability, and let the market value the parts properly. At HP that was the HPE separation; here it’s segmentation that, at minimum, makes Venmo cleanly visible — and at maximum, makes it cleanly sellable.

His public messaging since taking the seat has been careful and revealing: “execute with greater speed and precision,” “hold ourselves accountable for consistent delivery quarter on quarter,” “prioritize what matters most, simplify where necessary.” This is operator vocabulary, not visionary vocabulary. After three years of strategic narratives that didn’t translate to results, that shift in tone is welcome.

He has already committed $400 million in 2026 to revamp the branded checkout experience, and on the Q1 call announced a $1.5 billion cost program over 2-3 years to fund the reinvestment.

Jamie Miller, CFO and COO

Miller is the kind of CFO you want during a turnaround. She joined PayPal in November 2023 after serving as Global CFO of EY, where she led the separation and aborted IPO of EY’s consulting arm. Before that she was CFO of Cargill, and before that she spent more than a decade at GE — including as CFO of the entire company and CEO of GE Transportation. She also currently sits on Qualcomm’s board.

Her remit at PayPal expanded in 2025 to include the COO role, which is a strong signal the board was already grooming her for broader operational responsibility. When Chriss was fired, she was the natural interim choice, and she remains the operational backbone underneath Lores. She received a $3 million cash retention award to stay through the transition — which tells you the board considers her departure a real risk and is paying to prevent it.

Miller has spent her career inside complex, multi-segment industrial businesses. PayPal — with its branded checkout, Braintree, Venmo, BNPL, merchant services, and emerging stablecoin/AI offerings, now formally split into three reportable segments — is exactly that kind of business.

What Lores is being paid

The comp structure is, in my view, the single most underappreciated piece of this story. Here’s the package, pulled directly from the 8-K and proxy:

• Base salary: $1.45 million. Standard.

• Target annual bonus: 200% of base ($2.9M target). Standard.

• “Make-whole” RSUs: $20 million, vesting ratably over three years. Time-based — he gets this just for showing up, compensating him for HP equity he forfeited.

• 2026 annual RSUs and PSUs: $33 million combined ($16.5M each), on the same terms as other senior executives.

• 2027-related front-loaded RSUs: $11 million, vesting starting on the second anniversary.

• One-time Stock Price PSUs: $25 million target value, up to $62.5 million at maximum.

That last piece is the one that matters. The Stock Price PSU is the cleanest alignment instrument I’ve seen on a major turnaround in a long time. The hurdles are measured against a baseline stock price of $42.58, on the average closing price over any consecutive 60-calendar-day period, ending between the third and fifth anniversary of his March 1, 2026 start date. So between March 2029 and March 2031.

In plain English: Lores does not get a meaningful chunk of his pay unless the stock at least gets back to where it was at the end of 2024. He doesn’t get the stretch payout unless PYPL more than doubles. He doesn’t get the maximum unless it nearly triples.

To be fair to the bear case: most of his compensation isn’t actually performance-gated. The $20M make-whole and $33M annual grant are largely time-based. So the headline “$89M package” overstates the alignment. But the upside — the part that turns a $5M-a-year executive package into a $50M+ outcome — is entirely tied to a stock that needs to roughly double from where the package was struck. The proxy itself flags the design intent: “highly performance-oriented, with no sign-on cash component, and with a special performance-based equity award tied to rigorous stock price hurdles over a period up to five years.”

When you’re trying to figure out whether a CEO actually believes in his business, look at how he’s paid for the upside. Lores wins big only if shareholders win first.

That is exactly the alignment you want.

The Chriss exit

For completeness: Chriss received severance under PayPal’s standard “termination without cause” plan. There was no scandal, no clawback, no acrimony in the public filings — just a board that decided execution wasn’t fast enough and acted accordingly.

What actually broke (and is it fixable?)

To understand the opportunity, you have to understand the wound. The wound is branded checkout.

Branded checkout is the yellow PayPal button you see on millions of e-commerce sites. When a consumer clicks it instead of typing in a card, PayPal earns a meaningfully higher take rate than it does on its other businesses. Mizuho’s Dan Dolev put it bluntly: the yield on the branded button is multiples of any other product in the portfolio.

That button was in trouble. The deceleration over the last six quarters:

• Q4 2024: branded checkout TPV up ~7%

• Q1 2025: ~6%

• Q2 2025: ~5%

• Q3 2025: ~5%

• Q4 2025: 1%

• Q1 2026: 2% (currency-neutral)

Q1 2026 was the first sequential acceleration in five quarters. It’s a small one — going from 1% to 2% is not victory — but the direction matters. The bear case required this number to keep falling toward zero or negative. So far, it hasn’t.

Now the broader bear interpretation, which is still legitimate: Apple Pay, Google Pay, and Shop Pay are systematically eating PayPal’s lunch, especially on mobile, and especially with younger consumers who never built the muscle memory of clicking the yellow button. In the U.S., PayPal’s user base is forecast to grow less than 1% in 2026 to roughly 92 million, while Apple Pay closes in on 90.5 million and Google Pay reaches 55 million. Stripe Link is doing similar damage on the merchant side. Several analysts downgraded the stock after Q4 2025, citing exactly this view. Canaccord cut to Hold, arguing PayPal’s “core e-commerce proposition is losing relevance.” HSBC cut to Hold. Mizuho flagged that “the competitive nature of payments now appears more intense than we realized.”

And the agentic commerce risk is real, not theoretical. The FY2025 10-K explicitly added “autonomous AI agents” as a demand risk for PayPal services for the first time. Management thinks this matters; investors should too. The 5-10 year question is whether the checkout button is even the right unit of analysis if AI agents are doing the buying.

The bull interpretation, with Q1 evidence:

1. The branded checkout product is genuinely behind. Most PayPal merchants are still using older integrations. The new Fastlane product, biometric authentication (now at ~36% adoption), and one-click flows are still being rolled out. As of late 2024, only about 25% of U.S. checkout traffic was using the newer innovations. The other 75% is operating on what is effectively 2018 technology. The Q1 acceleration to 2% came before most of the $400M reinvestment has hit.

2. Q1 confirmed the rest of the portfolio is working — and accelerating. Venmo TPV grew 14% (sixth consecutive quarter of double-digit growth). Unbranded/PSP volume accelerated to 11%, up from 7% the prior half. Enterprise payments grew in the mid-teens. Other value-added services revenue grew 10% to $852M. The narrative is “branded is broken,” but the rest of the portfolio is now growing faster than branded by a substantial margin.

3. Active accounts inflected positive. 439 million as of Q1 2026, up 1% YoY. This was negative for two years. The customer base is no longer shrinking — it’s quietly growing again.

4. Lores reset expectations honestly. Q2 2026 guidance: high-single-digit EPS decline. That’s the kitchen-sink reset, even though it was framed as a reiteration of the FY26 number. New CEOs almost always set a low bar; Lores set one without explicitly cutting the year, which is harder and arguably more credible.

5. The Venmo reorg gave the bull case a near-term catalyst it didn’t have before. “Someday Venmo gets monetized” is now “Venmo is reportable as a standalone segment within months and Stripe is publicly interested.” Even if no transaction happens, the segment-level disclosure forces investors to value Venmo on its own. At $1.7B+ of revenue growing 20%, Venmo at fintech multiples alone is worth a meaningful chunk of the current EV.

The honest version of the operational picture: branded checkout is structurally challenged but probably not terminally broken, and the rest of the portfolio is doing better than the headline suggests. The Q1 print didn’t resolve the debate; it just shifted the burden of proof modestly toward the bulls.

The capital returns engine

This is where PayPal becomes genuinely interesting at $45.50, regardless of what you think about branded checkout.

The business throws off cash. That hasn’t been in dispute through the entire deceleration, and Q1 2026 reinforced it: $1.72 billion in adjusted free cash flow in the quarter alone, on track for the at-least $6 billion management has guided to for the year.

Buybacks

In 2024, PayPal returned $6 billion to shareholders by repurchasing approximately 92 million shares. The board authorized a fresh $15 billion repurchase program alongside the Q4 2024 results, on top of $4.86 billion still remaining from the prior 2022 authorization. So roughly $20 billion of authorized buyback capacity at the start of 2025.

In 2025, the company executed another $6 billion of buybacks, retiring approximately 86 million shares. In Q1 2026 alone, PayPal repurchased $1.5 billion of stock. Management has guided to another $6 billion of buybacks for the full year. At today’s market cap of ~$41 billion, a $6 billion annual buyback is roughly 14.5% of shares outstanding being retired every year. There is no other large-cap fintech doing this.

The shares-outstanding count tells the story:

• End of 2023: ~1.07 billion shares

• End of 2024: ~989 million shares

• End of 2025: ~921 million shares

• After Q1 2026: ~900 million shares

That’s a 16% reduction in share count in 27 months. If the program continues at this pace through 2028 — and at today’s price, $6B retires ~130M shares per year — share count would drop below 700 million.

A note on the math, because this is where buyback theses go wrong: the per-share benefit only compounds at this pace if the stock stays cheap. If PYPL re-rates to $80, $6B buys back fewer shares. The bull case has to choose between fast share retirement (stock stays cheap) and multiple expansion (stock re-rates). You can’t have both fully. What you can have is a substantial mix — meaningful share retirement plus meaningful re-rating, which is the realistic path to a double from here.

Dividend

PayPal initiated a quarterly dividend of $0.14 per share in 2025, the first in its history. That’s roughly $500 million annually and a 1.2% forward yield at the current price. Modest in isolation, but symbolically important: management is now committed to returning cash via two channels and has indicated long-term targets of distributing 70-80% of FCF to shareholders annually.

Balance sheet

Approximately $14.8 billion in cash and investments against $11.6 billion in debt. Net cash positive. There is no leverage problem. The buyback can continue at $6 billion annually for years without requiring additional operating cash flow. The credit rating is investment grade and management has explicitly committed to maintaining it.

What’s underappreciated: the optionality stack

The article I wrote ten days ago undersold three optionality assets that the Q1 quarter and recent disclosures have made more important.

1. PayPal Bank. In December 2025, PayPal applied to the Utah Department of Financial Institutions and the FDIC to charter PayPal Bank as a Utah industrial loan company. If granted — multi-year, uncertain — this lets PayPal directly earn net interest income on the float underlying customer balances rather than splitting it with partner banks. With ~$40B+ flowing through customer balances, this is a structural margin story that is in zero sell-side models.

2. PYUSD and the GENIUS Act. Congress enacted the Guiding and Establishing National Innovation for U.S. Stablecoins Act in July 2025, creating a federal framework for stablecoin regulation. PayPal is now formally a “digital asset service provider” under GENIUS, and PYUSD is the largest stablecoin issued by a regulated U.S. public-company partner. With federal regulatory clarity in place, this graduates from experiment to product line — particularly for cross-border payments, where stablecoin economics could meaningfully undercut Xoom and remittance competitors.

3. The Venmo carve-out. Even if Lores never sells Venmo, the segment reporting that begins later this year forces every PYPL investor to value Venmo on its own. At $1.7B revenue and 20% growth, comparable fintech multiples put Venmo’s standalone value somewhere in the $15–25B range — against a current PYPL enterprise value of ~$38B. The math implies the rest of the company trades for less than 7x EPS once you back out a fairly valued Venmo. That’s the sum-of-parts argument the segment reporting will make visible.

None of these are the thesis. The thesis is the $6B buyback retiring 14% of the float per year. But these are the call options stapled to the thesis, and they’re more valuable than the article wrote ten days ago.

The valuation case, three ways

The simple way. Market cap ~$41B. Free cash flow ~$6B. That’s a 14.5% FCF yield on a business with a net-cash balance sheet, a global brand with 439 million active accounts, and a CEO with a credible turnaround track record. Even if FCF declines 20% over five years, you’re still earning a yield well above the market.

The buyback math. If PayPal repurchases $6B annually at average prices in the $50-60 range over the next three years, the share count drops from ~900M today to roughly ~660M by year-end 2028. If FCF holds flat at $6B, FCF per share goes from ~$6.70 today to ~$9.10. At a 10x multiple, that’s $91/share — roughly a double from here, with no help from the operational turnaround. This is, not coincidentally, almost exactly the stretch hurdle in Lores’ comp package. The catch worth naming: this math assumes PYPL stays cheap enough to retire that many shares; if it re-rates faster, you get less share retirement and more multiple expansion. Either way, you arrive somewhere similar.

The operational re-rating. If Lores executes anything close to his HP playbook and branded checkout stabilizes at 4-5% growth — half of what it was at the end of 2024 — non-GAAP EPS could compound at low-to-mid teens over three years. Add the $1.5B cost program. Add Venmo segment monetization. Reasonable 2028 EPS of $7.00 at a 13x multiple (still below the 10-year average) puts the stock at $90+. The buyback adds another layer on top.

The base case and bull case rhyme because the buyback does so much of the work. The bear case requires the buyback to actually break, which requires the cash flow to break, which requires multiple parts of the business to deteriorate simultaneously while a CEO with a strong track record fails to course-correct. That’s possible. It’s not the most likely path.

Catalysts from here

1. Q2 2026 earnings (early August). Lores guided down hard for Q2 — high-single-digit EPS decline. That’s the bar. Anything that beats it materially is a positive surprise. Watch branded checkout: a third consecutive quarter of stabilization or modest acceleration would shift the narrative meaningfully.

2. Lores’ first formal strategy presentation (likely Q2 or Q3 2026). New CEOs typically announce a strategic plan within their first 90-120 days. The reorg announcement was the first half of this; the financial framework around it is the second half. The HP playbook — simplify, focus, cut cost, reinvest selectively, return cash — applied to PayPal is a credible 18-month story. Watch for explicit deprioritization of any product areas. Lores is famous for being willing to cut things, which is something Chriss was reportedly slower to do.

3. Venmo segment financials (later in 2026). The first time PYPL discloses Venmo as a standalone segment will be a price-discovery event. Whatever the segment-level margin and growth profile look like will be tested directly against fintech comps.

4. Branded checkout inflection. Q1 2026 showed the first hint. The $400M reinvestment is going into the checkout experience, biometric/passkey adoption, and merchant-side improvements. If branded TPV growth re-accelerates from 2% back to mid-single digits, the multiple re-rates.

5. Continued buyback execution. The boring catalyst, but the most reliable one. Every quarter that produces another $1.5B in buybacks at sub-$55 prices is a quarter where the math compounds quietly in your favor.

The asymmetry is the point. To lose meaningfully from here, you need branded checkout to actively decline (not just decelerate), the buyback to be cancelled, and Venmo/BNPL/merchant services to all stall. That’s possible, but it requires several bad things to happen simultaneously to a business still generating $6B in annual cash, with a new CEO whose entire $25-62M upside depends on the stock at least getting back to its 2024 levels.

The risks worth respecting

I want to be fair to the bear case, because the bears have been right for five years.

• Take rate compression is structural. It went from 2.03% in 2021 to 1.62% in Q1 2026 — 41 bps of erosion in five years. At $1.79T of TPV, every 1 bp is roughly $180M of revenue. As Braintree, Venmo, and enterprise PSP grow faster than branded, the blended take rate keeps falling. This is a slow grind that doesn’t go away with a new CEO.

• Five years of flat operating income. Look hard at the 10-Ks: non-GAAP operating income has been ~$6.4B every year from 2021 to 2025. The entire EPS growth has come from the share count dropping ~20%. The bull case requires either operating income to inflect or the buyback to keep doing the work. If both fail, the stock is a value trap.

• Apple Pay and Shop Pay aren’t going away. They are getting better, and they have distribution PayPal can’t match. Apple Pay is the default on every iPhone shipped.

• Venmo monetization keeps getting promised and partially delivering. The Venmo debit card growth (TPV +50%) is genuinely encouraging, but the broader “Venmo as a second growth engine” story has been “next year” for a long time. The reorg makes it harder to hide if Venmo doesn’t deliver — but also harder to ignore if it does.

• Agentic commerce. If AI agents do the buying, the checkout button matters less. PayPal has hedges — PYUSD, an AI API, agentic-commerce partnerships — but it’s early. Management put this risk in the 10-K for the first time in 2025; they’re not pretending it isn’t real.

• CEO transitions create execution risk. Lores is unproven at PayPal specifically. Five years on the board helps but doesn’t substitute for operating experience inside the company. The HP playbook may not translate cleanly to a payments business with very different competitive dynamics.

• Three CEOs in three years cuts both ways. The bull frame is “the board is acting.” The bear frame is “the board has now publicly admitted twice in three years that it doesn’t know what to do.” Both are true.

• Macro risk. PayPal is consumer-spend-exposed. A real recession could pressure TPV across the board, and FCF guidance assumes a relatively normal consumer environment.

The honest version of this thesis: I don’t know whether branded checkout reaccelerates beyond 2% to anything that matters. Nobody does, including Lores on day 65. But I don’t need to know, because the buyback gives me time to find out — and the new management is being paid in a way that makes their interests indistinguishable from mine.

Bottom line

PayPal at 8x earnings — with a 14.5% FCF yield, retiring 14% of its shares annually, sitting on a net-cash balance sheet, with a respected operator just installed as CEO whose entire $25-62 million upside is gated on the stock recovering, and with a fresh segmentation that turns Venmo from someday-optionality into near-term catalyst — is the kind of setup value investors wait years for.

It’s cheap because the business has real problems. It’s investable because the capital allocation works regardless of whether those problems get fully fixed, and because the people running it now are paid only to fix them.

The market priced in continued decline today and sold it off another 10% on a beat. The buyback is pricing in survival. The CEO comp package is pricing in a double. The Venmo carve-out is pricing in optionality that wasn’t there a month ago.

