AppLovin Lands on a Tax-Loss List Beside McDonald’s

Evercore ISI added AppLovin to a 40-stock tax-loss screen on September 7, grouping the advertising platform with McDonald’s, IBM, and Tesla. Strategists led by Julian Emanuel said funds may “make room” for OpenAI and Anthropic listings by selling names that already hurt.

AppLovin stock closed at $312.01 on September 8, more than 50% lower year to date and 58% below its 52-week high of $745.61. The operating business did not break. The ticker did, after a $16 million revenue miss in August, and that is why it now sits on a harvest list beside burger chains and mainframe software.

Evercore’s Screen Mixes AppLovin With McDonald’s

The note is easy to misread as an AI-for-AI rotation. It is a mechanical filter run on the Russell 3000. Evercore wanted stocks with a market value above $5 billion, a year-to-date drop of at least 10%, a gain of less than 20% from their 2026 low, and negative earnings-revision trends over three months. Forty names cleared. AppLovin cleared because it already failed those tests, not because ChatGPT is about to replace in-app ads.

The shortlist also includes AST SpaceMobile and First Solar. American Express, Lowe’s, and HCA Healthcare appear on longer recaps of the same screen. That is a loser list with a calendar attached. Mutual funds often realize losses in October to offset gains. Emanuel’s desk expects that process to start earlier so cash is ready if a mega listing prices.

HOW APPLOVIN CLEARS THE FILTER

A fund that needs a 1% or 2% slice of a giant listing has to raise cash from something it already owns. AppLovin’s float is large, the name is liquid, and the tax lot is underwater. That combination matters more to a September trading desk than whether AXON and Claude share a research paper.

A $16 Million Miss Already Did the Selling

AppLovin reported second-quarter results for the June period on August 5. Revenue was $1,924 million, up 53% from $1,259 million a year earlier. That print sat inside the company’s own range of $1,915 million to $1,945 million. It still came in about $16 million shy of the $1.94 billion consensus, and the stock treated the gap as a thesis break.

Shares closed August 6 at $335.67, down 19.66% from $417.80. They later tagged a 52-week low of $297.50 on August 24. Net income rose 55% to $1,267 million. Adjusted EBITDA rose 58% to $1,614 million, an 84% margin. Diluted earnings were $3.76. Free cash flow was $863.3 million. None of those figures look like a company in retreat. The multiple still compressed as if AppLovin were a tired ad network, while the cash engine kept printing software margins.

Co-founder and chief executive Adam Foroughi put the shortfall on model timing. Advertiser demand, he said, was not the problem.

This quarter came down to timing. Our pace of meaningful model improvement was lighter than normal during the quarter, and the next step up in model performance landed just after quarter end.

Adam Foroughi, Co-Founder and CEO, Q2 2026 earnings call

He added that nothing in the quarter pointed to weaker demand or a tougher competitive field. Third-quarter guidance still cooled the tape: revenue of $2,055 million to $2,085 million, or 46% to 48% growth, and adjusted EBITDA of $1,710 million to $1,740 million at an 83% margin. After years of beating the number, a quarter that merely landed inside the company’s own range was enough to refile the stock as a loss.

THE RERATE THAT FILLED THE TAX LOT

1. January 2026: Evercore ISI’s equity desk initiates AppLovin at Outperform with an $835 target, when the stock sat at $668.63.
2. June 23, 2026: AppLovin opens its ads product to all advertisers, ending a 14-year closed system, with the June quarter still open.
3. August 5, 2026: Q2 lands $16 million short of Street sales, with Foroughi citing a lighter model cadence.
4. August 6, 2026: The stock falls 19.66% to $335.67.
5. August 24, 2026: Shares print a 52-week low of $297.50.
6. September 7, 2026: Evercore’s strategy desk drops AppLovin onto the 40-name tax-loss screen.

The same firm that once sold a path to $835 is now telling funds that this ticker is a convenient place to raise cash. The August session did the economic work. The September note put a seasonal date on the next round of selling.

The Ad Engine Is Still Throwing Off Cash

AXON is a reinforcement-learning auction engine, not a chatbot. It scores demand against mobile and, increasingly, commerce inventory. AppLovin also runs MAX, the in-app bidding layer, and Adjust for measurement. Foroughi told investors in May that the company had been a closed platform for 14 years and that a worldwide sign-up in June would change the trajectory. The opening happened inside the second quarter. Sequential revenue still rose only about 4%.

That is the awkward part of the self-serve story. Opening the pipe did not produce a visible second-quarter pop. First-quarter revenue had been $1,842 million, up 59%, with an 85% adjusted EBITDA margin. The June quarter slowed on growth and ticked down a point on margin, and guidance points to another step down in the rate, not a stall.

WHAT THE JUNE QUARTER STILL PAID OUT

• Revenue: $1,924 million, up 53% year on year, with first-half sales of $3,766 million, up 56%.
• Cash: $869.0 million from operations and $863.3 million of free cash flow, even with tax and interest timing that held conversion down.
• Buybacks: 1.1 million Class A shares retired or withheld for $551.3 million, about $501 a share, leaving 335 million shares outstanding.
• Headcount: 898 people against trailing-twelve-month net income of $4.41 billion and trailing sales of $6.83 billion.

