APP — AppLovin Corporation

AppLovin has built the highest-margin advertising business in the world, with 84 percent adjusted EBITDA margins and $3.95 billion in free cash flow generated in fiscal 2025. An active SEC investigation into whether those margins rest on unauthorized device-tracking practices — the precise mechanism of competitive advantage — makes the thesis uninvestable until resolved. Interesting at a price that reflects the risk; not actionable today.

---

Apple's App Tracking Transparency framework, introduced in 2021, was supposed to permanently impair mobile advertising. It largely did what it promised. Within a year of the ATT rollout, most mobile ad networks reported CPM declines of 20 to 40 percent as the identifier-based audience matching that had underpinned mobile programmatic advertising collapsed. Meta disclosed losing an estimated $10 billion in 2022 revenue to ATT disruption. The broader mobile ad ecosystem retrenched, laid off engineers, and reconfigured its targeting infrastructure around a world where Apple would no longer permit the persistent device identifiers that had served as the industry's universal user key. The consensus view was that anyone dependent on identifier-based mobile targeting was structurally impaired.

What happened next was not anticipated. AppLovin, then a mid-sized mobile gaming and advertising-technology company, launched AXON 2.0 — its second-generation AI optimization model — in late 2023. Where its predecessor relied on traditional audience-matching signals, AXON 2.0 purported to operate on behavioral data from app events: in-game actions, in-app purchase sequences, level completions, session length patterns — signals that do not require a persistent user identifier and therefore survive ATT intact. The financial results that followed were extraordinary in a way that deserved scrutiny alongside celebration. Revenue grew 70 percent in fiscal 2025 to $5.48 billion. Adjusted EBITDA reached $4.51 billion, an 82 percent margin on annual revenue. Free cash flow was $3.95 billion. An advertising company with 84 percent EBITDA margins has no precedent at this scale.

Mobile advertising is a $262 billion market growing at roughly 23 percent annually, and the in-app segment — where AppLovin competes — is its most structurally attractive corner. Unlike display advertising, where a user has chosen to visit a site, in-app advertising reaches users inside applications they are actively using: games, utilities, social apps. The advertiser pays only for measurable outcomes — an install, a purchase, a registration — in a real-time bidding environment where an auction resolves in milliseconds every time an ad slot becomes available. The winner of that auction is determined by a predicted probability of conversion multiplied by the advertiser's bid. In this environment, competitive advantage compounds: the buyer with the best prediction model wins auctions, drives actual results, attracts more advertiser demand at higher bids, which generates more outcome data to improve the model further. Scale and model quality reinforce each other. Google and Meta dominate digital advertising broadly because they have spent decades building this flywheel. AppLovin's claim is that in the specific niche of in-app mobile performance marketing, it has built a sufficient data flywheel to compete with — and in certain measurable dimensions outperform — both platforms.

AppLovin was founded in 2012 and went public in 2021 as a hybrid business: a mobile gaming portfolio and an advertising technology platform. The gaming apps were not merely a revenue stream; they were a first-party data laboratory, providing AppLovin with training signal for its optimization models without depending on external data providers. By 2024, management had concluded that the gaming portfolio was a drag on financial quality rather than an amplifier of it, and executed the separation: in June 2025, the entire Apps segment was sold to Tripledot Studios for $800 million — $400 million in cash plus 20 percent equity in the acquirer — leaving AppLovin as a pure-play advertising technology business. The remaining company is effectively three things that function as one: AppDiscovery (the demand-side platform through which advertisers buy mobile installs and in-app actions), MAX (the mediation layer through which app publishers manage ad revenue across demand sources), and Adjust (the mobile attribution platform that measures outcomes). AXON, the AI model, sits at the intersection of all three, drawing on signals from advertisers, publishers, and measurement simultaneously to optimize in real time.