I think survival is a low bar to clear. The rest is upside.

Disclosure: This is a newsletter, not investment advice. Do your own work. Numbers are pulled from PayPal’s most recent SEC filings (8-Ks, 10-K, DEF 14A proxy), Q1 2026 earnings release and call dated May 5, 2026, the offer letter for Enrique Lores filed February 2026, and analyst commentary as of early May 2026. Stock prices, share counts, and authorization balances move; verify before acting. I may or may not own PYPL — disclose your own position before publishing.

It owns a near-monopoly. Roughly 90% market share in its core category. The product is so embedded in U.S. financial infrastructure that virtually every consumer in the country has interacted with it, often without knowing.

Its non-GAAP operating margins are 65%. Its software platform has a 136% dollar-based net retention rate, which puts it in the top decile of all enterprise software businesses. Combining its earnings growth and aggressive buybacks, it has delivered a 5,333% total return for shareholders over the last 14 years, versus 379% for the Russell 3000 over the same period. According to the company’s own proxy filing, fewer than 1% of Russell 3000 companies have matched this performance over its CEO’s tenure.

And the stock is down roughly 50% from its high in late 2024.

This is not happening because the business is broken. The business just reported its best quarter in history. Revenue grew 39% year over year. The most-watched line item — a particular high-margin product I will detail in the paid section — grew 127%. Earnings per share grew 60%. The company raised full-year guidance.

It is happening because of three fears, all of which I think are overstated.

Fear one: AI will commoditize software

This is the SaaSpocalypse narrative that has hit nearly every software stock since late 2025. The thesis is that large language models will collapse the cost of building enterprise software, which means incumbents lose pricing power, which means SaaS multiples deserve to compress.

There is something to this argument for some categories. But for the company I am writing about, it is essentially wrong. The reason is regulatory, not technical.

The product this company sells is governed by fair lending laws — a body of regulation that requires explainability, prohibits the use of certain predictive variables (redlining is the obvious example), and demands that consumers receive specific reason codes when they are denied credit. These requirements exist for good reasons, and they are not going away.

Black-box AI models cannot easily meet these requirements. We have a recent precedent: when Upstart attempted AI-driven underwriting under a CFPB no-action letter, the arrangement was eventually terminated after the agency raised disparate-impact concerns. The regulator wants reason codes. The regulator wants explainability. The regulator does not want a neural network deciding who gets a mortgage in a way it cannot audit.

This means the moat here is not technical, it is legal. And legal moats are the most durable kind.

Fear two: A new competitor will take share

There is a free, government-backed alternative to this company’s flagship product. It has been around for 20 years. The federal regulator overseeing the U.S. mortgage market just opened the door for lenders to use it. The credit bureaus that distribute the incumbent product also own the competing product, and they have priced it at $0.99 to try to win share.

This sounds bad. And on the surface it is the bear case.

But here is what I learned from reading the most recent earnings call transcript carefully. The CEO was asked directly what market share the competing product has actually achieved across the verticals where it has been competing for two decades — credit cards, auto loans, personal loans, non-conforming mortgages. His answer: roughly 2%. In the non-conforming mortgage market specifically, where it has been free to compete: zero share.

Why? Because the market that ultimately matters here is the secondary market — the one where mortgages get bundled, securitized, and sold as bonds. Those bonds are priced based on credit risk, and bond investors will not accept paper priced on a model that has only ten years of historical data and has never been tested through a full credit cycle. The CEO disclosed on the call that out of recent securitization batches, “something like 20 mortgages” used the competing product. Less than 0.1%.

A free competitor that has been pushed by powerful distribution partners for 20 years and gone nowhere is not the threat investors are pricing in. It is a known unknown that has remained known and unknown.

Fear three: Regulators are coming

This one is real. A sitting U.S. senator is running an active antitrust investigation. The federal regulator overseeing this market is publicly negotiating prices with the company. The Trump administration’s housing officials have explicitly cited this company as a target for cost reduction.

I take this seriously. But two things matter.

First, antitrust cases against pure software royalty businesses are nearly impossible to win. There is no bundling, no tying, no exclusive dealing. The competing product is freely available. The “monopoly” exists because customers prefer the product, which is the literal definition of competition working.

Second, the company has already adapted. They restructured their pricing in October 2025 to a direct license program that bypasses the credit bureaus. They cut their headline price to $0.99 for the new version of the product — but added a $65 funding fee tied to closed loans. The CEO told analysts on the recent earnings call that the company is “revenue neutral” between the new pricing model and the old one. They conceded the headline number while structurally capturing more downstream value. This is exactly the kind of pricing innovation that monopoly rents enable.

What the math actually says

Forward price-to-earnings sits around 25x on a business growing earnings per share at roughly 35% this year. That gives you a PEG ratio under 1, which for a 90%-market-share infrastructure royalty business is genuinely rare.

Compare that to the right peer set. S&P Global trades around 22-25x forward earnings and grows EPS in the high single digits. Moody’s trades around 25-27x and grows in the mid teens. MSCI trades around 27x and grows in the low double digits. This company trades at the lower end of the multiple range despite having more than triple the growth rate of any of them.

The reverse DCF is even more striking. At today’s price, the market is pricing in roughly 9% annual revenue growth for the next 10 years if margins hold. The company just printed 39%. Even if you assume meaningful margin compression to account for regulatory pressure, the market is still pricing in only 13% revenue growth for a decade. That is well below what the business has done in any year of the last five.

Three independent valuation models — Morningstar, the Wall Street analyst consensus, and GuruFocus — all cluster between $1,940 and $2,020 per share. The stock currently sits near $1,010. That is not a small dispersion. That is the market and the analytical community looking at the same business and arriving at fundamentally different answers.

The capital allocation flywheel

There is one more thing I want to share before the paywall, because it is the part of this story that gets the least attention.

Over the last two decades, this company has reduced its share count from 74 million in fiscal 2005 to roughly 23 million today. That is approximately a 69% reduction. They are not paying dividends. They are not making significant acquisitions. They are taking nearly every dollar of free cash flow and buying back stock, and they have been doing it consistently across multiple bull and bear markets.

In the most recent quarter, they bought back $605 million of stock at an average price of $1,251 per share — the largest single-quarter buyback in company history. After the quarter ended, they bought another $170 million at an average price of $1,040 per share. The CEO said on the call that current prices look “opportunistic.”

The CEO personally owns roughly 363,000 shares — approximately 1.52% of the company, worth around $370 million at current prices. He has not sold a single share since November 2025. In the six months ending September 2025, by comparison, he sold around $63 million worth of stock. The change is not subtle.

Insiders who believe in the business buy when it is cheap. Insiders who do not, sell.

The last time I saw a setup this clean in a 90%-share infrastructure business, the stock more than doubled in 18 months.

The rest of this post is for paid subscribers.

Below the paywall:

• The company name and ticker, with current price and market cap

• A pricing decomposition that shows exactly how this company has compounded one specific product’s price by 16.7x over eight years

• A discounted cash flow model with bear, base, and bull scenarios — and the specific share price targets each implies

• A reverse DCF showing exactly what the market is currently pricing in

• The specific catalysts I expect to drive the stock over the next 12 months, with dates

• A position-sizing framework with hard stop triggers

• The five risks I am taking most seriously and the data I am tracking on each one

• A side-by-side comparison with five financial-infrastructure peers

• The regulatory timeline that explains why the current selloff is the buying opportunity, not the warning sign

Read more

This is what a fallen growth stock looks like when the market has stopped believing the growth story.

But here’s what makes PYPL interesting at this price: even after assuming the growth thesis is dead, the math on capital returns alone is striking. The company is generating roughly $6 billion of free cash flow on a market cap of about $46 billion — a ~13% FCF yield — and returning essentially all of it to shareholders through buybacks and a newly initiated dividend.

If you bought at today’s price and the business simply held flat for five years, you’d own a meaningfully larger slice of a still-profitable cash machine. That’s the entire thesis in one sentence.

The question is whether “holds flat” is realistic. And the answer to that question depends almost entirely on a leadership team that just got rebuilt from the top.

How we got here: the five-year fall

To understand why PayPal trades where it does, you need the brief history.

PayPal was a pandemic darling. From the start of 2020 to mid-2021, the stock more than tripled to $310. The narrative was that COVID was permanently accelerating e-commerce, that PayPal would be the default checkout for a digitized world, and that Venmo would eventually be monetized into a second growth engine. Active accounts surged. Total payment volume surged. The company guided to 750 million active accounts by 2025, more than double its eventual peak.

Then everything broke at once.

E-commerce growth normalized as physical stores reopened. The 750 million target was quietly abandoned, then explicitly retracted. PayPal admitted it had been adding low-quality, incentivized accounts that didn’t transact. Take rates began compressing because the fastest-growing piece of the business — Braintree, the unbranded card processing operation that powers checkout for Uber, DoorDash, and others — operates at a fraction of the margin of branded checkout. Apple Pay went from a niche product to a default behavior on every iPhone. Shop Pay processed more volume each quarter than seemed possible. Stripe Link began offering one-click checkout that didn’t require the consumer to ever see PayPal.

Three CEOs have now tried to fix this.

Dan Schulman, who ran PayPal from the 2015 eBay spin-off through 2023, leaned into growth at any cost during the pandemic and then struggled to pivot. He was replaced in September 2023 by Alex Chriss, an Intuit veteran hired specifically to refocus the company on profitable growth. Chriss had a coherent plan: deprioritize low-margin Braintree, reinvigorate branded checkout, launch a one-click product called Fastlane to compete with Apple Pay, monetize Venmo via the debit card, and restart the buyback machine. He delivered on the capital returns immediately — $6 billion of buybacks in 2024 — but the operational turnaround was slower than the board’s patience allowed.

In Q4 2025, branded checkout volume growth — the single most important metric in the entire company — decelerated to 1%, down from 7% a year earlier. The board fired Chriss on February 2, 2026. CFO Jamie Miller stepped in as interim. On March 1, 2026, Enrique Lores — the outgoing CEO of HP Inc. and a five-year PayPal board member — took the chair.

This is where the current investment opportunity begins.

The new management team

If you’re going to underwrite a turnaround, you need to underwrite the people running it. The current bench is genuinely strong, and the comp structure tells you exactly what they’re being paid to do.

Enrique Lores, President and CEO

Lores spent more than 30 years at HP, the last six as CEO. His track record is built on three things: he architected the separation of HP and HPE in 2015 to improve agility, he simplified HP’s cost structure during a brutal period for PCs and printing, and he delivered six consecutive quarters of revenue growth in a business everyone had written off as a melting ice cube. He pushed HP into subscriptions, services, and AI-enabled hardware while protecting cash generation.

Importantly, he is not new to PayPal. He has been on the PayPal board for five years and served as Chair of the board. He didn’t need a learning ramp on the strategic issues; he was in the room when the previous CEO presented the plans that didn’t work. That dual experience — sitting in board meetings hearing what wasn’t getting done, and now being the person who has to do it — is unusual and useful.

His public messaging since taking the seat has been careful and revealing: “execute with greater speed and precision,” “hold ourselves accountable for consistent delivery quarter on quarter,” “prioritize what matters most, simplify where necessary.” This is operator vocabulary, not visionary vocabulary. After three years of strategic narratives that didn’t translate to results, that shift in tone is welcome.

He has already committed $400 million in 2026 to revamp the branded checkout experience.

Jamie Miller, CFO and COO

Miller is the kind of CFO you want during a turnaround. She joined PayPal in November 2023 after serving as Global CFO of EY, where she led the separation and aborted IPO of EY’s consulting arm. Before that she was CFO of Cargill, and before that she spent more than a decade at GE — including as CFO of the entire company and CEO of GE Transportation. She also currently sits on Qualcomm’s board.

Her remit at PayPal expanded in 2025 to include the COO role, which is a strong signal the board was already grooming her for broader operational responsibility. When Chriss was fired, she was the natural interim choice, and she remains the operational backbone underneath Lores. She received a $3 million cash retention award to stay through the transition — which tells you the board considers her departure a real risk and is paying to prevent it.

Miller has spent her career inside complex, multi-segment industrial businesses. PayPal — with its branded checkout, Braintree, Venmo, BNPL, merchant services, and emerging stablecoin/AI offerings — is exactly that kind of business.

What Lores is being paid

The comp structure is, in my view, the single most underappreciated piece of this story. Here’s the package, pulled directly from the 8-K and proxy:

• Base salary: $1.45 million. Standard.

• Target annual bonus: 200% of base ($2.9M target). Standard.

• “Make-whole” RSUs: $20 million, vesting ratably over three years. This compensates him for HP equity he forfeited by leaving. Time-based — he gets this just for showing up.

• 2026 annual RSUs and PSUs: $33 million combined ($16.5M each), on the same terms as other senior executives.

• 2027-related front-loaded RSUs: $11 million, vesting starting on the second anniversary.

• One-time “Stock Price PSUs”: $25 million target value, up to $62.5 million at maximum.

That last piece is the one that matters. The Stock Price PSU is the cleanest alignment instrument I’ve seen on a major turnaround in a long time. Here are the actual hurdles, against a baseline stock price of $42.58:

These are measured on the average closing price over any consecutive 60-calendar-day period, ending between the third and fifth anniversary of his March 1, 2026 start date. So between March 2029 and March 2031.

Translate that into plain English: Lores does not get a meaningful chunk of his pay unless the stock at least gets back to where it was at the end of 2024. He doesn’t get the stretch payout unless PYPL more than doubles. He doesn’t get the maximum unless it nearly triples.

He has $20M of “make-whole” RSUs that are essentially guaranteed, which is fair compensation for leaving HP. But the upside — the part that turns a $5M-a-year executive package into a $50M+ outcome — is entirely tied to a stock that needs to roughly double from where the package was struck. The proxy itself flags this design intent: “highly performance-oriented, with no sign-on cash component, and with a special performance-based equity award tied to rigorous stock price hurdles over a period up to five years.”

When you’re trying to figure out whether a CEO actually believes in his business, look at how he’s paid. Lores is structurally short the stock if PayPal trades sideways for five years. He only wins if shareholders win first.

That is exactly the alignment you want.

The Chriss exit

For completeness: Chriss received severance under PayPal’s standard “termination without cause” plan and remains an employee in a non-officer capacity through March 2, 2026, to assist with the transition. There was no scandal, no clawback, no acrimony in the public filings — just a board that decided execution wasn’t fast enough and acted accordingly. The fact that Chriss is staying through the handoff (and that Miller is being paid to stay) suggests the transition is being run cooperatively rather than chaotically.

What actually broke (and is it fixable?)

To understand the opportunity, you have to understand the wound. The wound is branded checkout.

Branded checkout is the yellow PayPal button you see on millions of e-commerce sites. When a consumer clicks it instead of typing in a card, PayPal earns a meaningfully higher take rate than it does on its other businesses. Mizuho’s Dan Dolev put it bluntly: the yield on the branded button is multiples of any other product in the portfolio.

That button is in trouble.

Here is the deceleration over the last five quarters:

• Q4 2024: branded checkout TPV up ~7%

• Q1 2025: ~6%

• Q2 2025: ~5%

• Q3 2025: ~5%

• Q4 2025: 1%

This is the chart that got Chriss fired. The bear interpretation is straightforward: Apple Pay, Google Pay, and Shop Pay are systematically eating PayPal’s lunch, especially on mobile, and especially with younger consumers who never built the muscle memory of clicking the yellow button. In the U.S., PayPal’s user base is forecast to grow less than 1% in 2026 to roughly 92 million, while Apple Pay closes in on 90.5 million and Google Pay reaches 55 million. Stripe Link is doing similar damage on the merchant side. And agentic AI commerce — bots that complete purchases on your behalf — could bypass the checkout button entirely over a 5-10 year horizon.

That’s the structural decline thesis. It isn’t crazy. Several analysts downgraded the stock after Q4 2025, citing exactly this view. Canaccord cut to Hold, arguing PayPal’s “core e-commerce proposition is losing relevance.” HSBC cut to Hold, citing low confidence in the company’s ability to stabilize branded volumes. Mizuho flagged that “the competitive nature of payments now appears more intense than we realized.”

Now the bull interpretation. There are real reasons to believe the deceleration is partially fixable:

1. The branded checkout product is genuinely behind. Most PayPal merchants are still using older integrations. The new Fastlane product, biometric authentication, and one-click flows are still being rolled out. As of late 2024, only about 25% of U.S. checkout traffic was using the newer innovations. The other 75% is operating on what is effectively 2018 technology. If even half that base gets upgraded in 2026-2027, the experience improves materially.

2. Q4 2025 included real macro and comp headwinds. U.S. retail was soft, international FX was a drag, and Q4 2024 had been a particularly strong comparison. Even bears acknowledge it’s hard to fully separate cyclical from structural.

3. The non-branded businesses are working. Venmo revenues grew 20% in 2025 to $1.7 billion, powered by 50% TPV growth on the Venmo debit card. Buy now, pay later volume hit $40 billion, up 20%. Enterprise business payments has now posted seven consecutive quarters of profitable growth. The narrative is “branded is broken,” but most of the rest of the portfolio is actually accelerating.

4. Lores has $400M to throw at the problem. The product roadmap is funded. New CEOs almost always reset expectations downward and kitchen-sink the bad news on their first earnings call so they have a low bar to beat. Expect that on May 5.

5. Active accounts stopped declining. Total active accounts hit 438-439 million as of mid-2025, up 2% year-over-year. This was negative for two years. The customer base is no longer shrinking.

The honest version of the operational picture: branded checkout is 50% structurally challenged and 50% fixable with better product execution. We won’t know the split for 18-24 months. The investment question is whether the rest of the thesis works while we wait.