Chief financial officer Matt Stumpf has kept the capital plan simple: fund the models, then buy back stock. Net leverage at quarter end was about 0.1 times trailing adjusted EBITDA. The SEC closed its data-collection inquiry with no recommended action on the same call. Those are not the footnotes of a business that needs an OpenAI listing to explain the next dollar of profit. They are the footnotes of a business whose stock is now cheaper than its August narrative.

Why OpenAI Is Not a Substitute for AppLovin

Anthropic, PBC, confidential draft registration with the SEC on June 1, after a $65 billion Series H that valued the company at $965 billion. OpenAI submitted a confidential S-1 on June 8 and said it had not decided on timing. “It may be a while because there are things we want to do that are likely easier as a private company,” the company wrote. Private-market marks still sit near $852 billion for OpenAI. Neither deal has priced. Evercore still talks as if the calendar could tighten before year end, and as if SpaceX’s June record might not survive that wave.

Those listings, if they come, will soak up tens of billions of dollars of new-issue demand. They will not take AppLovin’s auctions. OpenAI is selling model access, subscriptions, and a new ChatGPT ad layer. Anthropic is selling Claude into enterprises. AppLovin is selling the right to bid on a user in a game or on a storefront, scored thousands of times a second. A portfolio can hold all three. It cannot swap the cash flows.

The confusion is the AI sleeve. Once a stock is filed under that label, it becomes a funding source for the next AI event, even when the product is an ad server. Foroughi has been blunt about how perishable that stack is. “If technology goes faster than what you’ve developed, you gotta throw away what you have and rebuild it,” he said in August, describing a rebuild cycle measured in a couple of years. That is an operator talking about model decay. It is not a prospectus for a foundation-model IPO.

Store Pixels Keep Climbing After the Slide

The part of the story the September note does not capture is the commerce build. AppLovin spent 2025 and early 2026 opening a consumer-advertising track beside gaming. Foroughi has called that product about a year and a half old. In the second quarter, consumer advertiser spend finished 28% above fourth-quarter 2025 levels, in a seasonally slow period. MAX publisher earnings, he said, still grew at a double-digit sequential rate, and AppLovin’s share of publisher waterfalls held steady.

Independent store trackers who count AXON Pixel installs keep posting higher weekly totals. Those figures are not company-reported, and they cover only shops that carry the tag, but the direction is not a collapse.

AXON PIXEL COUNTS FROM STORE TRACKERS

• June 28: 10,225 live store installs, just after the public opening.
• August 9: 10,908 installs, a gain of 683 in six weeks, while the stock was still digesting the miss.
• September 6: 12,020 installs, a further 1,112 in four weeks, with the weekly pace picking up rather than fading.

A rising pixel count does not prove that e-commerce will match gaming in dollars. It does show that merchants are still wiring the tag after a 20% down day and a trip to $297.50. Gaming saturation remains the live operating risk. If that vertical slows faster than commerce ramps, growth can keep decelerating even as the pixel graph looks healthy. That is a product question. It is not an IPO-allocation question.

The Research Desk Still Says Outperform

Evercore’s own AppLovin analysts did not move to a sell. They cut the target to $510 from $630 and kept Outperform, with early checks that the third quarter had opened firmer after the late model lift. That $510 sits next to the $508.39 average target from 33 analysts compiled on the quote tape, about 63% above the September 8 close. Bank of America went Neutral at $400 after the print. Piper Sandler stayed Neutral and cut to $325. The Street is not of one mind. It is cheaper than it was in January, and it is still, on average, a buy list with a wide spread.

WHERE THE READS DIVERGE

• Evercore strategy: AppLovin is a liquid, underwater large cap that funds can sell to seed OpenAI and Anthropic paper and to harvest losses before October.
• Evercore research: AppLovin remains Outperform at $510, with the miss framed as model timing and the consumer track still scaling.
• The post-print cuts: Several firms lowered targets, and at least one major desk stepped to Neutral, arguing that a 30% long-term growth pitch is now harder to defend.

The 52-week range of $297.50 to $745.61 is the simplest picture of that split. At $312.01 the stock trades at 17.26 times forward earnings on that tape, a multiple that would have looked pedestrian when AXON was being sold as an unstoppable AI compounder. It looks less strange next to an 84% margin and a 53% growth print. The AI label that carried the shares toward $745.61 is the same label that now makes them easy to ticket as “loss harvest, proceed to IPO.”

October is when mutual funds usually finish that harvest. Anthropic has been discussed as a possible October listing and has not set a date. OpenAI has said the timing may slip. AppLovin’s next full earnings print, on its normal cadence, lands after that tax window is already open. Funds that want the first look at a lab listing still have to wait for a price. The tax calendar does not.

Disclaimer: This article is news reporting and analysis of AppLovin, Evercore ISI’s screen, and related public filings. It is informational only and is not investment advice, a recommendation to buy or sell AppLovin or any other security, or a prediction of IPO pricing for OpenAI or Anthropic. Readers should consult a licensed financial adviser or tax professional who can consider their own holdings, cost basis, and risk limits before acting on tax-loss selling or new-issue allocations. Share prices, targets, tracker counts, and listing timetables reflect the sources cited and can change with the next session or the next SEC filing.