The financial case for a genuine competitive advantage is strong. The platform processes over 2 million real-time ad auctions per second, drawing on behavioral data from more than 1 billion devices. Revenue per installation across the app ecosystem surged 75 percent in 2025 despite only modest growth in total mobile app installations — meaning AXON's model is extracting materially more value from the same underlying traffic, which is the definition of performance improvement rather than market growth. Publishers using the MAX platform report monetization improvements sufficient that more than half of some publishers' user acquisition spend routes through AppLovin's ecosystem. The vertical integration of the AppDiscovery-MAX-Adjust stack creates a closed data loop that competing point solutions cannot replicate: every outcome is observed, every signal is returned to the model, and every iteration improves prediction marginally. Years of this compounding — first in gaming, now expanding into e-commerce — produce the margins reported.

The problem is that this account of AXON's competitive advantage has a parallel account that the SEC is currently investigating.

In October 2025, the SEC opened an investigation into AppLovin following a whistleblower complaint and short-seller reports from Fuzzy Panda Research and Culper Research. The investigation is examining whether AppLovin violated service agreements with platform partners — Meta, Amazon, Google — by harvesting device identifiers through its SDK without authorization, enabling targeting granularity that a model operating purely on first-party app-event signals could not achieve. The specific technical mechanism alleged is identifier bridging: using a combination of device characteristics — screen resolution, operating system version, device model, hardware configuration — to construct a persistent proxy identifier that circumvents ATT by stitching together fragments that no single prohibited identifier provides alone. AppLovin's SDK is embedded in hundreds of thousands of mobile applications, providing potential access to device-level data across essentially the entire installed app base on iOS and Android. If that access was used to collect data beyond what users consented to in those applications' privacy disclosures, or beyond what Apple's post-ATT rules permit, or beyond what platform service agreements allow, then AXON's performance edge is not a pure product of legitimate model training.

As of April 2026, the SEC has confirmed the investigation is "still active and ongoing," with enforcement officials from its cyber and emerging technologies unit assigned. The agency has declined to release correspondence, citing potential interference with enforcement proceedings. Fuzzy Panda and Culper Research's allegations — which include click spoofing, forced app installations, and unauthorized harvesting of Meta's proprietary ad data to train AXON — remain unretracted as of this writing. AppLovin discontinued its Array product, which installed applications on users' devices without explicit user consent, following public reporting; the company has not disputed that Array operated as described. The CapitalWatch allegation about money laundering connections to a major shareholder was formally retracted and apologized for in February 2026, but CapitalWatch explicitly stated that its position on AppLovin's underlying business practices was unchanged by that retraction.

An important analytical distinction: the question is not whether AppLovin has a data advantage — the financial results prove it does. The question is whether that advantage is reproducible under the regulatory constraints that an SEC enforcement outcome or platform partner response would impose. If AXON's performance is partly attributable to unauthorized identifier ingestion, then the compliant version of AXON is a different, meaningfully inferior product. The business that generated $3.95 billion in free cash flow in fiscal 2025 may not be the business that operates in fiscal 2027 after enforcement and remediation. The investor cannot currently distinguish between these two scenarios with any confidence.

The competitive landscape amplifies this concern. Meta's Advantage+ automation, powered by a social graph of 3.3 billion users and Apple's own SKAdNetwork conversion data for opted-in users, is the natural winner of a fully ATT-compliant mobile advertising world. Meta has already begun building direct SDK relationships with mobile publishers — relationships that AppLovin currently controls through MAX — and has the scale and engineering capacity to develop a competing mediation and attribution stack if the business opportunity justifies the investment. Google's Performance Max similarly leverages search intent signals that require no device identifier. Both companies possess the data assets and the distribution that would make a compliant mobile performance marketing product very difficult for AppLovin to defend against at equivalent scale.

The financial profile merits examination on its own terms before turning to valuation. Fiscal 2025 revenue of $5.48 billion grew 70 percent from fiscal 2024's $3.22 billion. Adjusted EBITDA of $4.51 billion at an 82 percent margin grew 87 percent year-over-year, implying fiscal 2024 adjusted EBITDA of approximately $2.41 billion on $3.22 billion in revenue — already a 75 percent margin at scale. GAAP net income was $3.33 billion, aligning closely with adjusted earnings and confirming that the margin profile is not manufactured through non-recurring charge normalization. Stock-based compensation runs approximately $400 million annually, roughly 7 percent of revenue, partially offset by buybacks. The GAAP-to-adjusted reconciliation is clean. The company carries approximately $3.5 billion in net debt against $3.95 billion in annual free cash flow, a leverage ratio that is not concerning at these cash flow levels.