The capital returns engine

This is where PayPal becomes genuinely interesting at $50, regardless of what you think about branded checkout.

The business throws off cash. That hasn’t been in dispute through the entire deceleration. The bear case requires you to believe FCF is going to fall meaningfully — and even most bears don’t model a sharp decline before 2027-2028.

Buybacks

This is the headline number. In 2024, PayPal returned $6 billion to shareholders by repurchasing approximately 92 million shares, reducing the average share count by roughly 6%. The board authorized a fresh $15 billion repurchase program alongside the Q4 2024 results, on top of $4.86 billion still remaining from the prior 2022 authorization. So roughly $20 billion of authorized buyback capacity at the start of 2025.

In 2025, the company executed another $6 billion of buybacks, retiring approximately 84 million shares on a trailing-12-month basis as of mid-year. Management has guided to another $6 billion of buybacks in 2026.

At today’s market cap of ~$46 billion, a $6 billion annual buyback is roughly 13% of shares outstanding being retired every year. There is no other large-cap fintech doing this.

The shares-outstanding count tells the story:

• End of 2023: ~1.07 billion shares

• End of 2024: ~1.00 billion shares

• Mid-2025: ~940 million shares

• Today: ~920 million shares

That’s a 14% reduction in share count in 24 months. If the program continues at this pace through 2028, share count would approach 600 million — a one-third reduction from today.

Dividend

PayPal initiated a quarterly dividend of $0.14 per share in 2025, the first in its history as a standalone public company. That’s roughly $500 million annually and a 1.1% forward yield at the current price. Modest in isolation, but symbolically important: the company is now committed to returning cash via two channels, not one, and management has indicated long-term targets of distributing 70-80% of FCF to shareholders annually.

Balance sheet

Approximately $14.8 billion in cash and investments against $11.6 billion in debt. Net cash positive. There is no leverage problem. The buyback can continue at $6 billion annually for years without requiring a single additional dollar of operating cash flow. The credit rating is investment grade and management has explicitly committed to maintaining it.

Putting it together

If you own PayPal at $50, here is what is happening to your ownership stake mechanically, assuming the buyback continues at the guided pace:

• Year 1: your stake gets larger by ~13% (share count drops)

• Year 2: another ~13%

• Year 3: another ~13%

Compounded, after three years, you own roughly 33% more of the company than you did at purchase. If FCF holds flat at $6B, your FCF per share goes from $6.50 today to $9.70. If the multiple holds at 9x, the stock is at $87. If the multiple expands modestly to 12x — still well below the historical average — the stock is at $116. Both outcomes assume the operational story does nothing to help.

This is what people mean when they say a stock is “cheap.” Cheap doesn’t mean low absolute price. Cheap means the math works even when the narrative doesn’t.

Long-term targets management has actually committed to

The 2025 proxy laid out explicit medium-term targets that are still operative under Lores. These are worth knowing because they frame the stock-price hurdles in his comp package:

• Total Payment Volume: grow faster than e-commerce over time

• Branded checkout TPV: grow in line with e-commerce (so roughly mid-to-high single digits)

• Transaction Margin Dollars: high single-digit growth by 2027, with longer-term ambition for 10%+

• Operating expense: grow at less than half the rate of TM$ growth

• Non-GAAP EPS: low-teens+ growth by 2027, with ambition for 20%+ longer-term

• FCF: grow in line with net income

• Capital return: distribute 70-80% of FCF annually, with near-term focus on buybacks

If Lores delivers even the lower end of these targets, the stock should approach his $68 threshold hurdle without breaking a sweat. If he delivers the upper end, $100 stretch is in play. The targets and the comp package are pointed at the same outcome.

That is rare. Most management comp packages are decoupled from public targets in ways that make them functionally meaningless. This one is designed to align.

The valuation case, three ways

The simple way. Market cap ~$46B. Free cash flow ~$6B. That’s a 13% FCF yield on a business with a net cash balance sheet, a global brand with 438 million active accounts, and a CEO with a credible turnaround track record. Even if FCF declines 20% over five years, you’re still earning a yield well above the market.

The buyback math. If PayPal repurchases $6B annually at an average price of $55 over the next three years, the share count drops from ~920M today to roughly ~600M by year-end 2028. If FCF holds flat at $6B, FCF per share goes from ~$6.50 today to ~$10. At even a 10x multiple, that’s $100/share — a double from here, with no help from the operational turnaround. This is, not coincidentally, exactly the stretch hurdle in Lores’ comp package.

The operational re-rating. If Lores executes anything close to his HP playbook and branded checkout stabilizes at 4-5% growth — half of what it was at the end of 2024 — non-GAAP EPS could compound at low-to-mid teens over three years. The 2025 plan called for $4.95 to $5.10 in 2025 EPS; if 2028 EPS reaches $7.00 and the multiple expands from 9x to 13x (still below the 10-year average), the stock approaches $90-95. Add the buyback effect on share count and you can see how $100 becomes the base case rather than the bull case.

The base case and bull case rhyme because the buyback does so much of the work. The bear case requires the buyback to actually break, which requires the cash flow to break, which requires multiple parts of the business to deteriorate simultaneously while a CEO with a strong track record fails to course-correct. That’s possible. It’s not the most likely path.

The catalyst

There are four catalysts worth watching in 2026:

1. May 5, 2026: Q1 earnings. This is Lores’ first quarter — even though he only took the seat March 1, the print and the call will be his first major message to the market. New CEOs almost always kitchen-sink and reset. Expect conservative full-year guidance, possibly an explicit reset of branded checkout expectations, and a clearer articulation of the 2026 reinvestment plan. Sentiment is so bad that simply stabilizing branded checkout growth at 1-2% would be received as good news.

2. Lores’ first strategic update (likely Q2 or Q3). New CEOs typically announce a strategic plan within their first 90-120 days. The HP playbook he ran was unsexy and effective: simplify, focus, cut cost, reinvest selectively, return cash. Apply that to PayPal and you get a credible 18-month story. Watch for explicit deprioritization of any product areas — Lores is famous for being willing to cut things, which is something Chriss was reportedly slower to do.

3. Branded checkout inflection. The $400M reinvestment is going into the checkout experience, biometric/passkey adoption (currently around 36%), and merchant-side improvements. If branded TPV growth re-accelerates from 1% back to mid-single digits — even just halfway back — the multiple re-rates. You won’t see this in Q1; you might see early signs in Q2 or Q3.

4. Continued buyback execution. The boring catalyst, but the most reliable one. Every quarter that produces another $1.5B in buybacks at sub-$55 prices is a quarter where the math compounds quietly in your favor. Management has guided to $6B for 2026 and has $15+ billion of authorization remaining.

The asymmetry is the point. To lose meaningfully from here, you need branded checkout to actively decline (not just decelerate), the buyback to be cancelled, and Venmo/BNPL/merchant services to all stall. That’s possible, but it requires several bad things to happen simultaneously to a business still generating $6B in annual cash, with a new CEO whose entire $25-62M upside depends on the stock at least getting back to its 2024 levels.

The risks worth respecting

I want to be fair to the bear case, because the bears have been right for five years.

• Take rate compression is real. As Braintree (low-margin) grows faster than branded (high-margin), the blended take rate falls. This is a slow grind on margins that doesn’t go away with a new CEO.

• Apple Pay and Shop Pay aren’t going away. They are getting better, and they have distribution PayPal can’t match. Apple Pay is the default behavior on every iPhone shipped in the world.

• Venmo monetization keeps getting promised and partially delivering. The Venmo debit card growth (TPV +50%) is genuinely encouraging, but the broader “Venmo as a second growth engine” story has been “next year” for a long time.

• Agentic commerce is a real wildcard. If AI agents do the buying, the checkout button matters less. PayPal has hedges — a stablecoin (PYUSD) and an AI API — but it’s early. This is also the risk that bears bring up most often without a clear timeline, which makes it both real and difficult to position against.

• CEO transitions create real execution risk. Lores is unproven at PayPal specifically. Five years on the board helps but doesn’t substitute for operating experience inside the company. The HP playbook may not translate cleanly to a payments business with very different competitive dynamics.

• The share-count math depends on the stock staying cheap. Ironically, the buyback thesis works better if the operational story takes longer to fix, because the company keeps repurchasing at low prices. If the stock re-rates quickly, the buyback’s per-share impact diminishes. This is a strange thing to wish for, but it’s how the math works.

• Macro risk. PayPal is consumer-spend-exposed. A real recession could pressure TPV across the board, and FCF guidance assumes a relatively normal consumer environment.

The honest version of this thesis is: I don’t know whether branded checkout reaccelerates. Nobody does, including Lores on day one. But I don’t need to know, because the buyback gives me time to find out — and the new management is being paid in a way that makes their interests indistinguishable from mine.

Bottom line

PayPal at 9x earnings — with a 13% FCF yield, retiring 13% of its shares annually, sitting on a net-cash balance sheet, with a respected operator just installed as CEO whose entire $25-62 million upside is gated on the stock recovering — is the kind of setup value investors wait years for. It’s cheap because the business has real problems. It’s investable because the capital allocation works regardless of whether those problems get fully fixed, and because the people running it now are paid only to fix them.

The market is pricing in continued decline. The buyback is pricing in survival. The CEO comp package is pricing in a double.

I think survival is a low bar to clear. The rest is upside.

Disclosure: This is a newsletter, not investment advice. Do your own work. Numbers are pulled from PayPal’s most recent SEC filings (8-Ks, 10-K, DEF 14A proxy), earnings releases through Q4 2025, the offer letter for Enrique Lores filed February 2026, and analyst commentary as of late April 2026. Stock prices, share counts, and authorization balances move; verify before acting. I may or may not own PYPL — disclose your own position before publishing.

As of April 14, 2026, NOW trades at roughly $89 per share on a split-adjusted basis, a market cap of approximately $93 billion against a business that just reported $13.3 billion in trailing revenue, $4.6 billion in free cash flow, and guided 2026 subscription revenues of $15.5 billion at 19.5 to 20% constant-currency growth. The stock is down roughly 45% from its highs. The IGV software ETF is off about 27% from its September 2025 peak. UBS just downgraded NOW to Neutral with a $100 target, citing AI disruption risk, and the SaaS-pocalypse narrative is eating everything in sight.

This is the article I want to write. Not because it is easy or because the conclusion is obvious. The bear case here is real, is thoughtful, and needs to be engaged with seriously before arriving at any conviction. But after reading every available annual report, every material quarterly filing, all of the investor materials, and mapping what AI actually does to ServiceNow’s architecture, I think the market has the story almost exactly backwards. ServiceNow is not the victim of the AI wave. It is one of the two or three companies best positioned to be its landlord.

Here is the case.

What ServiceNow Actually Is

Most descriptions of ServiceNow get it subtly wrong. They call it an IT service management company, or a workflow automation company, or an enterprise SaaS company. All of those are accurate but they miss the thing that matters.

ServiceNow is the system of record and system of action for how large enterprises get work done. It is not a database. It is not a chat tool. It is not a ticketing system, though it started as one. It is the connective tissue that sits between an enterprise’s people and its systems, and it owns that layer with a depth of integration that has no real parallel in software.

The company was founded in 2004 by Fred Luddy on a simple premise: enterprise software was designed for computers, not for people, and the result was that work got lost in email, phone calls, and spreadsheets rather than flowing through defined processes. The Now Platform was his answer. It gave large organizations a single place where any request, any incident, any change, any workflow could be tracked, routed, automated, and resolved. The user interface was built to feel like a consumer app. The backend was built to be the most powerful workflow engine ever made for the enterprise.

It worked. By the time Bill McDermott arrived as CEO in January 2020, ServiceNow already had about $3.5 billion in revenue and a reputation as the stickiest software in enterprise IT. McDermott, who built SAP into a global powerhouse, took one look at what ServiceNow had and immediately saw something bigger: this was not an IT company. It was a platform that could run every workflow in every department of every large organization on earth. The Total Addressable Market was not the $20 to $30 billion ITSM market. It was a TAM that analysts now size at $275 billion and growing.

The expansion has been methodical. ServiceNow started with ITSM (IT service management), moved into ITOM (IT operations management), then into HR service delivery, then customer service management, then security operations, then legal service delivery, then field service management, then industry-specific modules for financial services, healthcare, manufacturing, telecom, and government. Each expansion leverages the same platform, the same data model, the same workflow engine. Customers who start with ITSM find that adding HR or CSM costs them almost nothing in integration work because it all runs on the same Now Platform they already deployed. This is a platform-extension flywheel that compounds with every new module.

The Financials: A Decade of Compounding You Cannot Argue With

Let me walk through the numbers with discipline, because they are genuinely extraordinary.

ServiceNow crossed $1 billion in revenue in 2015. It was the second enterprise SaaS company ever to do so, after Salesforce. From there, growth has been relentless.

Revenue trajectory (all figures in billions, rounded):

• 2015: $1.0B

• 2016: $1.4B

• 2017: $1.9B

• 2018: $2.6B

• 2019: $3.5B

• 2020: $4.5B

• 2021: $5.9B

• 2022: $7.2B

• 2023: $9.0B

• 2024: $11.0B

• 2025: $13.3B

That is a compound annual growth rate of approximately 30% from 2016 to 2025. At $13 billion in annual revenue. There is not another enterprise software company at this scale growing this consistently. Salesforce grew at this rate for a period, then decelerated meaningfully as it scaled. Oracle never did it. SAP never did it. Microsoft did it, but Microsoft is a different animal entirely. ServiceNow’s growth rate has barely bent in a decade.

More importantly, look at what the profitability profile looks like now:

Full year 2025 results:

• Total revenue: $13.3B (up 21%)

• Subscription revenue: approximately $12.8B

• Non-GAAP subscription gross margin: approximately 85%

• Non-GAAP operating margin: 31% (up 150 basis points year-over-year)

• Free cash flow: $4.6B (up 34% year-over-year)

• Free cash flow margin: 35% (beat guidance by 100 basis points)

This is a business that generates $4.6 billion in free cash flow on $13 billion in revenue. That is a 35% FCF margin at $13 billion of scale, still growing at 21%. The company has over $10 billion in cash and investments on the balance sheet against roughly $2.4 billion in debt, so net cash is around $7.65 billion.

2026 guidance: Subscription revenue of $15.53 to $15.57 billion (19.5 to 20% constant-currency growth). Operating margin of 32% (up 100bps). Free cash flow margin of 36% (up 100bps). The company is expanding margins while compounding revenue at 20%-plus. The Rule of 54 (revenue growth rate plus FCF margin) that management cites is not marketing puff. It is a real measurement of capital efficiency that puts ServiceNow in a category of one among companies at this scale.

Now look at the contracted revenue visibility. As of Q4 2025, remaining performance obligations (RPO) stood at $28.2 billion, up 26.5% year-over-year. Current RPO, which is revenue contracted and expected to be recognized within the next 12 months, stood at $12.85 billion, up 25% year-over-year and 21% in constant currency. This $12.85 billion current RPO effectively pre-loads 2026 revenue. ServiceNow enters every year with more than 90% of the following year’s revenue already under contract.

The customer base is also telling. 2,109 customers spend more than $1 million annually with ServiceNow, up 12% year-over-year. Nearly 600 customers spend more than $5 million annually. The net new ACV growth is accelerating, not decelerating. The renewal rate is consistently 98 to 99%. When a customer renews at 99% and 70% of those renewals expand the contract value, you have a machine that compounds revenue per customer while also adding new customers. The net revenue retention rate sits above 120%.

This is the financial profile of one of the best enterprise software businesses ever built.

The Moat: Why Nobody Leaves

The renewal rate of 98 to 99% is not a marketing number. It is an economic fact that tells you everything about the nature of ServiceNow’s competitive position.

When a large enterprise deploys ServiceNow, they do not just install software. They rebuild their entire process architecture around the Now Platform. The company’s ITSM module becomes the nervous system of IT. The HRSD module becomes how every employee interaction with HR happens. The CSM module becomes how customer service flows. The CMDB (Configuration Management Database) becomes the system of record for every IT asset in the enterprise. After a full deployment, ServiceNow is not software that a company uses. It is the operating system that the company runs on.

This creates switching costs that are among the highest in all of enterprise software. Replacing ServiceNow means rearchitecting every workflow, retraining every employee, migrating a decade of process data, and reconfiguring every integration to every other system in the enterprise. Independent estimates suggest a full ServiceNow replacement project takes two to four years and costs tens of millions of dollars in consulting fees, migration work, and productivity loss. The typical Fortune 500 company is not going to go through that. It is simply not a rational economic decision.

Morningstar assigns ServiceNow a “wide” economic moat, which is their highest rating. They derive it specifically from switching costs. This is correct, but it understates the full picture. ServiceNow’s moat is actually three-layered.

The first layer is the switching cost moat described above. The second layer is the platform-network moat. ServiceNow has built an ecosystem of over 1,800 certified third-party applications in its App Store. It has deep integrations with Microsoft, Salesforce, SAP, Oracle, AWS, Google Cloud, and every major enterprise software platform. Each integration makes ServiceNow more central to the enterprise IT architecture and makes it harder to remove without destroying all of those adjacent integrations as well. The third layer is what might be called the data moat. ServiceNow’s CMDB contains a live, constantly updated map of every technology asset and process relationship in the enterprise. This data is irreplaceable. You cannot copy it out and hand it to a competitor, because the competitor’s platform would not know what to do with it. The data is native to ServiceNow’s architecture.

