AppLovin: Buying the Growth the Market Stopped Explaining

In December 2025 AppLovin closed at $733.60, the highest price in its history. Seven months later it changes hands near $390. Nothing in the reported numbers went backwards over that stretch. Revenue grew. Margins hit a record. The company bought back a billion dollars of its own stock in a single quarter. What changed was the story people were willing to believe about how the money is made.

That is the whole trade. You are not being asked to forecast a turnaround or model a recovery. You are being asked whether a business that converts roughly two thirds of every dollar of revenue into net income deserves a discount because four short sellers and one federal regulator have questions about its data practices.

What you are actually buying

AppLovin is no longer the mobile-gaming roll-up most people still picture. The entire apps and games portfolio was sold to Tripledot Studios in June 2025 for about $400 million in cash plus a stake of roughly 20% in the buyer. What is left is a single business: an advertising engine called Axon that decides which ad to show, to whom, at what bid, across a network of more than 140,000 mobile apps reaching over a billion people a day. Around it sit MAX, the auction that sells publisher inventory, plus Adjust for measurement and Wurl for connected TV.

The first quarter of 2026 set the tone. Revenue of $1.84 billion grew 59% year over year, the adjusted EBITDA margin reached a company-record 85%, free cash flow was $1.29 billion in a single quarter, and net income of $1.21 billion put the net margin at 65%. Guidance for the second quarter, given on 6 May, called for revenue of $1.915bn to $1.945bn (still 52% to 55% growth) with the EBITDA margin holding at 84% to 85%. Management expects free cash flow to settle around 75% of EBITDA for the full year once the tax and interest timing that flattered the first quarter washes out.

Those are software margins on an advertising business. Return on invested capital sits near 120%, because there is barely any capital in the ground to earn it on. That is the point.

Fewer installs, far more money

One line in the first quarter tells you more than the headline growth rate. Revenue rose 59% while the volume of app installs the platform delivered fell 18%. The company sold less and collected far more.

That only happens if advertisers are getting better outcomes per dollar and are willing to bid higher for the same impression. Axon is a bidding engine: it takes a return-on-ad-spend target from the advertiser and buys inventory against it. When the model gets better at predicting who converts, the advertiser hits the same target at a higher price, the auction clears higher, and AppLovin keeps the spread.

It also explains the margin. Serving 18% fewer installs costs less, not more, so the extra revenue drops almost intact to EBITDA. A pricing gain compounds differently from a volume gain: volume needs more inventory, more users, more supply, while price only needs a better model. This is why the margin keeps setting records rather than flattening out, and it is the part of the business the bears have never really disputed.

June 2026: the gate came off

For fourteen years AppLovin sold through people. An advertiser wanting access had to be onboarded, referred, walked through a campaign by a salesperson. The self-serve version of Axon Ads went live on 1 October 2025 but stayed behind a referral wall, which management used to control how fast new spend arrived.

At the end of June 2026 that wall came down worldwide, alongside a dedicated e-commerce product. Any advertiser can now sign up, generate creative with the platform's own tools, and run campaigns without speaking to anybody.

The reason this matters is arithmetic. Web advertising — meaning e-commerce, not apps — reached roughly a $1 billion annual run rate while still behind the referral gate. Management puts the non-gaming advertising market at five to ten times the size of gaming. The bottleneck was never demand; it was the roughly 57% of qualified leads that made it all the way to a live campaign under a human-led process. Self-serve is the fix for a funnel problem, and it costs almost nothing to run. There is a second effect worth naming: Axon targets on contextual signals rather than user-level identifiers, which sidesteps the iOS privacy rules that constrained identity-dependent platforms.

What the price implies

At $396 the market capitalisation is about $133 billion against 336 million shares. Trailing twelve-month earnings per share came in near $11.64 after the first quarter, so the stock trades on roughly 34 times what it has already earned. On my estimate for this calendar year — the first quarter delivered $3.56, the second is guided to something close to $3.75 — full-year earnings land somewhere near $15.50 a share. That puts the multiple around 25 times current-year earnings for a business growing above 50%.

Buybacks do real work here. The company retired 2.23 million shares for $1 billion in the first quarter alone and had about $2.3 billion of authorisation left. Share count is down roughly 2% over a year, and every dollar of that is funded from cash the business already generates rather than debt.

