Adobe's Acrobat AI Is Real — the Question Is Whether It Converts

Adobe just posted what looks like a clean quarter. On September 10 it reported record revenue of $6.76 billion, up 13% from a year ago, non-GAAP earnings of $6.13 a share (up 15%), a record $2.52 billion of operating cash flow, and it raised its full-year guidance. That is the shape of a company on the upswing. Yet the stock spent the year going the other way: Adobe is down about 29% year to date near $249, against a 52-week high above $370.

A beat-and-raise that keeps the shares sliding is not a paradox. It usually means the market is looking at a different number than revenue. For Adobe, that number is net new annual recurring revenue.

The number the market is watching

Adobe's book of recurring subscription revenue — its ARR — reached $27.5 billion, up about 11% from a year ago. Steady. But the new ARR added in the quarter, the figure that shows whether growth is accelerating or braking, fell roughly 36–37% year over year.

Management says the decline is a choice, not a failure. Adobe is deliberately giving away its AI features free to build a giant user funnel and convert it into paid subscriptions later. That choice is the whole story hiding inside the "Acrobat AI" headline.

The raw numbers behind the bet are enormous. Monthly active users across Adobe's business and consumer products — Acrobat and Express — passed 900 million, up more than 25% year over year. Acrobat AI Assistant's monthly active users doubled quarter over quarter. Creative freemium users passed 100 million, up over 70%. This is Adobe's one genuine moat in the AI fight: PDF is the closest thing to a universal document format, nearly every knowledge worker has Acrobat or Reader installed, and no rival can replicate that distribution surface overnight.

And monetization has begun. "AI-first" ARR — standalone AI products and credit packs — crossed $650 million, up more than 150% year over year, and business and consumer subscription revenue grew 15%. The catch is scale: $650 million is still only about 2.4% of Adobe's $27.5 billion ARR total. It is growing explosively from a base that remains, for now, small.

The contest was never about smarter AI

The framing question — can Acrobat AI outpace Microsoft and Google? — asks the wrong thing. Microsoft and Google are not beating Adobe on raw AI quality; they are beating it on distribution and price. Microsoft pushed a mostly-free "Copilot Chat" tier beneath its paid Copilot, and Google has folded Gemini into Google Workspace plans at little or no extra cost. For a business that already pays for Microsoft 365 or Google Workspace, an assistant that summarizes a document is a feature it already holds. That is the pressure Adobe actually faces — not being out-built, but being bundled around.

Adobe's counter is that it owns the document itself: the file format, the editing, the intelligence that lives in the PDF, the layer Microsoft and Google only bolt on top. Its new Productivity Agent carries that argument, converting a document into a slide deck, a report, or a podcast — workflows the suites do not match at the same depth.

The single number that separates bull from bear

This is what separates the two readings of the stock. The bull case: Adobe expanded AI usage explosively and is front-loading user acquisition into future conversion — the freemium pattern that has minted fortunes in consumer software. The bear case: Adobe is trading today's paid growth for monthly active users, as one analyst put it after last quarter's results, and the conversion is unproven while Microsoft and Google give the equivalent away.

If the funnel converts, net new ARR re-accelerates in the quarters ahead, and today's roughly 14-times forward earnings looks cheap for a low-double-digit-growing, 89%-gross-margin franchise. If it does not convert — if free users take the AI and stop there — the stock is cheap for a reason. There is already one measured example of that risk: last quarter Adobe traced roughly $70 million of a net-new-ARR shortfall to customers abandoning its stock-photo subscriptions for generative image tools. Cannibalization is not hypothetical here; it is booked.

Two further risks sharpen the judgment. Non-GAAP operating margin has drifted from about 47% to roughly 44% through 2026, as GPU costs and a lower-priced, AI-heavy user mix eat into profitability — the freemium funnel is not economically neutral. And Adobe is newly under different leadership, with Anil Chakravarthy slated to take the CEO seat from Shantanu Narayen; the market has marked the shares down partly to see that new team execute.

So forget "outpacing" Microsoft and Google. That was never a fair race, and it is the wrong lens — the question is which stage each player owns. Adobe owns the document and the funnel; the platforms own the default distribution. Demand is not the issue for Adobe: its AI usage numbers are real and visibly improving. The issue is the shape of the return — a cheap multiple that already prices in durable-growth doubt, set against an operating question that is genuinely still open. The honest verdict is that Acrobat AI is no longer a headline; it is a measurable bet whose answer arrives the moment net new ARR turns back up.