Add to this the fact that over 85% of the Fortune 500 already uses ServiceNow, and nearly 60% of the Global 2000, and the moat becomes self-reinforcing from a sales perspective. Every new enterprise executive who joins a company that uses ServiceNow and loved it at their last company is a natural expansion vector. The global partner ecosystem, including Accenture, Deloitte, KPMG, Cognizant, and EY, all have tens of thousands of certified ServiceNow practitioners whose livelihoods depend on the platform growing. The ecosystem is not just a distribution channel. It is a structural moat that makes it harder for competitors to get a foothold.

None of this is to say ServiceNow has no competition. BMC, Ivanti, Jira Service Management, and Freshservice all compete for pieces of the ITSM market. Salesforce competes for customer service workflows. Workday competes for HR service delivery. Microsoft, with its Azure and Power Platform ecosystem, is the most formidable potential competitor given its own deep enterprise integration. But the evidence from ten years of 98 to 99% renewal rates across a $13 billion revenue base is that none of these competitors have found a way to dislodge ServiceNow once it is in place. The switching costs are too high.

AI: Threat or Tailwind?

This is the question the market is pricing as existential right now, and it deserves a careful answer.

The bear case, as articulated by UBS analyst Karl Keirstead in his April 2026 downgrade, runs as follows. AI agents from Anthropic, OpenAI, and others can now perform workflow automation tasks autonomously. If a Claude agent can handle an IT help desk ticket, route it, resolve it, and close it without any human involvement, do enterprises need ServiceNow’s seat-based licenses? Budget is finite. As AI infrastructure spend inflects in 2026, traditional software spend is under pressure. The “seat compression” narrative holds that AI reduces the number of humans in the loop, and therefore reduces the number of seats that per-user SaaS models can sell. This is a real risk for companies whose pricing is tied to human headcount.

The problem with applying this thesis to ServiceNow specifically is that it misunderstands what ServiceNow actually does. ServiceNow is not a human replacement software company. It is not a tool that helps humans do manual work more efficiently. It is a workflow orchestration platform, and AI agents need orchestration more, not less.

Consider what happens when an enterprise deploys AI agents to handle IT tickets. The agent receives the ticket. It diagnoses the issue. It attempts a resolution. If it cannot resolve the issue, it escalates to a human. It creates a record of the interaction. It updates the knowledge base. It triggers compliance logging. It interfaces with other systems (Active Directory to reset a password, an asset management system to check if the user’s hardware is under warranty, a change management system to record what was changed). Where do all of these actions happen? On ServiceNow. ServiceNow is not the thing being replaced by the AI agent. ServiceNow is the platform that the AI agent runs on.

In fact, this is precisely what ServiceNow’s own AI agents do internally. The company has disclosed that AI agents now resolve 90% of its internal IT support requests and 89% of customer support requests autonomously. Resolution times have fallen sevenfold. ServiceNow is not a company that AI automation is threatening. It is a company that AI automation is demonstrating as the platform of choice for deploying AI at enterprise scale.

The product evidence supports this. Now Assist, ServiceNow’s GenAI product line that launched in 2023, is described as the fastest-growing product introduction in company history. By December 2024, ServiceNow had nearly 1,000 AI customers and had already surpassed $200 million in annual contract value for its Pro Plus AI solution. Now Assist service desk deals grew over 150% quarter-over-quarter in Q4 2024. In Q4 2025, Now Assist net new ACV more than doubled year-over-year. The AI products are not cannibalistic to the core platform. They are incremental and highly accretive.

The agentic AI strategy is even more compelling. ServiceNow launched AI Agent Orchestrator, AI Agent Studio, and thousands of pre-built AI agents in March 2025 with the Yokohama platform release. The AI Agent Orchestrator allows enterprises to manage teams of AI agents from different vendors (including Anthropic and OpenAI) working together across complex, multi-step enterprise workflows. This is the explicit counterplay to the “AI disruption” narrative: ServiceNow is positioning itself as the control tower where all enterprise AI agents, whether built by ServiceNow or third parties, are governed, monitored, and orchestrated.

The Moveworks acquisition for $2.85 billion, which closed in December 2025, deepens this strategy. Moveworks had built the leading AI assistant and enterprise search product for employees, with nearly 5.5 million enterprise users across 350+ customers. Its technology adds a natural-language front door to ServiceNow’s automation engine: instead of navigating a portal, an employee asks a question and Moveworks’ AI resolves it by triggering the right ServiceNow workflow. The combination of Moveworks (conversational AI front end and enterprise search) plus ServiceNow (workflow execution and data backbone) is a full-stack AI work platform. The combined entity is well ahead of anything Anthropic or OpenAI has commercially deployed in enterprise workflow automation.

The partnership ecosystem with AI hyperscalers also reflects strategic positioning rather than defensiveness. ServiceNow announced partnerships with Anthropic and OpenAI in January 2026 to integrate their models into the ServiceNow platform. Notably, these AI companies are partnering with ServiceNow rather than trying to displace it. They are not building enterprise workflow automation platforms. They are building models and wanting them to be used in enterprise workflows. ServiceNow is the thing they need to partner with to get enterprise distribution.

The consumption-based monetization transition is also worth noting. Beginning in 2025, ServiceNow began embedding AI Agents into its Pro Plus and Enterprise Plus SKUs rather than charging separately for them. Management was explicit about the rationale: accelerate adoption now, monetize increasing usage over time as consumption-based pricing phases in. This is the same playbook Snowflake, Twilio, and AWS ran. Forgo short-term revenue recognition to build usage habits, then capture the monetization as the usage base scales. The free cash flow guidance going to 36% in 2026, with further expansion expected, suggests management has high confidence the transition is accretive long-term.

The UBS concern about “seat compression” is valid for companies whose entire value proposition is human productivity tooling at the seat level. It is not valid for ServiceNow, whose value proposition is process orchestration across the entire enterprise architecture. If anything, the deployment of AI agents multiplies the volume of workflows, API calls, data events, and automation actions that flow through ServiceNow’s platform. That is revenue growth, not compression.

The TAM is expanding, not contracting. Analysts now size ServiceNow’s addressable market at $275 billion by 2026, up from $220 billion, driven by expansion across ITSM, CSM, HR, and AI. The company sits at $13.3 billion in revenue against a $275 billion TAM. Penetration is still under 5%.

Asset Underwrite: What Is This Company Worth?

Balance Sheet: The company holds approximately $10.3 billion in cash and investments against $2.4 billion in total debt. Net cash position is roughly $7.65 billion, or approximately $7.38 per share (split-adjusted). The balance sheet is pristine. There are over 2,000 U.S. and foreign patents, a global partner ecosystem representing effectively incalculable enterprise distribution value, and a brand that is the gold standard of enterprise ITSM. These are hard assets that do not appear on the balance sheet.

Revenue Visibility: Current RPO of $12.85 billion is contracted revenue expected to be recognized in the next 12 months. This $12.85 billion gives us enormous visibility into 2026 revenue. The company has guided $15.5 billion in subscription revenue for 2026 and has a track record of beating guidance consistently for years.

Customer Cohort Economics: With 2,109 customers at over $1 million ACV and 600 approaching $5 million or more, the upsell trajectory within the existing base alone (70% of renewing customers grow their ACV) represents a multiyear growth engine that is entirely separate from new customer acquisition.

Now let me build the DCF:

I will run three scenarios: Base, Bull, and Bear.

Assumptions across all scenarios:

• Weighted average cost of capital (WACC): 10%

• Terminal growth rate: 3.5%

• Shares outstanding: approximately 1.04 billion (split-adjusted)

• Net cash: $7.65 billion

BASE CASE (most likely scenario)

Revenue assumptions:

• 2026E: $16.0B (guided $15.5B subscription plus professional services, slight beat)

• 2027E: $19.0B (18.5% growth, mild deceleration)

• 2028E: $22.3B (17.5% growth)

• 2029E: $25.8B (16% growth)

• 2030E: $29.2B (13.5% growth)

• Terminal year growth: 3.5%

FCF margin assumptions:

• 2026E: 36% (guided)

• 2027E: 37.5%

• 2028E: 39%

• 2029E: 40%

• 2030E: 41%

Free Cash Flow:

• 2026E: $5.76B

• 2027E: $7.13B

• 2028E: $8.70B

• 2029E: $10.32B

• 2030E: $11.97B

Terminal value (Gordon Growth, 3.5% terminal growth, 10% WACC):

• Terminal FCF: $11.97B x 1.035 / (0.10 - 0.035) = $190.5B

• Terminal value PV: $190.5B / (1.10)^5 = $118.4B

Sum of discounted FCFs (2026-2030):

• 2026: $5.76B / 1.10 = $5.24B

• 2027: $7.13B / 1.21 = $5.89B

• 2028: $8.70B / 1.33 = $6.54B

• 2029: $10.32B / 1.46 = $7.07B

• 2030: $11.97B / 1.61 = $7.43B

• Sum: $32.17B

Total Enterprise Value: $32.17B + $118.4B = $150.6B Add net cash: $150.6B + $7.65B = $158.2B Per share intrinsic value: $158.2B / 1.04B shares = $152 per share

BULL CASE (AI amplifies growth and margins)

In the bull case, Now Assist and the agentic AI product line accelerate revenue growth rather than decelerate it. Consumption-based AI pricing begins flowing meaningfully into revenues by 2027. The Moveworks integration proves transformational for CRM penetration, adding a new growth vector. FCF margins expand faster as AI internally automates a significant portion of G&A and support costs.

Revenue assumptions:

• 2026E: $16.3B

• 2027E: $20.0B (22.5% growth, reacceleration)

• 2028E: $24.2B (21% growth)

• 2029E: $28.9B (19.5% growth)

• 2030E: $34.1B (18% growth)

FCF margins:

• 2026E: 37%

• 2027E: 40%

• 2028E: 43%

• 2029E: 45%

• 2030E: 47%

Free Cash Flow:

• 2026E: $6.03B

• 2027E: $8.00B

• 2028E: $10.41B

• 2029E: $13.01B

• 2030E: $16.03B

Terminal value (3.5% growth, 10% WACC):

• Terminal FCF: $16.03B x 1.035 / 0.065 = $255.1B

• Terminal PV: $255.1B / (1.10)^5 = $158.5B

Sum of discounted FCFs:

• 2026: $5.48B

• 2027: $6.61B

• 2028: $7.83B

• 2029: $8.91B

• 2030: $9.96B

• Sum: $38.79B

Total Enterprise Value: $38.79B + $158.5B = $197.3B Add net cash: $197.3B + $7.65B = $204.9B Per share intrinsic value: $204.9B / 1.04B = $197 per share

BEAR CASE (AI compression materializes, growth decelerates sharply)

In the bear case, the UBS thesis is right. Budget compression in non-AI software is real, significant, and sustained. cRPO growth decelerates to 16% by end of 2026 as UBS predicts. CSM, which represents roughly 10% of revenue, suffers measurable seat compression as AI agents replace customer service headcount. Operating margin expansion slows or stalls as the company invests heavily to stay relevant. However, the core ITSM and ITOM franchise holds because the switching costs are too high to overcome even in a difficult environment.

Revenue assumptions:

• 2026E: $15.5B (at the low end of guidance, no beat)

• 2027E: $17.7B (14% growth)

• 2028E: $19.8B (12% growth)

• 2029E: $21.6B (9% growth)

• 2030E: $23.4B (8% growth)

FCF margins:

• 2026E: 33% (investment headwinds)

• 2027E: 33.5%

• 2028E: 34%

• 2029E: 35%

• 2030E: 36%

Free Cash Flow:

• 2026E: $5.12B

• 2027E: $5.93B

• 2028E: $6.73B

• 2029E: $7.56B

• 2030E: $8.42B

Terminal value (2.5% terminal growth, reflecting slower long-term outlook):

• Terminal FCF: $8.42B x 1.025 / (0.10 - 0.025) = $115.0B

• Terminal PV: $115.0B / (1.10)^5 = $71.4B

Sum of discounted FCFs:

• 2026: $4.65B

• 2027: $4.90B

• 2028: $5.06B

• 2029: $5.18B

• 2030: $5.23B

• Sum: $25.02B

Total Enterprise Value: $25.02B + $71.4B = $96.4B Add net cash: $96.4B + $7.65B = $104.1B Per share intrinsic value: $104.1B / 1.04B shares = $100 per share

The current market price of approximately $89 per share is below even the bear case intrinsic value of $100. In the bear case, which assumes cRPO growth falls to UBS’s prediction, CSM suffers real seat compression, and revenue growth decelerates to single digits by 2029, ServiceNow is still worth more than it is trading for today.

At the base case, the stock is worth $152. At the bull case, $197. The probability-weighted expected value, assigning rough weights of 25% bear, 55% base, and 20% bull, is:

($100 x 0.25) + ($152 x 0.55) + ($197 x 0.20) = $25 + $83.60 + $39.40 = $148 per share

Against a current price of $89, that is a 66% expected return before time value.

Price Targets and Positioning

12-month price target: $130 to $145

This is based on the expectation that Q1 2026 results (reporting April 22, 2026) will show continued beat-and-raise execution, that the SaaS sentiment cycle turns as the AI narrative matures from disruption panic to actual implementation realities, and that the company’s own AI monetization begins showing up more clearly in the numbers through 2026.

2-year price target: $160 to $180

3-year price target: $200 to $230

The range widens with time as the AI monetization uncertainty resolves in one direction or another. The numbers above are pre-split-adjusted equivalents based on the current share count of approximately 1.04 billion shares.

At $89, the stock trades at approximately 17x 2026E free cash flow. The historical trading range for ServiceNow has been 30 to 60x FCF. Even at 25x 2026E FCF of $5.76B, the stock would be worth $138. The multiple compression from 40-50x to 17x has been extreme and appears to reflect peak AI disruption fear rather than a fundamental repricing of the business based on actual financial results.

The Risks I Take Seriously

Any honest investment case needs to grapple with what could make it wrong. Here are the risks I think are actually material, not the ones that show up in the 10-K boilerplate.

The budget compression risk is real. The UBS analyst is not making this up. Enterprise CIOs are reallocating budgets toward AI infrastructure and data platforms. In some cases, that is at the expense of traditional software renewal and expansion. The key question for ServiceNow specifically is whether this compression affects renewal rates or expansion rates within existing contracts. If 98 to 99% renewal holds but expansion slows, revenue growth moderates. If renewal rates tick down even 100 basis points, that is an enormous headline and a major re-rating catalyst. I am watching the Q1 and Q2 2026 earnings for any signal in cRPO growth rates below guidance.

Microsoft is a genuine long-term threat. Copilot embeds AI workflow automation into the Microsoft 365 ecosystem that every enterprise already runs. If Microsoft decides to build deep ITSM and process automation into Teams and Copilot, it competes directly with ServiceNow’s core value proposition. The integration tax for ServiceNow is high, but the integration tax for replacing Teams or Office is higher. Microsoft has every incentive to expand up the stack. ServiceNow and Microsoft have a partnership (the two platforms integrate and there are joint go-to-market agreements), but Microsoft is also the most natural existential threat in the category.

The consumption-based pricing transition creates near-term revenue recognition uncertainty. By embedding AI agents into Pro Plus and Enterprise Plus SKUs rather than charging separately, ServiceNow deferred incremental near-term ACV to build usage. If usage-based monetization takes longer to materialize than expected, the 2027 revenue numbers could disappoint relative to what a simpler seat-expansion model might have produced.

Execution risk in the Moveworks integration. The $2.85 billion acquisition is the largest in ServiceNow’s history and was paid at approximately 20x revenue. Integration of an AI-native company into a workflow platform company is not trivial. Retention of Moveworks’ engineering talent, harmonization of the product roadmap, and actual delivery of the combined agentic platform are all execution risks over the next 18 to 24 months.

The Verdict

ServiceNow is one of the ten best businesses in American software. It has a decade of compounding financial performance, a switching-cost moat that is wide and defensible, a TAM that is expanding not contracting, a product line that is ideally positioned to benefit from rather than be disrupted by agentic AI, and a balance sheet with $10 billion in cash.

The stock is down 45% from its highs because the market has priced in a SaaS-pocalypse scenario where AI agents make enterprise workflow software obsolete. That scenario misunderstands ServiceNow’s role in the enterprise stack. ServiceNow is not the software that AI replaces. ServiceNow is the platform that AI runs on.

At $89 per share, with a bear case intrinsic value of $100 and a base case of $152, the risk-reward is the most compelling it has been in years. The market is offering you the chance to buy a business that generates $4.6 billion in free cash flow, is contracted for $12.85 billion in revenue over the next 12 months, and has guided 20% growth in 2026 at a price that implies roughly 17x that free cash flow.

This is not a perfect business. The AI transition creates genuine short-term uncertainty about the pace of expansion revenue. Microsoft remains a structural shadow. The consumption monetization transition introduces some near-term opacity into financial modeling. But these are manageable uncertainties in a business that has consistently delivered at or above guidance for years, through COVID, through interest rate cycles, through competitive challenges from BMC and Salesforce and Microsoft. The operational excellence at ServiceNow is not accidental. It is institutional.

The stock will likely remain volatile through the first half of 2026 as the broader SaaS sector digests the AI narrative. Q1 2026 earnings on April 22nd will be important. But for investors with a 2 to 3 year time horizon and the ability to sit through noise, NOW at $89 is a compelling entry point.

My conclusion is unambiguously bullish. The company is worth somewhere between $130 and $200 per share depending on how the AI monetization transition plays out. At $89, you are being paid to own one of the best franchise businesses in enterprise software at a once-in-a-decade discount.