A discounted-cash-flow model starting from about $4.8 billion of free cash flow this year, letting growth fade from a first-year rate down to a terminal rate over five years, is not heroic in its defaults. Assume growth roughly halves within two years and lands in the low teens by 2031, apply a discount rate well above the risk-free rate, and the shares are worth about a fifth more than they cost. Push the first-year rate to the 50% the company is currently printing and the number moves a long way; cut it to 15% and the current price is roughly right. That asymmetry is the argument.

Three ways this ends

Cash-flow models are useful for testing sensitivity, not for setting targets on a business this early in a product transition. The scenarios below work off earnings for 2027 and a multiple the market has actually paid for this stock at various times.

Weighting those at 20 / 50 / 30 gives a probability-adjusted value near $520, about a third above the current price. Sell-side consensus sits higher still, around $654, with targets running from $406 at the cautious end to $860 at the aggressive one. I would not lean on that consensus: the same analysts had the stock as a top pick at $700.

Why the discount exists

None of the above is worth anything if the earnings are not real, or not repeatable. Four separate short-selling firms — Fuzzy Panda, Culper Research, Muddy Waters and CapitalWatch — have published against this company since early 2025. The claims range from improper collection of user identifiers and breaches of app-store terms, to installing apps without consent, to an allegation that two large shareholders run money through both sides of the auction. The chief executive has called the reports false and misleading, and no charges have followed. The specific risks:

• The SEC investigation is real and still open. Bloomberg reported it in October 2025; the stock fell 14% that day. In February 2026 the regulator described the probe as active and ongoing. It concerns data-collection practices and was reportedly triggered by a whistleblower. No wrongdoing has been established, but an open federal investigation into the core mechanism of the business is not a footnote.
• Platform dependency cuts deep. Morningstar's objection is the sharpest: if advertisers ever had to choose between AppLovin's tools and access to Meta or Google inventory, most would take the latter. AppLovin does not own the rails it runs on.
• The e-commerce ramp is already wobbling. On 13 July a Bank of America note flagged that AppLovin's e-commerce footprint expanded more slowly in June than expected. The stock fell 11.2% in a session. That is the exact metric this thesis rests on, and the first read on it was soft.
• The multiple is a second bet. Beta is about 2.45. A stock that fell 46% from its high in seven months without an earnings miss can do so again for reasons that have nothing to do with the business.

I would hold the position anyway, in size that assumes some of this lands. The bear case is mostly about how the numbers are produced, while the bull case is about whether they keep arriving — and for six straight quarters they have arrived, through two short reports and one federal investigation. Short interest has fallen to under 4% of shares outstanding.

How I am buying it

Second-quarter results land on 5 August, and a stock with a beta of 2.45 heading into a print on the exact metric that spooked the market three weeks ago is not something to take a full position in. So: a starter now, then adds only on declines of 20% or more — a starter of ~1.5% at $396 before Q2, a second 1.5% near $320 (below the 52-week low of $343), 2% near $255 (roughly 16× current-year earnings) making an average price of $301.

The cap is the part that does the work. A thesis with an open regulatory investigation in it does not get to be a large position, however good the margins look. Filling the whole ladder would mean the market has repriced the stock at half of today's level, and at that point the question is whether the invalidation list above has been triggered — not whether the shares look cheap.

Investment view: Buy, staged entry, position capped at 5% given the open SEC probe.

Prices are as of 22–23 July 2026 and will be stale quickly. Estimates for 2026 and 2027 earnings are the author's own, built from reported results and company guidance, not consensus figures. This article represents an investment thesis, not financial advice.

Sources

AppLovin, first-quarter 2026 shareholder update and earnings call, 6 May 2026.

AppLovin Q1 2026 earnings call transcript, The Motley Fool.

Coverage of the June 2026 self-serve and e-commerce launch (ppc.land, Common Thread Collective).

Valuation and share statistics (stockanalysis.com, Macrotrends).

SEC investigation reporting — Bloomberg via Morningstar and Forbes.

Short-seller allegations and status (Money Morning, Yahoo Finance).

July 2026 analyst activity and the 13 July e-commerce note (CNN Markets, CNBC).