Period	Revenue	Adj. EBITDA	Margin	Free Cash Flow
FY2024 (full year)	$3.22B	~$2.41B	~75%	~$2.07B
Q1 2025	$1,484M	$776M	52%*	—
Q2 2025	$1,260M	$1,020M	81%	—
Q3 2025	$1,405M	$1,158M	82%	—
Q4 2025	$1,658M	$1,400M	84%	$1,310M
Q1 2026 (guided)	$1,745–1,775M	$1,465–1,495M	84%	—

*Q1 2025 includes the gaming Apps segment (lower margin); gaming portfolio divested June 2025.

The Q4 2025 data is the clearest view of the business at full run-rate, without the gaming segment and with AXON operating at what management describes as its current ceiling of optimization. The $1.31 billion in free cash flow on $1.66 billion in revenue represents 79-cent cash conversion on the revenue dollar. Annualized, the platform as currently configured generates approximately $5.2 billion in free cash flow. These are the numbers a long investor is buying when purchasing AppLovin at the current price — but also the numbers that are materially impaired if the SEC investigation leads to business model changes.

Adam Foroughi co-founded AppLovin and remains CEO with approximately 10.88 percent of shares outstanding and effectively 93 percent of voting control through a super-voting share structure. The capital allocation record is genuinely strong: $2.58 billion in share repurchases during fiscal 2025, reducing diluted shares from 346 million to 341 million, executed at prices well above the current trading range. The gaming divestiture was rational — extracting $800 million from a declining-revenue business to focus capital on the highest-return asset. Management guided Q1 2026 revenue of $1.75 billion at 84 percent EBITDA margins, representing 5-7 percent sequential growth from Q4 2025, suggesting no sign of demand-side deterioration in the advertising business as of February 2026.

The insider selling record requires honest evaluation. Sixty-five selling transactions against zero open-market purchases in the disclosed period is a pattern, not noise. The transactions were executed through 10b5-1 plans, which reduces but does not eliminate informational content. The analytical answer — that Foroughi sold at prices substantially above the current level and retains a large stake — is partially satisfying but not fully. It is accurate that the remaining 10.88 percent ownership represents a compelling alignment with long-term shareholders. What it does not explain is why no one in management has purchased shares in the open market at prices 37 percent below the peak, when insiders at companies with genuine long-term conviction generally do exactly that.

The growth runway centers on e-commerce. AppLovin launched Axon Ads Manager, a self-serve e-commerce advertising platform, in October 2025. By Q4 2025, e-commerce advertisers were generating an annualized revenue run rate of approximately $1 billion from roughly 600 active customers. Self-serve weekly spending growth was reported at 50 percent week-over-week in early cohorts. The e-commerce advertising market is a $271 billion opportunity currently dominated by Meta and Google, and AXON's hypothesis is that the same machine learning model that predicts mobile game install intent can predict consumer purchase intent in retail categories.

The early evidence is genuinely mixed. AppLovin reports strong conversion rates from early customers, and management describes the e-commerce pipeline as the next phase of the company's growth. The advertiser quality in the early cohort, however, is heavily weighted toward direct-response businesses in supplements, weight loss, and adjacent consumer categories — advertisers whose performance marketing economics tolerate aggressive tactics that premium consumer brands do not. Ridge, an accessory brand, reduced AppLovin from 30 percent of its marketing budget in December to 5 percent by mid-February after observing conversion metrics and new customer visits deteriorate. Jones Road, a beauty brand, spent up to $50,000 per day on the platform and pulled back in November after concluding the efficiency trailed Meta's for their customer acquisition profile. These are individual data points from early testing rather than definitive structural evidence — but the brands that are pulling back are exactly the brands AppLovin needs to attract for the e-commerce expansion to justify the premium valuation.