Disclosures and Important Notes

This article represents the author’s independent analysis and opinion. It is not investment advice. All financial figures are sourced from ServiceNow SEC filings (10-K, 10-Q, 8-K), ServiceNow investor relations materials, and publicly available research. The DCF model presented here makes numerous assumptions that could prove incorrect. Readers should conduct their own due diligence and consult a qualified financial advisor before making any investment decision. All share prices and market data are as of approximately April 14, 2026, and reflect the split-adjusted share price following ServiceNow’s December 2025 5-for-1 stock split.

Sources: ServiceNow 10-K filings 2015-2024, Q4 and Full-Year 2025 8-K earnings release (January 28, 2026), Q1 2025 earnings release (April 23, 2025), Q2 2025 earnings release (July 2025), Q3 2025 earnings release (October 2025), ServiceNow investor presentations, Morningstar research, UBS analyst note (April 2026), Wikipedia ServiceNow corporate history, Everest Group Moveworks acquisition analysis, Cyntexa ServiceNow statistics report.

The market is penalizing ADSK for near-term disruption from a sales force restructuring while ignoring the structural margin expansion story unfolding in real time. This is a classic case of Wall Street confusing short-term friction with long-term impairment. We believe ADSK deserves a 43%+ rerating from current levels over the next 12-18 months — and that is a conservative base case.

This is not a momentum bet. It is a value and quality compounding story in one of the most defensible software franchises on earth.

What Autodesk Actually Does — And Why It Matters

Autodesk has been building software for people who design and make things since 1982. AutoCAD, the company’s flagship product, became the standard drafting tool for architects, engineers, and designers globally over four decades. That dominance is not nostalgia — it is an installed base moat that takes a generation to dislodge.

The company now serves four primary verticals through a cloud-first subscription platform:

• Architecture, Engineering, Construction & Operations (AECO) — $3.6B in FY26 revenue, 22% growth. This is the crown jewel: Revit for BIM, AutoCAD Civil 3D for infrastructure, and the Autodesk Construction Cloud (ACC) suite anchored by Autodesk Build, BuildingConnected, and BIM Collaborate Pro.

• AutoCAD & AutoCAD LT — $1.8B in FY26, 14% growth. The cash cow. Hundreds of millions of professionals have been trained on AutoCAD. The switching cost is not software cost — it is human capital.

• Manufacturing (MFG) — $1.4B in FY26, 16% growth. Fusion (formerly Fusion 360) has emerged as a genuine threat to the Dassault/PTC duopoly in mid-market CAD/CAM/CAE. Inventor continues to serve industrial manufacturing.

• Media & Entertainment — $332M in FY26, 5% growth. Maya and 3ds Max dominate high-end animation and VFX. Slower growth, but sticky and margin accretive.

The genius of the business model transformation completed over the past five years is this: Autodesk converted virtually its entire revenue base from perpetual licenses to cloud subscriptions. Today, 97% of revenue is recurring. The customer is locked in not through contractual force but through workflow integration — thousands of project files, templates, trained teams, and supply chain integrations built on Autodesk’s formats. You don’t switch away from that on a Tuesday.

The Financial Transformation: Four Years of Evidence

The numbers tell a story of deliberate, sustained compounding. Here is the progression across four annual reports:

That 54% free cash flow surge in FY26 is the headline number that the market has not yet priced. It is not a one-time event — it reflects the normalization of billing patterns after a multi-year transition from upfront multi-year contracts to annual billing, and the first wave of margin expansion from go-to-market restructuring. The FY27 FCF guidance of $2.7-2.8B represents another 15%+ increase.

Subscriptions grew to 7.79 million in FY25, up ~516K year-over-year. Remaining Performance Obligations (RPO) hit $8.3 billion in Q4 FY26, up 20% year-over-year, with current RPO of $5.5B. That is nine months of forward revenue visibility already contracted. Businesses do not build this kind of backlog unless customers are deeply committed.

The Elephant in the Room: The Accounting Investigation — And Why It’s Over

In April 2024, Autodesk disclosed an internal investigation into its free cash flow and non-GAAP operating margin practices. The stock fell sharply. The SEC and the U.S. Department of Justice for the Northern District of California both opened inquiries. Activist investor Starboard Value, which held over $500 million in ADSK shares, issued a scathing letter accusing management of manipulating metrics.

Here is what actually happened, and why the overhang has fully cleared:

The investigation found that management had, during FY2022 and FY2023, pushed enterprise customers toward multi-year upfront billing arrangements to help meet free cash flow targets — a practice that was disclosed in aggregate terms but not quantified. Management also timed certain discretionary spending and collections to influence reported free cash flow. This was poor governance and sloppy disclosure, but it was not accounting fraud. No revenue was fabricated. No expenses were hidden permanently. No restatement was required.

The CFO who oversaw the period (Deborah Clifford) was reassigned to a strategy role, then ultimately departed. In August 2025, both the SEC and the USAO formally closed their investigations with no charges and no penalties. The case is over.

The new CFO, Janesh Moorjani, who joined in December 2024 after a distinguished career at VMware and Nutanix, has explicitly flagged margin expansion as his primary mandate. His credibility is clean. The governance remediation has been completed. The class action lawsuits remain in early stages but are unlikely to represent material economic liability given the SEC/DOJ outcomes.

What the market did: it de-rated the stock on the investigation and never fully re-rated it on the resolution. That is the opportunity.

The New Transaction Model: Transition Friction Masking Structural Upside

Between FY2024 and FY2026, Autodesk executed a fundamental overhaul of how it transacts with customers — shifting from an indirect channel model where resellers and distributors handled billing, to a direct billing model where Autodesk invoices customers directly (the ‘new transaction model’).

This created significant noise in reported metrics: Billings were inflated during the rollout as the new model brought forward revenue recognition. GAAP revenue was temporarily elevated. And the percent of revenue through indirect channels fell from 63% in FY2024 to approximately 53% by FY2026 completion. The distortions made year-over-year comparisons nearly impossible to read cleanly.

As of Q4 FY2026, the new transaction model rollout is complete in all markets. By Q2 FY2027, the comparison period will also reflect the new model, eliminating the distortion entirely. This is the clarifying moment for investors. The underlying business, stripped of transition noise, is growing revenue at 8-10% organically and expanding margins aggressively.

The strategic benefits of direct billing are not cosmetic. Autodesk now has complete visibility into its end-customer base, pricing, and usage patterns for the first time in its history. This enables far more precise upselling, better renewal management, and — critically — the potential for consumption-based pricing models tied to AI usage. That option did not exist before.

The Margin Expansion Story: Best in the Industry Is the Target

This is the most underappreciated part of the thesis. Autodesk’s current non-GAAP operating margin of 38% sounds reasonable for a software company, but it is dramatically below what the business is structurally capable of.

Compare to peers: ServiceNow operates at ~44% non-GAAP operating margins. Cadence Design Systems runs at ~40%+. Oracle’s cloud business generates margins above 50%. Adobe generates 45%+ non-GAAP margins. Autodesk’s 38% reflects a business that has been investing heavily in cloud infrastructure, platform development, and a go-to-market model that had significant duplication between direct and channel layers.

Management has now explicitly guided to ‘GAAP margins among the best in the industry’ as the terminal target. For FY2027, GAAP operating margin guidance is 26-28% — a 400-600 basis point improvement over FY2026’s 22%. Non-GAAP operating margin guidance is 38.5-39%, up from 38%.

The restructuring completed in FY2025/FY2026 (approximately 9% of workforce across two tranches — ~1,350 in February 2025 and ~1,000 in early 2026) eliminated significant redundancy in customer success, marketing operations, and sales overlayers. The savings are flowing directly to operating leverage, not back into headcount.

Management plans to host an investor day in H2 FY2027 to lay out the full long-term margin roadmap. We believe this will be a catalyst. Our model suggests non-GAAP operating margins of 42-44% are achievable by FY2029, which on $9B+ of revenue would imply $3.8-4.0B of non-GAAP operating income and $3.5B+ of free cash flow.

The Growth Engines: Construction Cloud and AI

Autodesk Construction Cloud (ACC)

Construction is the biggest TAM expansion story in the company’s history. The global construction industry — a $13 trillion sector — has historically been one of the least digitized industries on earth. Productivity growth in construction has lagged every other major industry for 50 years. Autodesk’s ACC suite (Autodesk Build, BuildingConnected, BIM Collaborate Pro, Takeoff, Cost, and the recently acquired Rhumbix) is attacking this directly.

Make revenue — which captures ACC cloud subscriptions — grew 22% in FY2026 to $796M, accelerating from 25% in FY2025. This is the fastest-growing product family and now represents over 11% of total revenue, up from essentially zero in FY2020. The construction segment added nearly 400 net new logos in a single quarter. Major wins like Power Design and Cleveland Construction (displacing competitive solutions) demonstrate that Autodesk is now winning displacement deals, not just greenfield.

The ACC platform strategy creates compounding switching costs. A general contractor using Autodesk Build for project management, BIM Collaborate Pro for design coordination, BuildingConnected for preconstruction bidding, and GCPay for payments has their entire project life cycle on Autodesk’s platform. Moving that stack is a multi-year organizational transformation. The NRR holding in the 100-110% range despite macro headwinds in construction is evidence of this stickiness.

Project Bernini and the AI Opportunity

Autodesk’s AI strategy is not about chatbots layered on top of existing products. It is about fundamental workflow automation built on proprietary training data accumulated over 40 years of customers submitting projects through Autodesk’s cloud infrastructure.

Project Bernini — Autodesk’s generative 3D AI model — is the most visible piece of this. Bernini enables non-technical users to generate manufacturable 3D designs from text or 2D inputs, dramatically expanding the addressable user base beyond trained CAD professionals. The Fusion product’s AutoConstrain tool, which uses AI to suggest geometry constraints, has achieved a roughly 50% acceptance rate — meaning it is saving meaningful time on tasks that previously required manual expertise.

More importantly, Autodesk’s CFO confirmed in early 2026 that the company is developing additional foundation models trained on customer datasets that are delivering ‘stronger and more cost-effective results’ than general-purpose AI models for Autodesk’s specific domains. This is a key differentiation: generic AI cannot match domain-specific models trained on millions of real construction plans, manufacturing drawings, and architectural designs.

The monetization model for AI is still evolving, but the Flex consumption platform — token-based pay-as-you-go access — gives Autodesk a natural mechanism to charge for AI usage above base subscription tiers. As AI tools drive higher usage frequency and broader adoption, we expect Flex to become a material revenue contributor by FY2029.

The Moat: Four Concentric Rings of Defense

Autodesk’s competitive position is sometimes underestimated because the company has faced perpetual competitive threats — from Dassault, Bentley, PTC, and more recently Trimble — without losing meaningful market position. Here is why:

1. The Installed Base / File Format Lock-In

AutoCAD’s .DWG file format is the de facto standard for construction and engineering documentation. Hundreds of millions of drawings exist in this format. Every AEC firm has decades of project archives in DWG. Even if an architect wanted to switch to a competitor, their supplier chain — contractors, consultants, fabricators — uses Autodesk formats. This is a network effect disguised as a file format.

2. The Education Moat

Autodesk provides free software to over 100 million students and educators through its Education Community. The IIT Bombay memorandum of understanding signed in Q4 FY2025 is one example of hundreds of university partnerships globally. Engineers graduate knowing Autodesk products, firms hire them, and the cycle perpetuates. Competitors cannot buy this; it took 40 years to build.

3. The Enterprise Data Moat

The transition to cloud subscription has given Autodesk something it never had before: complete visibility into how customers use its products. Every click, every workflow, every collaboration event is now data. This enables AI training, predictive renewal analytics, and targeted upsell that competitors with legacy perpetual-license models simply cannot replicate.

4. The Platform Ecosystem

The Autodesk Platform Services (APS, formerly Forge) connects over 3,000 third-party developers building on Autodesk’s APIs. This creates a flywheel: more integrations make Autodesk more valuable, more value attracts more developers, and the ecosystem becomes its own switching cost. Bentley and Trimble have nothing comparable in construction.

Valuation: Cheap Relative to Quality and Growth

At approximately $237 per share and a market cap of roughly $50 billion, here is how ADSK stacks up:

A 19x forward non-GAAP P/E for a business growing free cash flow at 15-20% annually, with 97% recurring revenue, an 8.3B RPO backlog, dominant market share in structural growth markets, and AI optionality is genuinely cheap. For comparison: ServiceNow trades at 45-50x forward non-GAAP earnings. Salesforce trades at 28-30x. Workday trades at 28x. Even Cadence and Synopsys — slower-growth EDA companies — trade at 35-40x.

The consensus analyst price target is approximately $340-$360, representing 43-52% upside from current levels. We find the bull case — $380-400 — entirely reasonable if the FY2027 margin expansion plays out as guided and the AI narrative gets attached.

Our base case price target of $340 represents approximately 27x FY2027 non-GAAP EPS of $12.43. This is a modest premium to the market multiple for a company of this quality growing FCF at mid-teens annually. Even applying a 25x multiple — a discount to peers — implies a target above $310.

The Bull Case: $380

Upside scenario assumes: FY2027 margins beat guidance at 40%+, AI monetization begins to show up in billings growth re-acceleration, and construction recovery in Europe and the U.S. drives AECO above 25% growth. Multiple re-rates to 30x FY2027 non-GAAP EPS.

The Bear Case: $210

Downside scenario assumes: Sales force restructuring creates more disruption than expected, new transaction model causes customer attrition, macro deterioration in construction markets pressures net new logos, and margins disappoint. At $210, the stock trades at roughly 17x FY2027 non-GAAP EPS — implying the market views this as a zero-growth utility, which is not credible given the construction cloud trajectory.

What the Market Is Missing: The Five Mispricing Factors

1. The FCF Normalization is Being Misread as a One-Time Benefit

The 54% FCF surge in FY2026 is correctly attributed in part to the new transaction model bringing forward cash collections. But the base business FCF — stripping that out — is also growing 20%+ as margins expand. The market is discounting FY2027’s $2.75B FCF guidance as unsustainable. We believe it is the floor, not the ceiling.

2. The SEC/DOJ Closure is Underpriced

Both government investigations closed in August 2025 with zero penalties. The psychological discount applied during the investigation period never fully unwound. Investors who were burned by the disclosure are still on the sidelines. That is not rational capital allocation — it is recency bias creating an opportunity.

3. The Go-to-Market Disruption is Temporary and Self-Inflicted

Management explicitly guided for ‘temporary risk to billings and revenue’ as it operationalizes the sales optimization plan. This is management being unusually transparent about short-term headwinds. The stock is being punished for that transparency. The restructuring is designed to create a leaner, more productive sales organization. The disruption window is 2-4 quarters, not permanent.

4. Construction Cloud is Not Yet in the Model

Most equity models still treat ADSK as primarily a Design software company with ACC as an emerging add-on. ACC is now $796M and growing at 22%+. In three years, it will represent 20%+ of revenue at premium SaaS multiples. The embedded optionality is enormous.

5. AI Will Create a New Monetization Layer

Autodesk is uniquely positioned for AI because it has 40 years of proprietary design data, a specific technical domain where generic LLMs fail (3D geometry, manufacturing tolerances, construction codes), and a Flex consumption platform ready to monetize incremental AI usage. This optionality is valued at approximately zero in the current stock price.

Risks Worth Watching

Construction Market Cyclicality

Autodesk’s AECO business is partially exposed to construction starts. A deeper-than-expected construction downturn driven by sustained high interest rates could slow net new logos and put pressure on renewal rates. Management’s NRR metric (currently 100-110%) is the canary in the coal mine here.

Go-to-Market Disruption

The second phase of the sales optimization plan — channel partner integration and digital self-service buildout — has not yet started. If it is botched, it could create multi-quarter revenue disruption. The departure of the CRO during the restructuring adds integration risk.

AI Disruption Risk

The longer-term question: could AI design tools from NVIDIA, Google, or specialized startups displace Autodesk workflows over a 5-10 year horizon? We think the answer is no — for the same reason it has not happened yet. Design and engineering is not a generalist AI task; it requires domain expertise embedded in proprietary models that Autodesk is actively building. But this is a risk that needs monitoring.

Macro Sensitivity

ADSK is global — 44% of revenue from international markets. Currency headwinds (the new transaction model somewhat amplifies FX exposure), geopolitical disruption, and a global construction slowdown could all pressure results. The company has not paid a dividend and does buybacks at a modest pace relative to FCF, limiting the yield cushion in a down market.

The Bottom Line: A Tier-1 Software Franchise at a Tier-3 Multiple

Autodesk is the standard-bearer for design and construction software globally. Its competitive position — built on 40 years of installed base, educational programs, and increasingly on cloud data and AI — is as defensible as any in enterprise software. The business has successfully transformed from a legacy perpetual license model to a 97% recurring revenue, cloud-first subscription business. Free cash flow is growing at 20%+ annually. Margins are expanding toward best-in-class. The regulatory overhang is fully resolved.

And the stock trades at 19x next year’s non-GAAP earnings.

We have reviewed four annual reports, every quarterly earnings call from the past two years, the investor day presentations, and all available sell-side research. The consensus is Buy with a $340-360 target. Our view is that the consensus is right for the wrong reasons — most models are not fully capturing the margin expansion potential, the ACC growth trajectory, or the AI monetization optionality.

The stock is sitting near the bottom of its 52-week range ($215-$329), below its 200-day moving average, and at its lowest non-GAAP P/E multiple in years. When accounting scandals close, restructuring disruptions end, and FCF normalization becomes visible in real-time numbers, the stock re-rates. That process has already begun — ADSK rose 6% on its February 2026 earnings release. We are in the early innings.

ADSK is a BUY. Price target $340. Time horizon 12-18 months.

