Adobe's AI Leverage Is Real — It's Just Not Big Enough Yet

Adobe just handed the bull case its best quarter in years: record revenue, raised guidance, and a fast-growing pile of AI subscriptions. The stock dropped anyway — down roughly 13% over five days and about 29% for the year. That gap between a headline that says "the AI leverage is happening, don't miss it" and a tape that keeps selling every beat-and-raise is the whole story in one chart. The leverage is real. It just isn't big enough yet to carry a $27.5 billion recurring-revenue base — and the market has stopped pretending otherwise.

What the bull case actually has

Start with what Adobe reported on September 10 for its fiscal third quarter. Revenue came in at $6.76 billion, up 13% year over year, with non-GAAP earnings per share of $6.13, up 15%. Total annual recurring revenue reached $27.5 billion, and the company raised its full-year revenue target to about $26.6 billion and its EPS target to about $24.48. More than a billion monthly active users, over 100 million of them on free creative tiers that grew more than 70%. On the face of it, this is a company executing, not defending.

The marquee number is what the company calls "AI-first" ARR — subscriptions tied to newer AI products like Firefly, Acrobat's AI assistant, and GenStudio rather than to the legacy tools. That crossed $650 million in the quarter, up more than 150% year over year. It had only just crossed $500 million in the prior quarter, up threefold. On the surface, that trajectory is the "leverage" the bulls keep pointing to: a new revenue stream compounding faster than the legacy business.

Now hold the $650 million against the $27.5 billion it sits inside. AI-first ARR is roughly 2.4% of total annual recurring revenue. A revenue stream can triple and still be a rounding error against the base it is meant to carry — and that base is the actual problem.

Notice what the reported growth is buying. In the spring, Adobe closed its acquisition of Semrush, a marketing-analytics firm contributing around $480 million of the ARR it counts. Strip that out and the underlying organic growth in Adobe's own business is closer to single digits — around 8% to 9% on what was guided toward 10% to 12% on a reported basis. So some of the "leverage" headline is a capital allocation decision, not a product one.

Then there's the mechanism by which the AI money is actually collected. Adobe does not sell Firefly as a standalone product so much as it rations generative output through "credits" bundled into plans, sold as paid credit packs, or captured by nudging users into higher-priced Creative Cloud tiers. That is a repricing and tier-shifting model more than it is a net-new customer model, which is why the critical unproven link is whether the giant free user funnel — the 100 million-plus freemium users — converts into paid AI seats rather than consuming free credits forever. The company set a million ARR records without yet demonstrating that conversion.

This is the two-ledger problem in miniature. The execution ledger — revenue, EPS, reported ARR — looks fine, because it does. The underlying ledger — organic growth, free-to-paid conversion, and the next quarter's guidance — is where the doubt sits. And each quarter the doubt wins on the margin: this time Q4 guidance of $6.80 to $6.85 billion in revenue came in with a midpoint just below what Wall Street had already discounted. A beat-and-raise stops feeling like a beat when every expectation has been pulled forward to meet it.

Cheap is not the same as a bargain

None of this is the AI-competitor apocalypse scenario the bears warned about for two years. Adobe is not being displaced by Canva or an open model in the quarterly numbers; it is growing, in the low teens, against a mature base. The discount investor has made the correct, if uncomfortable, observation: the franchise is intact but no longer compounding, so it now trades at roughly 13 times trailing earnings and about 10.5 times EV/EBITDA — against Microsoft at roughly 27 times earnings. That multiple is not an invitation; it is the market pricing in an erosion-by-a-thousand-specs future where a maturing creative base, an organic-growth shortfall, and a leadership transition (Shantanu Narayen hands the CEO role to Anil Chakravarthy in December) each shave a little off the terminal value.

So the "don't miss it" framing is the part to distrust. Adobe is not a product whose AI story is unproven — the story is proven and small. What is unproven is whether a credit-based, tier-shifting slice of revenue can grow from 2.4% of ARR into something that changes what a $98 billion company is worth, at a pace that outruns a core business settling into single digits. That is an execution question with a price tag attached, and the market has decided, real AI curve or not, that it does not yet have to pay up for it.

The leverage Adobe talks about is happening. The economics of that leverage have not arrived yet. Until the free funnel converts and organic growth stops leaning on acquisitions, the honest reading is not "you'll regret missing it" — it's that the market's skepticism is a reasonable price to pay while the company proves its AI money is new money, not a relabeling of the old.