E-commerce currently represents approximately 10 percent of AppLovin's revenue. The gaming advertising business that built the machine generates the remaining 90 percent. At this stage, the market is pricing AppLovin largely on what the e-commerce expansion could become — a business that extends AXON's performance model into the largest and fastest-growing segment of digital advertising. That is a reasonable thing to price. It is not a reasonable thing to pay 40 times free cash flow for.

AppLovin's shares traded at approximately $430-465 in mid-April 2026, down approximately 38 percent from the January 2025 peak of $745.61. At $465, the market capitalization is approximately $158 billion against fiscal 2025 free cash flow of $3.95 billion — a price-to-free-cash-flow multiple of approximately 40 times. Trailing GAAP P/E is approximately 43 times on $3.33 billion in net income. Forward estimates based on Q1 2026 guidance imply a forward P/E of approximately 24-27 times. Enterprise value is approximately $158 billion against $4.51 billion in adjusted EBITDA — roughly 35 times EV/EBITDA. These are not cheap multiples for a business with active regulatory uncertainty; they reflect the market continuing to accord AXON meaningful growth credit despite the SEC overhang.

On normalized pre-tax terms: fiscal 2025 GAAP pre-tax income was approximately $4.0 billion (net income $3.33 billion at roughly an 18-19 percent effective tax rate). With 341 million diluted shares outstanding, normalized pre-tax EPS is approximately $11.70. At $465, the stock trades at approximately 40 times normalized pre-tax earnings. The 15-times threshold at which a purchase requires no growth to be justified — paying $175 for the current earnings stream — is not in the discussion at current prices. Even the forward estimate on accelerating 2026 earnings still places the stock at approximately 25-30 times pre-tax normalized earnings. The investor is paying a substantial premium for growth that has not yet occurred, in a business whose competitive advantage is under federal investigation.

The bull case at these prices requires believing simultaneously that: the SEC investigation resolves without material business model restrictions; e-commerce advertising scales from $1 billion to $5-10 billion ARR over three to five years; AXON's gaming moat remains intact through whatever regulatory remediation is required; and Meta and Google do not respond to AppLovin's e-commerce ambitions by building competing mobile performance marketing capabilities that outperform a potentially constrained version of AXON. All four conditions may be true. None is certain enough to underwrite at 40 times free cash flow.

The intelligent bear on this stock argues that the SEC investigation is not peripheral noise but central to the thesis: that AXON's extraordinary margins exist precisely because the model does things that a fully compliant version could not do, and that the market is pricing AppLovin as if the investigation will be dismissed when the regulatory trend for unauthorized data collection has moved in a consistently more severe direction over the past decade. The answer that AppLovin's proponents offer — that AXON works on behavioral signals alone and the SEC investigation will ultimately confirm this — may be correct. But it requires betting on the outcome of an active federal investigation against an enforcement unit that has explicitly stated the probe is ongoing, on the basis of a company that has already acknowledged discontinuing one product (Array) that installed apps on users' devices without consent. That is not a high-confidence foundation for a 40-times-free-cash-flow valuation.

What would change this picture: SEC investigation closure with a finding that the business practices were compliant, or a consent decree that requires modest modifications without material revenue impact, combined with six to eight quarters of e-commerce revenue demonstrating that AXON generalizes to premium consumer brands at scale, at a price corrected to reflect earnings uncertainty — something in the range of 20 to 25 times normalized pre-tax earnings, which implies a purchase price in the $230-290 range on current earnings. At that price, with regulatory clarity, the e-commerce expansion becomes optionality the investor is not paying for. At the current price, without regulatory clarity, the optionality is already embedded and the risk is uncompensated.

The cash flows are real. The AXON model has demonstrably produced something extraordinary. The price, at 40 times free cash flow with an open federal investigation into the source of that performance, is not defensible for an investor who requires knowing what they own.