Disclosure & Disclaimer

This article is for informational and educational purposes only and does not constitute investment advice. The author may hold positions in securities discussed. All financial data sourced from SEC filings, company press releases, and publicly available research. Past performance is not indicative of future results. Always conduct your own due diligence and consult with a qualified financial advisor before making investment decisions. Price data as of early April 2026.

Wall Street has spent the last three months terrified of enterprise software. The iShares software ETF is down over 21% year-to-date. Traders started calling it the “SaaSpocalypse” — a term coined at Jefferies to describe a selloff that has erased an estimated $2 trillion in software market capitalization. The narrative is everywhere: AI agents will kill SaaS, per-seat licensing is dead, these companies are structurally impaired.

I think the narrative is half-right — and that getting the other half right is how you make serious money over the next 18 months.

Here is the thesis in one paragraph: The market is correctly identifying that these companies are too fat. It is incorrectly concluding that they are therefore worthless. The real trade is that AI doesn’t just threaten these companies — it also hands them the precise tool they need to fix their single biggest problem. And when they fix it, the stocks will do what over-cost, over-hated companies always do at generational lows: they rip.

The Philosophy: Buying Cost, Not Revenue

There are two ways to make a stock go up. You can grow the top line, or you can shrink the cost line. Wall Street spends 95% of its time focused on revenue growth. That’s why it consistently misses the other opportunity — the one that’s actually more reliable, more predictable, and more binary in its outcome.

When a company’s cost structure is wildly out of step with its revenue base, you are sitting on a coiled spring. The costs are observable today. The savings math is not complicated. The only variable is whether and when management pulls the trigger. And when they do, the savings flow directly and immediately to free cash flow — not through the revenue line with all its uncertainty and lag, but straight to the bottom line, the next quarter.

Think about what that means in practice. A company that cuts $1 billion in annual operating costs doesn’t need to sell a single additional dollar of product. Every dollar of savings is essentially a dollar of new earnings. At a 20x earnings multiple — modest for a growing software company — a billion dollars in cost cuts is worth $20 billion in market cap. That math is why stocks can surge 20–30% in a single day when a major restructuring is announced. The market is not reacting to news. It is repricing a certainty that was always there, hidden inside the cost structure, waiting to be unlocked.

This is not a new trade. It is one of the oldest trades in value investing. Buy companies that are priced as if their current bloated cost structure is permanent. Wait for the inevitable moment when reality asserts itself and management acts. Collect the re-rating.

What is new is the scale of the opportunity. Because AI has simultaneously done two things: it has created the narrative that justifies the cuts to a skeptical board and public, and it has created the actual tools to execute those cuts faster and more deeply than any prior restructuring wave. The combination of cover and capability is rare. When it shows up, you pay attention.

The Setup: Multi-Year Lows Across the Board

To understand the opportunity, you need to understand how extreme the current dislocation is.

We are not talking about stocks that are down 10 or 15 percent from recent peaks. We are talking about category-dominant businesses that have been cut in half — or worse — from highs reached just 12 to 24 months ago. Many high-profile software names have hit multi-year lows in 2026.Software price-to-sales ratios have compressed from 9x to 6x — levels not seen since the mid-2010s.

ServiceNow, one of the most admired enterprise software franchises ever built, was trading at over $211 a share in July 2025. Today it sits near $103. That is a 51% drawdown on a business that just beat earnings for the ninth consecutive quarter. Workday has fallen from an all-time high of $307to around $127 — a nearly 59% decline from peak. Intuit, which runs TurboTax and QuickBooks — products that are category monopolies with decades of pricing power — has fallen from a 52-week high of $813 to under $430, a collapse of nearly 50%.

These are not speculative companies. These are businesses with billions in recurring revenue, deeply embedded customer relationships, and operating models that have survived multiple economic cycles. They are being repriced as if they face existential destruction.

J.P. Morgan’s research describes the market’s logic as a fundamental contradiction: investors are holding two mutually exclusive fears simultaneously. On one hand, they fear AI will disrupt and replace enterprise software. On the other, they fear hyperscalers are spending too much on AI infrastructure and won’t generate expected returns. Both fears cannot be simultaneously true. If AI companies are going to obliterate software companies, those AI infrastructure stocks should be more valuable — not less. Bank of America called the resulting selloff “internally inconsistent,” likening it to the DeepSeek panic of January 2025 — which proved a buying opportunity within weeks.

The market is selling indiscriminately. That is exactly the environment where disciplined, specific stock selection generates the highest returns.

The Screen: Revenue Per Employee

The standard way to identify cost opportunity in these companies is to look at SG&A as a percentage of revenue. That works. But there is a sharper lens that gets closer to the actual mechanism of what is about to happen.

I call it the Revenue Per Employee Trap.

In knowledge-work software companies, a large portion of headcount is concentrated in go-to-market, customer success, implementation, and support — functions that generate revenue indirectly by helping customers adopt and expand their use of the product. For two decades, this was simply the cost of doing business. You needed humans to sell, onboard, train, and retain enterprise clients. There was no alternative.

AI has created an alternative. Not a partial one — a near-total one for the most repetitive, process-driven elements of these roles. Customer success managers doing quarterly business reviews from templates. Support agents triaging tier-one tickets. Implementation specialists running standardized onboarding workflows. Sales development reps qualifying inbound leads. These are not creative, judgment-intensive jobs. They are pattern-matching jobs. And pattern-matching is precisely what large language models do at a fraction of the cost.

The proof of concept is already in the public record. Salesforce cut around 4,000 customer support roles after deploying AI systems that handle a growing share of service requests, with CEO Marc Benioff stating that AI agents now handle approximately half of all customer interactions. This is not a projection. It is a reported fact from a Q4 2025 earnings disclosure. The technology works. The savings are real. The playbook is written.

The screen below applies three filters simultaneously: revenue per employee (a proxy for labor intensity of the business model), stock drawdown from the 52-week high (a proxy for board pressure to act), and an AI role-replacement index (an estimate of what share of the workforce sits in automatable functions). The companies that score highest on all three are the ones where the coiled spring is tightest.

How to read the screen: Revenue/Employee tells you how efficiently headcount translates to revenue. A high number in a services-heavy business model (like Intuit’s human tax experts) signals bloat in revenue-generating roles — not lean operations. Drawdown tells you how much board pressure exists to act. AI Replacement Index estimates the share of current roles that AI agents can realistically absorb within 18 months. The composite score weights all three equally and is meant to flag urgency, not investment quality.

The Four Highest-Conviction Names

ServiceNow (NOW)

ServiceNow is the name that gets the least attention in this context, which is exactly why it is the most interesting.

The company has built what is arguably the best enterprise workflow platform in the world. Its Now Platform is deeply embedded in how large organizations manage IT operations, HR processes, and customer service workflows. Switching costs are extraordinarily high. Revenue retention is exceptional. The business grew revenue 21% year-over-year in fiscal 2025 to $13.3 billion, and just beat earnings for the ninth consecutive quarter. By every fundamental measure, this is a business in excellent health.

And yet the stock has been cut in half. The 52-week high was $211.48. It now trades near $103. ServiceNow is on track for its largest quarterly decline on record. The reason is fear — fear that AI will erode the value of the workflows ServiceNow manages, and by extension, the software that manages them.

Here is the irony: ServiceNow’s own platform is precisely the technology that enterprises use to automate the workflows that AI is now targeting. The company is not a victim of disruption. It is a potential beneficiary — provided it adapts its cost structure to the new reality.

The numbers make the opportunity clear. ServiceNow had 29,187 employees as of December 31, 2025. Sales and marketing expenses alone ran at $4.39 billion in fiscal 2025, with G&A adding another $1.12 billion — a combined SG&A load of over $5.5 billion against $13.3 billion in revenue, a ratio of 41%. The company was aggressively hiring even as the market was collapsing around it; sales and marketing headcount costs alone rose $534 million in fiscal 2025. That hiring binge, combined with a stock that has since been cut in half, is exactly the setup that forces a reset.

A 25% workforce reduction — roughly 7,300 positions — at an estimated fully-loaded cost of $200,000 per employee would generate approximately $1.46 billion in annual savings. Against $13.3 billion in revenue, that pushes operating margins from the mid-20s toward 36% or better. At a 15x revenue multiple — well below historical norms for this business — that margin expansion alone implies a stock price significantly above today’s levels. ServiceNow’s next earnings call is scheduled for April 22nd. Watch closely.

Workday (WDAY)

Workday’s situation is perhaps the most straightforward on this list, and the most urgent.

The all-time high Workday stock price was $307.21, reached in February 2024. Today the stock trades around $127. That is a nearly 59% decline in just over two years, on a business that has continued to grow revenue at a double-digit pace throughout that entire period — $9.55 billion in fiscal 2026, up 13% year-over-year. The market has essentially decided that every dollar of Workday’s future earnings is worth roughly half what it thought two years ago. That is an extraordinary repricing of a mission-critical business.

The co-founder has returned as CEO. This is almost always a watershed moment. When Howard Schultz returned to Starbucks, when Steve Jobs returned to Apple, when Bob Iger returned to Disney, each brought a founder’s clarity and willingness to make bold bets that a hired executive simply cannot. A returning founder is not afraid of the company’s history, and is not politically beholden to the people who built the current structure.

Workday has already begun cutting. Combined with the February 2025 round that eliminated 1,750 positions, the company has now cut more than 2,100 jobs in just over 12 months. But 2,100 jobs out of 21,000 employees is incremental. The setup calls for something far more decisive — a single, clean, comprehensive restructuring that resets the cost base in one move and allows the company to grow into a leaner operating model. That is what the market is waiting to reward.

A 30% workforce reduction at a fully-loaded cost of $210,000 per head would generate roughly $1.3 billion in annual savings. At a 25x free cash flow multiple — conservative for a growing, mission-critical enterprise software business — that incremental cash flow represents over $30 billion in market cap, against a current market cap of approximately $35 billion. The stock could nearly double on cost action alone, before giving any credit to continued revenue growth.

Autodesk (ADSK)

Autodesk is the underappreciated name on this list. It doesn’t generate the same headlines as Workday or ServiceNow, but the setup is arguably cleaner.

Autodesk’s fiscal 2026 revenue came in at $7.21 billion, up 17.5% year-over-year. The business is accelerating. The products — AutoCAD, Revit, Fusion 360 — are the professional standards for architects, engineers, and manufacturers worldwide. There is no viable substitute. The competitive moat is as deep as it gets in software.

And yet the company has now executed two separate rounds of major layoffs in the span of 12 months: a 9% reduction (approximately 1,350 employees) announced in February 2025, followed by another 7% reduction (approximately 1,000 roles) in January 2026, concentrated in customer-facing sales teams. Starting from a base of 15,300 employees entering fiscal 2025, Autodesk has now cut roughly 2,350 positions — and the stock is still down 28% from its 52-week high despite strong revenue growth.

That is the tell. When a company with accelerating revenue and dominant market position cuts its sales force not once but twice in a year, it is not doing so because business is weak. It is doing so because AI is making those headcount levels unnecessary, and management has figured out that the math of cutting is better than the math of holding. The question is whether 16% cumulative cuts are the end of the story, or the beginning. Given the SG&A load and the continued stock underperformance, the pressure to go further is acute.

Intuit (INTU)

Intuit is the most counterintuitive pick on this list, and counterintuitive is usually where the money is.

The surface read on Intuit is that it is a quality compounder. TurboTax and QuickBooks are dominant, sticky products, and the company grew total revenue 16% to $18.8 billion in fiscal 2025, with GAAP operating income up 36%. That read is correct. What the market is also correctly identifying — but perhaps incorrectly pricing — is that Intuit built an enormous human labor infrastructure as a competitive moat. When the software alone wasn’t enough to justify premium pricing, Intuit wrapped it in humans: human tax experts through TurboTax Live, human bookkeepers through QuickBooks Live, armies of customer success agents managing small business relationships.

AI has now eliminated the strategic necessity of that wrapping. When Intuit’s own product can provide real-time, accurate, personalized guidance inside the interface, the army of human experts sitting behind it becomes an expensive legacy rather than a moat. The stock reflects this fear, having fallen nearly 47% from its 52-week high of $813.70 to under $430. That is a brutal drawdown on a business that still throws off extraordinary cash flows.

The numbers are stark. Intuit’s annual SG&A expenses were $6.6 billion in fiscal 2025 — against $18.8 billion in revenue, a ratio of 35%. For a company selling software products this mature and this dominant, that SG&A ratio is a function of the human service layer, not the product itself. Compressing it toward 25–28%, in line with a more efficiently run software business, would represent over $1 billion in annual savings and a meaningful re-rating of a stock that already trades at multi-year lows. Intuit has 18,200 employees. At fully-loaded costs of $180,000 per head, a 20% reduction unlocks approximately $655 million annually — before any benefit from the higher-margin AI-native service model that replaces the human one.

The Risk You Need to Understand

This is not a risk-free trade. No trade is. There are three things that can go wrong, and you should think about all of them before sizing a position.

The first is that management doesn’t act, or acts too slowly. The pressure is intense, but board cultures vary, and some CEOs will choose incremental cuts over decisive restructuring. The companies that nibble — doing repeated 2–3% reductions over 18 months rather than one clean reset — will not get the stock market reaction they need. Repeated rounds of small cuts destroy morale without delivering the financial clarity that investors are looking for.

The second is that cuts go too deep into revenue-generating functions. Cutting 30% of your customer success team sounds like a cost win until your net revenue retention rate drops five points a year later. The real structural threat to these businesses is not that AI kills their products overnight — it is that AI is eating their customers’ budgets, with every dollar redirected to AI infrastructure being a dollar not spent on another software seat. Management teams have to thread the needle between cutting enough to satisfy investors and cutting so much that they damage the customer relationships that make the business worth owning.

The third is that sentiment gets worse before it gets better. Short interest in mid- and large-cap software has surged over the past three months, with cybersecurity and SaaS companies seeing the biggest spike in bearish bets. A crowded short can stay crowded for a long time. These stocks can go lower before they go higher.

But here is the thing about that last risk: the more short interest piles in, the more violent the reversal when the catalyst arrives. A restructuring announcement at a heavily shorted company doesn’t just go up. It explodes. The short squeeze amplifies the fundamental re-rating. That dynamic is not a coincidence — it is a feature of this trade.

The Bottom Line

The SaaSpocalypse has created one of the most unusual setups in enterprise software investing in a generation. Companies with dominant products, deep competitive moats, and accelerating revenue are trading at multi-year lows because the market has decided their cost structures are permanent liabilities rather than fixable problems.

They are not permanent. They are fixable. The tools to fix them have never been more available. The board pressure to fix them has never been more intense. And the market reward for fixing them has never been more clearly telegraphed.

Prices have gotten so low that even longtime SaaS bears are now seeing buying opportunities. JP Morgan strategists have stated that the market is pricing in worst-case AI disruption scenarios that are unlikely to materialize, while Morgan Stanley’s software research chief has described the selloff as a sentiment-driven dislocation not justified by the underlying fundamentals.

The companies on this screen — ServiceNow, Workday, Autodesk, and Intuit at the top of the list — are not broken businesses. They are businesses with fixable cost structures, beaten-down stock prices, and enormous leverage to a single, highly predictable event: the moment their management teams decide that the math of cutting is better than the math of waiting.

That moment is coming. Buy before it arrives.

Meta Platforms hit an all-time high of roughly $796 per share last August. Today it trades around $530, a drawdown of more than 32 percent. The Magnificent Seven narrative that carried it to those heights has fractured under the weight of three simultaneous pressures: a historic surge in capital expenditure, a string of legal defeats, and a broader macro environment rattled by tariffs and Iran tensions.

The most immediate catalyst was a brutal week in the courtroom. On March 25, a Los Angeles jury found Meta and Google liable for social media addiction, ordering Meta to pay 70 percent of a $6 million award. The following day, Meta’s stock was down more than 11 percent for the week, the worst drop in the Magnificent Seven cohort that week. The day before the LA verdict, a New Mexico jury ordered Meta to pay $375 million for failing to protect children from predators on its platforms. Two adverse verdicts in 48 hours will do that to a stock.

Compounding the legal pain, Meta announced in January that its 2026 capital expenditure guidance is $115 to $135 billion, nearly double the $72 billion spent in 2025. Investors who had grown accustomed to the margin expansion story of the “Year of Efficiency” suddenly found themselves staring at a company spending more on infrastructure in a single year than the entire 2022 market cap of many S&P 500 members.

“I think that it’s the right strategy to aggressively frontload building capacity so that way we’re prepared for the most optimistic cases.” — Mark Zuckerberg, Q3 2025 earnings call

Then there is the macro backdrop. Trump’s tariff regime has weighed on advertising budgets, raised recession fears, and triggered a tech-wide sell-off. Meta fell 31 percent during the 2025 tariff shock, versus a 19 percent decline for the broader market. With a beta of 1.28, Meta reliably amplifies both market rallies and drawdowns.

The Business: What Meta Actually Does

Strip away the headlines and Meta is one of the most structurally dominant advertising businesses ever assembled. The company operates in two segments: Family of Apps (FoA) and Reality Labs.

The Family of Apps — Facebook, Instagram, WhatsApp, Messenger, Threads, and Reels — generated essentially all of the company’s $201 billion in revenue in 2025. Advertising is 97 percent of that total. Family daily active people reached 3.48 billion in Q2 2025, up 6 percent year over year, and hit 3.58 billion by Q4. For reference, that is roughly 44 percent of the entire human population engaging with a Meta product every single day.

The advertising engine runs on two levers: impressions and price per impression. In Q4 2025, ad impressions grew while average price per ad rose 6 percent. The AI-driven Advantage+ ad suite has been meaningfully improving return on ad spend for advertisers, which in turn allows Meta to charge more. This is the compound loop that drives the business: more users leads to more data, which leads to better AI targeting, which leads to higher advertiser ROI, which leads to higher prices and more impressions.

