Adobe Beats — and Falls. The Number Behind Both Is ARR.

On June 11 Adobe reported the kind of quarter most software companies would frame and mount on a wall. Record revenue of $6.62 billion, up 13% year over year. Non-GAAP earnings of $5.96 a share, above what the Street expected. Full-year guidance raised, again.

The stock fell 6% that day to a fresh 52-week low.

That single contradiction is the entire puzzle when Adobe reports Thursday, Sept. 10. The standard retail framing — buy ahead of earnings — starts in the wrong place. Adobe beats-and-raises almost every quarter now, and the market has stopped paying for it. The reason lives in a number most investors never open.

For a subscription company, current-quarter revenue is a lagging indicator. It reflects contracts signed years ago, now marching through renewals. The leading indicator — the figure that tells you whether next year's revenue is still growing — is net-new annual recurring revenue: the fresh recurring subscriptions layered on top of the base each quarter.

This is where Adobe's record quarter quietly cracked. Revenue rose 13% and total ARR reached $27.1 billion. But net-new ARR, excluding the recently acquired Semrush, came in at $560 million, down 3% from a year earlier, and Adobe cut its own organic ARR target by roughly $480 million, about two percentage points of growth.

The income statement accelerated while the leading indicator decelerated, in the same quarter. The market sold the metric that matters and kept the one that doesn't, and the multiple collapsed with it. By mid-June Adobe traded at about 7x forward enterprise value-to-EBITDA, nearly half the 15x of a year earlier. A growth-software premium had become a value-software discount in one year of AI dread.

The dread has a mechanism. Generative AI made the thing Adobe sells — creative output — abundant. If anyone can conjure a marketing image with a prompt, the market reasons, the pricing power behind Photoshop's per-seat fees erodes. That destruction thesis is what Wall Street spent the past year pricing into the stock.

Run the abundance-scarcity test on that same fact and it cuts the other way. When content becomes abundant, the scarce thing is output a company can ship commercially without an IP lawsuit attached. That is Firefly's entire pitch: a generative model trained on licensed material and offered with enterprise indemnification — the thing a Midjourney prompt can't give you. Adobe says its AI-first ARR tripled year over year and passed $500 million.

But under that bullish headline sits a real concession. To chase users, Adobe deliberately slowed its own ARR growth this year and flipped to freemium — giving products away now, betting free users convert to paid later. Management made the trade explicit, likening it to the Acrobat Reader playbook that built a $27 billion subscription business. Wall Street heard "growth slowing by design," and its downgrades weren't about the quarter: Wolfe, Evercore, and Stifel all cut ratings over the freemium shift one called "thesis-changing".

The honest tension: the bull case rests on a free-to-paid conversion that hasn't been proven at scale, and the bridge is uncrewed. The CFO left for another company four days after the last report; the CEO who ran Adobe for 18 years stepped down in March.

So the number Thursday must resolve is not the EPS beat. Analysts largely expect a by-the-book quarter — roughly $6.07 a share on about $6.69 billion of revenue — and Citi goes further, calling for a beat-and-raise while flagging real fiscal-2027 ARR risk. What changes the story is net-new Digital Media ARR, plus two tells: whether Adobe holds its full-year ARR guidance, and whether Firefly's ARR climbs toward the roughly $400 million benchmark that would prove the freemium bet is converting. If free users don't convert at the projected rate, the thesis breaks. If Firefly keeps compounding and net-new ARR reverses its decline, the market that already repriced Adobe for ruin now owns a cheaper, asymmetric bet.

That is the asymmetry the selloff created: a value-style multiple on a business still growing revenue 13%, whose failure case is mostly already in the price. Whether it is value or a value trap is a question of conversion — and that is precisely the data Adobe owes you on Thursday.