WhatsApp monetization remains the most underappreciated element of the Meta story. With over 3 billion monthly active users, WhatsApp is barely monetized today. Business messaging and click-to-WhatsApp ads are growing rapidly and could represent a multi-tens-of-billions annual revenue stream on their own.

Reality Labs lost $19.2 billion in 2025 on revenue of roughly $2.3 billion, pushing cumulative losses past $75 billion since 2020. The segment has been a source of investor frustration for years. However, the narrative is beginning to evolve: Ray-Ban Meta smart glasses sales tripled year-over-year in the first half of 2025. The $799 Ray-Ban Display, which launched in September 2025 and sold out within 48 hours, suggests the wearables thesis is real. Zuckerberg said on the Q4 call that 2026 will likely represent peak Reality Labs losses as the company gradually reduces deficits going forward.

True Cash Flows: Cutting Through the Capex Fog

The headline free cash flow number of $43.6 billion for 2025 looks impressive. But it needs context. The $115-135 billion capex guidance for 2026 represents a near-doubling of the $72 billion spent in 2025. If operating cash flow grows at roughly 20 percent and hits around $140 billion in 2026, and capex lands at the midpoint of $125 billion, free cash flow would compress to approximately $15 billion — a drop of more than 65 percent year over year.

This is the number that should terrify investors in the near term and excite them in the medium term. The compression is real and front-loaded. But capex is not consumed the moment it is spent; it gets depreciated over 10-15 years for data centers and 3-5 years for servers. The cash is spent now; the expense hits the income statement slowly. The gap between cash capex and depreciation is what Wall Street calls “capex overhang,” and Meta is sitting on a massive one.

• FY2025: Operating Cash Flow $115.8B minus Capex $72.2B = Free Cash Flow $43.6B

• FY2026 Est.: Operating Cash Flow ~$138B minus Capex ~$125B (midpoint) = Free Cash Flow ~$13B

• By FY2028, if capex normalizes to $80-90B and revenues compound to $280B+, FCF could rebound to $80-100B+. The question is whether you can live with the trough.

There is also the joint venture structure to understand. Meta entered a $27 billion joint venture with Blue Owl Capital for its Hyperion data center in Louisiana, with Blue Owl covering 80 percent of the cost. This off-balance-sheet structure reduces the direct cash burden somewhat. Similarly, the $100 billion agreement with AMD and the CoreWeave cloud deal spread some infrastructure costs to operating expenses rather than pure capex. The true economic capex picture is complex, but the cash demand is undeniably large.

Looking at maintenance versus growth capex matters here. Meta’s baseline infrastructure to support the current ad business probably requires $20-30 billion annually. Everything above that is discretionary AI investment — a bet on future revenue streams that do not yet exist at scale. Investors are effectively paying for a $100 billion per year AI lottery ticket, with a mature advertising business as the collateral.

The AI Bet: What Are They Actually Building

Meta’s AI strategy has undergone a significant pivot. The company spent years as the champion of open-source AI through its Llama model family, building enormous goodwill with the developer community. That era appears to be ending. After a lukewarm reception to Llama 4 and growing concern that DeepSeek had incorporated elements of Llama’s architecture, Meta is now developing a closed-source successor codenamed Avocado.

The strategic reorganization is equally notable. In July 2025, Zuckerberg launched Meta Superintelligence Labs (MSL), consolidating all AI research under a single banner with the explicit goal of building superintelligence. MSL is led by Alexandr Wang, who joined as part of a $15 billion acquisition of Scale AI. Meta has reportedly offered signing bonuses of up to $1 billion to recruit top researchers. Yann LeCun, the longtime chief AI scientist and “godfather of AI,” departed in late 2025 amid the strategic shift away from open-source.

Two flagship models are in development. Mango is a multimodal image and video generation model. Avocado is a reasoning-focused LLM targeting strong coding performance. Both were originally scheduled for a first-half 2026 internal release, though Avocado encountered training challenges that pushed it back.

The AI infrastructure buildout is staggering. Meta committed to spending $600 billion on U.S. infrastructure through 2028. The Hyperion data center in Louisiana is projected to eventually scale to 5 gigawatts. The Prometheus data center in Ohio is expected to be online in 2026. Meta signed a $100 billion multi-year deal with AMD for GPU infrastructure, a $27 billion compute deal with Nebius Group, a $14.2 billion deal with CoreWeave, and is reportedly in talks with Oracle for another $20 billion agreement.

The core thesis is that Meta’s 3.58 billion daily active users represent the world’s most valuable distribution platform for AI products. Meta AI, the assistant embedded across Facebook, Instagram, WhatsApp, and Messenger, has a built-in audience that no standalone AI company can match. If Avocado performs at the frontier level and Meta AI becomes genuinely useful, the monetization runway expands from advertising alone into AI services, premium subscriptions, and enterprise tools.

Meta has something no other frontier AI lab possesses: 3.58 billion daily active users who are already in the habit of opening Meta apps. If Meta AI reaches GPT-4 quality and becomes the default assistant for billions of users in emerging markets, the monetization potential through ads, subscriptions, and business tools is orders of magnitude larger than the current ad business alone. WhatsApp Business alone, at $10 per business per month across 300 million businesses, would be a $36 billion annual revenue stream.

Lawsuit Risk: The Legal Iceberg

The legal exposure facing Meta in 2026 is broad, complex, and genuinely material. Three distinct litigation vectors deserve attention.

The FTC Antitrust Case

The longest-running legal threat reached a significant turning point in November 2025 when U.S. District Judge James Boasberg ruled that Meta does not hold monopoly power in personal social networking, given competition from TikTok and YouTube. The court’s analysis focused on current market conditions rather than the state of the market in 2012 and 2014 when Meta acquired Instagram and WhatsApp, respectively. However, the FTC filed an appeal in January 2026 and the case continues. Even if the FTC wins on appeal, legal experts note that proving current competitive harm is a steep climb. The government still needs to demonstrate that Meta’s conduct is harming competition in 2026, not just relitigate acquisitions from a decade ago. The structural risk of forced divestiture of Instagram or WhatsApp is the tail risk that matters — low probability but existential in impact.

Youth Addiction and Harm Litigation

This is the more immediate and potentially more damaging legal frontier. The recent California and New Mexico verdicts are bellwether cases in what could become Big Tech’s tobacco moment. As of March 2026, more than 235 plaintiffs are in federal multidistrict litigation against Meta, over 250 school districts have filed suits, and more than 100,000 individual mass arbitration claims have been submitted. The design-defect legal theory — targeting platform architecture rather than user-generated content — sidesteps Section 230 protections in a way that could unlock enormous liability.

The individual damage awards so far are immaterial ($375 million in New Mexico, a few million in California). But the precedent is the point. Analysts estimate total industry-wide legal exposure at $10 to $50 billion depending on how appellate courts treat the design-defect theory. Meta has already warned in its 2026 10-K that youth addiction lawsuits and mass arbitration “could significantly impact” financial results. The tobacco parallel is imperfect but instructive: the industry settled for $246 billion spread over 25 years. A similar settlement trajectory for Big Social would be painful but survivable for a company generating $140 billion in operating cash flow.

European and Global Regulation

The EU AI Act imposes transparency obligations on Meta’s agentic tools. The Digital Markets Act creates interoperability requirements for dominant platforms. GDPR fines continue to accumulate. Meta has introduced location-based fees for European advertisers to offset compliance costs. European regulatory friction is a chronic drag rather than an acute risk, but it limits Meta’s ability to extract maximum value from its European user base.

The FTC case: Low probability of forced divestiture, but years of legal distraction and uncertainty. Youth addiction litigation: Potentially $10-50B in aggregate industry settlements over many years — painful but manageable given Meta's cash generation. The real risk is not the dollars but regulatory-mandated platform redesigns that could impair engagement metrics and advertising effectiveness.

Growth Trajectory: Where Is This Business Going

Meta’s revenue grew 22 percent in 2025, powered by AI-enhanced advertising and continued user growth. Q4 growth accelerated to 24 percent. Management guided Q1 2026 revenue of $53.5-56.5 billion, implying approximately 28-33 percent year-over-year growth at the midpoint — an acceleration. Consensus analyst estimates project 2026 revenue around $235-240 billion, implying roughly 17-19 percent growth.

The digital advertising market is Meta’s primary growth driver. AI-powered ad targeting through tools like Advantage+ has increased advertiser ROI, allowing Meta to raise effective CPMs. In Q4 2025, average price per ad grew 6 percent while ad impressions also grew — volume and price moving together is the dream scenario for an ad business.

Three incremental revenue vectors are worth tracking. First, WhatsApp monetization remains nascent. Click-to-WhatsApp advertising and business messaging APIs are growing fast from a small base. Second, Threads has scaled to hundreds of millions of users with minimal monetization so far; early ad tests have been promising. Third, AI Services — a premium Meta AI tier or enterprise Llama licensing — represents blue-sky upside that is not yet in any model.

Consensus growth forecasts call for roughly 15-20 percent revenue CAGR over the next three years. That implies revenues approaching $300-350 billion by 2028. If margins recover as capex normalizes, the earnings power of this business in 2028-2030 could be extraordinary.

DCF Valuation: What Is It Actually Worth

Building a DCF for Meta requires making honest assumptions about the capex trough, the revenue growth rate, and the terminal margin. Here is a grounded framework using three scenarios.

Key Assumptions for All Scenarios

Current price: approximately $530. Shares outstanding: approximately 2.53 billion. Discount rate: 10 percent (reflecting Meta’s market beta and risk profile). Terminal growth rate: 3.5 percent. The FY2026 FCF trough is painful but temporary; the question is the recovery trajectory.

Applying a DCF with a 10 percent discount rate and these terminal values, discounted back to today:

1. $350: -34% from current. Capex stays elevated through 2029. AI monetization disappoints. Legal costs escalate. Advertising decelerates to high single digits.

2. $730 +38% from current. Capex peaks in 2026. Ad revenue compounds at 16-18%. AI shows early monetization. FCF normalizes to $80-90B by 2028.

3. $1,100+107% from current. Avocado succeeds. WhatsApp monetizes. AI Services launches. Capex normalizes early. FCF hits $120B+ by 2028. Rerating to 30x.

The base case is $730, implying roughly 38 percent upside from current levels. The wide dispersion between bear and bull reflects genuine uncertainty about whether the capex spending generates returns. The consensus analyst target is around $840-875, suggesting Wall Street is pricing something between the base and bull. The average 1-year price target across Wall Street is $873, with a high of $1,200 and a low of $683.

Sum of the Parts: Breaking the Business Apart

Valuing Meta as a monolith misses the texture of what is inside. A sum-of-the-parts analysis reveals both the value buried in the core advertising business and the optionality embedded in Reality Labs and AI infrastructure.

The SOTP analysis is instructive but comes with important caveats. The 30x multiple on the core ad business feels defensible for a franchise growing 17-22 percent with 82 percent gross margins and near-monopoly positioning in social advertising. The optionality values on WhatsApp, AI, and glasses are speculative by nature. The key takeaway is that even if you assign zero value to everything except the core advertising business and apply a conservative 25x multiple on FY2026 earnings, you get to roughly $700-750 per share — comfortably above today's price.

The Fears: What Could Actually Break This

The bears have a real case. It centers on several scenarios that should not be dismissed.

The first and most plausible bear case is that the $125 billion in annual capex does not translate into meaningful new revenue streams for four to five years, during which time free cash flow remains suppressed, earnings multiples compress further, and investor patience runs out. This is the “Amazon in 2014” scenario — and it is worth noting that Amazon’s stock went nowhere for years before the AWS flywheel became undeniable.

The second fear is AI model irrelevance. Llama 4 was widely considered a disappointment relative to models from OpenAI, Google, and Anthropic. If Avocado also fails to achieve frontier performance, Meta risks spending $600 billion building the world’s most expensive second-tier AI infrastructure. Developer adoption of Meta’s models reportedly declined significantly following the open-source pivot uncertainty. Without a competitive model, the AI assistant and AI services strategies become difficult to execute.

The third fear is advertising cyclicality. Digital advertising is not recession-proof. If tariffs tip the U.S. economy into a meaningful slowdown in 2026, advertiser budgets will be cut. Meta’s entire operating leverage thesis depends on advertising continuing to compound at double-digit rates. A recession that cuts advertising growth to low single digits while capex stays at $125 billion would be genuinely painful.

The fourth fear is legal and regulatory in nature. The youth addiction litigation, if it reaches a Big Tobacco-scale settlement, could impose both financial costs and platform redesign mandates. If courts require Meta to fundamentally alter the engagement-optimization algorithms that make Instagram and Facebook so effective at retaining users, the advertising business model is impaired at its foundation.

The fifth fear is that TikTok returns to the U.S. market fully. Meta benefited enormously from TikTok’s near-banishment. Reels user growth, advertiser migration, and engagement metrics all reflected the reduced competitive pressure. If TikTok returns at full strength, some of that accreted value reverses.

Current Valuation and the Setup

Meta currently trades at roughly 25x trailing earnings and 19.7x forward earnings, with a PEG ratio of 0.88 — below 1, which conventionally suggests undervaluation relative to growth. The current P/E of approximately 22x is 22 percent below Meta’s 10-year historical average of 28x. The stock has a 52-week high of $794 and a 52-week low of $478. The average analyst price target is around $844, implying roughly 60 percent upside from the current price.

The forward multiple compression tells an interesting story. At $530 and with consensus 2026 EPS estimates around $29, the stock is trading at roughly 18x forward earnings. For a company growing revenue at 17-20 percent with 82 percent gross margins, a 18-19x forward multiple is genuinely inexpensive relative to history. The compression reflects the FCF trough from elevated capex, not a deterioration in the underlying business quality.

GuruFocus rates Meta’s GF Value at $754, calling the stock modestly undervalued with a GF Score of 99 out of 100. The EV/EBITDA of approximately 13.6x is also below what you would expect for a business with this growth profile and margin structure.

The stock repurchase program adds a floor element. In 2025, Meta spent $26.3 billion on buybacks and paid $5.3 billion in dividends. Share count declined 1.53 percent over the past year. Even with the FCF trough in 2026, Meta has $81.6 billion in cash and securities to deploy. Buybacks at current levels represent a meaningful yield on top of the fundamental growth.

The Setup: How to Think About Owning This

Meta in 2026 is a story of temporary cash flow pain in service of potentially permanent competitive advantage. The question every investor must answer is whether they believe: (1) the advertising business keeps growing at 15-20 percent; (2) the AI infrastructure eventually generates a return; and (3) the legal risks are manageable and not structurally impairing to the business model.

If you believe all three, the stock at roughly 19x forward earnings is arguably cheap. If you believe any of the three fails, the downside to the $350 bear case is material from here.

The setup has some characteristics that are genuinely interesting for patient capital. The capex cycle is well understood and widely feared, which means its peak is increasingly priced in. The legal verdicts this week caused the stock to sell off sharply, but the actual dollar amounts ($375 million in New Mexico, a few million in California) are rounding errors on a $43 billion free cash flow business. The market’s reaction appears to be pricing in the precedent risk, not just the specific damages. If appeals modify those precedents, there is sentiment upside.

The position sizing question matters. Meta’s beta of 1.28 means it amplifies market moves in both directions. In a macro environment with active Iran tensions, tariff uncertainty, and a Fed that may cut rates unevenly, that amplification cuts both ways. Averaging into a position over the next few quarters rather than committing a full allocation today would allow an investor to benefit from continued weakness while not missing a potential sharp recovery if the April 29 earnings call is clean.

The deepest insight about Meta right now is that the market is pricing in the worst of the capex cycle, the legal uncertainty, and the macro headwinds simultaneously. Rarely are three separate concerns maximally priced at the same time. That is when asymmetric setups appear.

The Bottom Line

Meta Platforms is a business generating $201 billion in annual revenue, growing at 22 percent, with 82 percent gross margins and 3.58 billion daily active users. Its core advertising engine is being strengthened by AI. Its next platform — AI glasses — is showing genuine commercial traction. Its management team has demonstrated, most recently in 2023’s “Year of Efficiency,” a willingness to make hard decisions and execute.

The risks are real. The capex is genuinely enormous. The legal exposure is broad and uncertain. The AI model competition is fierce. The macro environment is hostile to high-multiple growth stocks.

But at 19x forward earnings with a PEG below 1, the stock is pricing in a lot of bad news. Our base case DCF yields $730, roughly 38 percent upside. A full sum-of-the-parts analysis puts fair value closer to $968. Wall Street consensus is $844. The setup is for patient investors willing to live through a year of compressed free cash flow in exchange for optionality on one of the most consequential AI buildouts in corporate history.

This is not a trade. It is a 2-3 year thesis that requires conviction in the advertising business’s durability and at least some belief in the AI return on investment. Those who hold it with that time horizon have historically been rewarded. Those who need quarterly FCF validation will find the next few quarters uncomfortable.

Most investors think “value” and “tech” are opposites. We disagree. Some of the greatest wealth-compounding opportunities of the past two decades have been cheap tech stocks trading at a discount to intrinsic value. The trick is knowing where to look — and knowing how to look.

This post is a complete walkthrough of our due diligence process for software and technology stocks. We’ll cover how to screen for candidates, how to read the financials like a forensic accountant, how to build a simple valuation model, and how to know when to pull the trigger — and when to walk away.

STEP 1: SCREENING FOR CANDIDATES

The first job is to find ideas. The market is large and noisy. You don’t have time to analyze every company, so you need a disciplined screen that surfaces businesses worth studying.

We start with quantitative filters. These aren’t valuation screens — we’re not trying to find stocks that look cheap on a spreadsheet. We’re trying to find businesses that have real earning power but are temporarily in the penalty box.

Primary screens we run:

• Price-to-Free-Cash-Flow below 15x. For mature software companies generating real cash, this is the starting gate. Many great businesses have briefly touched this level after earnings misses or macro sell-offs.

• Enterprise Value-to-Revenue below 3x. For earlier-stage software with high growth, we tolerate losses but want to see revenue multiples that imply the market isn’t pricing in a recovery. Below 3x on a company growing 15%+ is interesting.

• 52-week lows with no change in fundamentals. This is our favorite hunting ground. When a stock hits a 52-week low, we ask: has the business deteriorated, or has the narrative deteriorated? These are very different things.

• Insider buying. When executives are buying their own stock in the open market with their own money, we pay close attention. This is the most reliable signal in the game.

• Share buybacks at depressed prices. A company aggressively repurchasing its own stock at 52-week lows is making an implicit statement about intrinsic value. We take that seriously.

[Practitioner Note] Tools like Stratosphere, Tikr, or Koyfin make these screens easy to run. We also manually read the “Frequently Cited Stocks” section of 13F filings from investors we respect — if a great investor is buying a beaten-up tech stock, we want to understand why before the next 13F cycle reveals it to everyone else.

STEP 2: UNDERSTANDING THE BUSINESS MODEL FIRST

Before we open a single spreadsheet, we spend time understanding what the company actually does and how it makes money. Sounds obvious. You’d be surprised how many investors skip this step and go straight to the P/E ratio.

In software and tech, business model quality varies enormously. A company selling perpetual licenses is fundamentally different from one selling annual SaaS subscriptions, which is different from a marketplace, which is different from a hardware company with a services attach rate. Each has a different risk profile, a different valuation framework, and different things you need to understand to form a view.

We ask a set of basic questions upfront:

1. What is the revenue model — recurring subscription, transaction-based, advertising, or one-time license?

2. Who are the customers, how large are contracts, and how long do they typically stay?

3. What is the switching cost for a customer who wants to leave? Is the product mission-critical or discretionary?

4. What is the competitive landscape — are there entrenched incumbents, well-funded newcomers, or is this company the dominant player in a niche?

5. Does the business have network effects, proprietary data advantages, or high customer concentration risk?

This part of the process is qualitative and messy. Read every earnings call transcript from the past three years. Read competitor filings. Read industry analyst reports. Talk to people who use the product if you can. The goal is to build conviction — or to kill the idea early — before you’ve invested hours in a financial model.

“We only buy what we understand. If we can’t explain the moat in two sentences, we don’t own it.”

STEP 3: DISSECTING THE INCOME STATEMENT

Now we open the 10-K. The income statement tells us about profitability, but in tech it can be deeply misleading if you read it naively. Here’s how we work through it.

Revenue quality and growth

We want recurring revenue, ideally ARR (Annual Recurring Revenue) or MRR-based. One-time revenue that gets counted alongside recurring revenue is a yellow flag. We disaggregate the revenue into its components and track each one separately across quarters. Is the growth coming from new customers or from expansion within existing accounts? Both are good, but for different reasons.

For SaaS companies, we pay close attention to the breakdown between subscription revenue and professional services. Professional services revenue is lower quality — it’s labor-intensive, lower margin, and doesn’t recur the way software does. A company inflating its headline revenue number with services attach is not as healthy as the top-line growth suggests.

Gross margins — the most important number in tech

Gross margin is the structural DNA of a tech business. Software gross margins above 70% indicate high-quality unit economics. Below 60%, we start asking hard questions. Below 50%, you’re essentially looking at a services business wearing a software valuation, and that’s a problem.

Gross margin benchmarks:

- World-class SaaS: 75–85%

- Good SaaS: 65–75%

- Mixed model: 50–65% (dig into the mix)

- Red flag: below 50% (structural concern)

Operating expenses: separating real costs from accounting fiction

GAAP operating expenses for software companies are badly distorted by stock-based compensation (SBC). This is one of the most important adjustments to make. SBC is a real economic cost — dilution is real — but it’s also non-cash and often used by management to obscure true profitability. We do two things: we adjust for SBC to understand cash economics, and we separately track SBC as a percentage of revenue to understand the dilution rate.

SBC above 15% of revenue for a mature company is a problem. It means management is compensating employees at the shareholders’ expense rather than earning returns for capital providers.

Adjusted Operating Income formula:

GAAP Operating Income

+ Stock-Based Compensation

+ Amortization of Acquired Intangibles

+ Restructuring Charges (if one-time)

= Cash Operating Income (our base metric)

We also watch R&D spend very carefully. In tech, R&D is the engine. A company that’s cutting R&D to hit near-term earnings targets is potentially starving its future. The question is always: is R&D being spent efficiently? We track R&D as a percentage of revenue over time and compare it to revenue growth. If R&D spending is rising but growth is slowing, the productivity of that spend is declining — that’s a sign of a business losing its innovative edge.

STEP 4: THE CASH FLOW STATEMENT — WHERE TRUTH LIVES

If the income statement is the face a company shows the world, the cash flow statement is what’s actually happening underneath. This is where deep value investors spend a disproportionate amount of their time.

Free cash flow is the only number that matters long-term

Free cash flow is operating cash flow minus capital expenditures. For most software companies, capex is low — servers, laptops, office equipment — which is precisely why software can be such a remarkable business: capital is not the binding constraint. But we verify this. We’re looking for companies where free cash flow is high relative to net income — ideally higher, which indicates quality earnings.

FCF Yield formula:

FCF Yield = Free Cash Flow / Market Capitalization

Target entry: FCF yield > 5–6% for mature software

Exceptional value: FCF yield > 8% with stable or growing FCF

Working capital dynamics and deferred revenue

For subscription software companies, deferred revenue is a gift. When customers pay annually upfront, the company collects cash before recognizing revenue — this is a source of float, similar conceptually to an insurance company’s float. Watch deferred revenue growth. If it’s growing alongside ARR, the billing model is healthy. If deferred revenue is shrinking while ARR is growing, customers are moving to monthly billing — that’s a cash flow headwind worth understanding.

The Rule of 40 — and why we use a stricter version

The Rule of 40 holds that a healthy software company’s revenue growth rate plus free cash flow margin should sum to at least 40. It’s a rough heuristic that tries to balance growth and profitability. We use it as a sanity check, not a buy signal. A company at 60% on the Rule of 40 that’s trading at 4x EV/Revenue is different from one at 60% trading at 15x — and deep value investors only care about the former.

Rule of 40 benchmarks:

- Above 60: Exceptional efficiency

- 40 (minimum threshold): Healthy benchmark

- 40–60 (deep value sweet spot): Priced below quality

- Below 40: Needs a catalyst — turnaround candidate

STEP 5: THE BALANCE SHEET — FORTRESS OR LIABILITY?

Deep value investing requires a margin of safety, and a strong balance sheet is a core source of that margin. We want companies that can weather a downturn, a competitive attack, or an economic recession without being forced into dilutive financing.

For software companies, the balance sheet analysis focuses on a few key areas. Net cash position — cash and equivalents minus all debt — tells us immediately how much cushion exists. A company with $500M in cash and no debt trading at $1B market cap has a very different risk profile than the same market cap with $400M in debt. The enterprise value — market cap plus net debt — is what you’re actually paying for the business.

We look at debt maturity schedules. A company with $300M in debt that matures in five years is fine. The same company with $300M maturing in 18 months during a period of elevated rates is in a very different position. Many value traps in tech have been created by investors who saw a cheap EV/EBITDA multiple without noticing the refinancing cliff sitting two years out.

Red flag checklist — balance sheet:

— Debt-to-EBITDA above 4x for a non-growing company

— Covenant-heavy credit agreements near the threshold

— Goodwill representing more than 50% of total assets (sign of acquisitive growth that may have been overpriced)

— Rapidly declining cash balance with no path to cash generation

— Any mention of “going concern” language in the auditor’s report — walk away immediately

STEP 6: SAAS-SPECIFIC METRICS — THE OPERATING DASHBOARD

For software companies, the GAAP financials are necessary but not sufficient. We also track a set of operating metrics that reveal the health of the underlying subscription engine. These are reported in investor relations sections, earnings supplements, and — for the best-run companies — the 10-K itself.

Net Revenue Retention (NRR)

NRR is the single most important operating metric in SaaS. It measures how much revenue you retain from existing customers after accounting for churn, contraction, and expansion. An NRR of 110% means that even if you acquired zero new customers this year, revenue would grow 10% from your existing base alone — through upsells, seat expansion, and price increases.

NRR benchmarks:

- Above 120%: Elite — best-in-class retention

- 110–120%: Strong — healthy expansion

- 100–110%: Acceptable — flat, needs new logos

- Below 100%: Problem — net churn, scrutinize immediately

Customer Acquisition Cost (CAC) and Payback Period

CAC payback period tells us how long it takes to recoup the cost of acquiring a customer through the gross profit generated by that customer. Under 18 months is very good. Over 36 months is a warning sign that the go-to-market machine is inefficient. We calculate this ourselves from the sales and marketing expense line divided by new ARR added in the period, adjusted for gross margin.

CAC Payback Period formula:

CAC Payback = (S&M Expense / New ARR Added) / Gross Margin %

Example: $10M S&M ÷ $5M new ARR = 2.0x CAC ratio

At 75% gross margin: 2.0 ÷ 0.75 = 2.67 years payback

→ Borderline acceptable; want to see this trending down over time

Magic Number

The magic number measures the efficiency of sales and marketing investment in generating new recurring revenue. We calculate it as net new ARR (current quarter ARR minus prior quarter ARR) times four, divided by prior quarter S&M expense. A magic number above 1.0 indicates you’re getting more than $1 of annualized new revenue for every dollar spent acquiring it. Below 0.75 suggests the go-to-market is struggling and needs to be fixed.

STEP 7: BUILDING A SIMPLE VALUATION MODEL

We are deliberately simple in our modeling. Complex DCF models with 15 tabs and precise projections out to 2040 give a false sense of precision. A five-year business is hard enough to model with reasonable accuracy. The goal of the model is not to predict the future — it is to understand what the market is implying about the future, and to judge whether that implied scenario is too pessimistic.

The reverse DCF approach

Rather than projecting forward, we work backward. Given the current stock price, what growth rate and terminal margin does the market need to assume to justify that price? If the implied scenario requires the company to grow 25% per year for 10 years to earn a fair return, that’s a high bar — any stumble and the stock falls. But if the implied scenario assumes zero growth forever and the company is still growing at 8%, the market is being irrationally pessimistic. That’s our opportunity.

Reverse DCF — implied growth formula:

EV = FCF × (1 - g/r) / (r - g) [simplified Gordon Growth Model]

Solve for g (growth) given:

· Current EV (market cap + net debt)

· Current FCF (trailing or forward estimate)

· r = discount rate (we use 10–12% for tech)

If implied g is less than 0% for a growing business → potential value opportunity

Scenario analysis — bull, base, bear

We model three scenarios: base case (the business performs roughly as it has historically), bull case (growth reaccelerates, margins expand to peer levels), and bear case (revenue growth halves, margins stay compressed). We assign rough probabilities and weight the outcomes. The key question is: if the bear case materializes, do we lose money in a permanent way, or is it a temporary drawdown in an otherwise sound business?

Deep value investing in tech requires a strong stomach for short-term pain. A stock can get cheaper after you buy it. What we’re trying to ensure is that the fundamental downside — the actual business value — is protected even in adverse scenarios. Price drawdown risk and fundamental risk are not the same thing.

STEP 8: MANAGEMENT QUALITY AND CAPITAL ALLOCATION

Great businesses run by bad management teams are investments that require the business to bail out the team. We prefer businesses where management thinks like owners — because they are owners.

We look at management’s track record on capital allocation. Have acquisitions created value or destroyed it? Has the company returned capital to shareholders intelligently — buying back stock when cheap rather than at peak prices? Are they empire-building with other people’s money, or are they disciplined stewards of the balance sheet?

Founder-led companies often score well here. Founders tend to think in longer time horizons, and they have direct personal stakes in permanent capital rather than just their comp plan. But founder-led doesn’t automatically mean good capital allocator — we’ve seen plenty of founders who love to acquire companies at multiples that never make economic sense.

Five things we look for in management:

1. Skin in the game. Does management own meaningful stock — not just options? Insider ownership above 5–10% for the combined leadership team aligns incentives with shareholders.

2. Acquisition track record. Pull the last three acquisitions. What were the purchase multiples? What was promised vs. delivered? Goodwill impairments are admissions of prior overpayment.

3. Shareholder communication quality. Do they give guidance they actually hit? Are earnings calls candid about problems? Management that’s never wrong is either lucky or lying. We want honest partners.

4. Compensation structure. Is executive pay aligned with shareholder returns, or is it largely fixed and structured to pay out regardless of outcomes? Proxy statements reveal a lot.

5. Capital allocation framework. Has the company articulated a clear hierarchy for capital allocation: organic growth first, then acquisitions, then buybacks? Improvised capital allocation is a red flag.

STEP 9: IDENTIFYING THE CATALYST

One of the oldest traps in value investing is buying cheap things that stay cheap. In deep value, we’re looking for businesses trading below intrinsic value, but we also want some reason to believe that the gap between price and value will close within a reasonable time horizon — typically two to four years.

Catalysts in tech can take many forms. A change in management — especially bringing in an operationally excellent CEO to replace a founder who was brilliant at building but poor at scaling. A business that has been through a significant restructuring and is now leaner, with a cost structure aligned to a lower-growth reality. A competitive scare that turned out to be overblown. A macro headwind — rising rates, enterprise budget freezes, FX headwinds — that is cyclical rather than secular.

Sometimes the catalyst is simply time: the market remembers the bad quarter but forgets the durable business underneath. Holding through the forgetting period requires conviction, which is why all the prior steps matter. You can’t hold through a 40% drawdown unless you have genuine, deep conviction in the underlying business value.

[Important distinction] We are not buying broken businesses hoping for a turnaround. A company with declining NRR, losing customers to better-funded competitors, with management that has missed guidance seven quarters in a row, is not a value opportunity — it’s a value trap. The question is always whether the business has enduring competitive advantages that have been temporarily obscured by noise, or whether the business itself is impaired.

STEP 10: SIZING AND PORTFOLIO CONSTRUCTION

Finding a great idea is only half the job. Position sizing is where the expected value gets converted into actual portfolio outcomes. We size positions based on conviction, valuation upside, and the severity of the downside scenario.

A high-conviction position where the bear case still implies a reasonable outcome might get 5–8% of the portfolio. A more speculative situation with a wider range of outcomes might get 2–3%. We never size so large that a single position failure becomes an existential problem for the portfolio. Permanent capital preservation is the first job.

We also think about correlation. Multiple beaten-up SaaS companies are likely to be correlated — they’ll all get hit in the same risk-off event, and they’ll all recover together. That’s fine if you’ve sized the aggregate exposure intelligently. We typically run 15–25 positions, concentrated enough to matter, diversified enough to survive.

PUTTING IT ALL TOGETHER: WHAT A REAL PROCESS LOOKS LIKE

Here’s a simplified example of how we’d run through a candidate. Imagine a vertical SaaS company that serves the construction industry. It came up on our screen because it hit a 52-week low after a disappointing quarter where guidance was cut due to a slowdown in commercial construction permitting — a macro headwind, not a company-specific one.

Step one: we understand the business. It sells project management and financial software to general contractors and subcontractors. Switching costs are high — contractors manage their entire job cost accounting on this platform. Once a contractor is live on the system, migration is painful and disruptive. That’s a moat we can hold in our head in two sentences.

Step two: we look at the financials. Gross margins are 72% and have been stable. NRR is 108% — lower than a year ago when it was 114%, but still above 100%, meaning existing customers are still spending more, just growing more slowly. FCF margin is 14% even in this slow environment. The company is profitable.

Step three: we run the reverse DCF. At the current stock price, the market is implying roughly 4% annual FCF growth for the next decade. But the vertical the company serves is one where software penetration is still in early innings, and the company has been compounding at 18% over five years. Even if it decelerates significantly to 10–12%, the stock is materially undervalued.

Step four: we check the balance sheet. Net cash of $180M, no meaningful debt, so the business can weather the construction slowdown without stress. Management has been buying back stock — $40M in the last two quarters at prices below current levels. That’s a good sign.

We take a 4% initial position. We’ll add to it if the stock goes lower and our thesis remains intact. The thesis is simple: commercial construction will eventually recover, and when it does, this company’s NRR will expand back toward 115%. At that point, the market will re-rate the multiple upward at the same time earnings power recovers. That double expansion — multiple re-rating plus earnings recovery — is exactly the kind of outcome that makes deep value investing in tech rewarding.

FINAL THOUGHTS

Deep value investing in technology is not about buying cheap junk and hoping it goes up. It is about finding genuinely excellent businesses — businesses with high gross margins, strong retention, real moats, and honest management — that are temporarily mispriced because the market has confused a cyclical problem with a structural one.

The work is primarily analytical: understand the business, verify the cash flows, stress-test the balance sheet, and judge management honestly. But ultimately, it requires a psychological toughness that the analysis alone cannot provide. You will buy things that go lower before they go higher. You will hold through earnings seasons where the stock drops 15% and everyone tells you you’re wrong. The process described here is designed to give you the conviction to hold — because you’ve done the work.

When the market sells high-quality tech indiscriminately, that is not a threat to the disciplined value investor. It is a gift. Learn to recognize it as such.