Adobe closed its purchase of Topaz Labs on September 23. The June announcement left out the price; the fiscal third-quarter 10-Q filled it in at about $340 million, mostly in cash and funded from cash on hand. ADBE ended the day near $240, a market value of roughly $95 billion. It’s down about 31% for the year, while the S&P 500 is up 14% and the software sector is flat. That puts the stock at about 9.8 times the midpoint of Adobe’s own fiscal 2026 guide for non-GAAP earnings per share ($24.45 to $24.50). At that price, $340 million would buy back about 1.4 million shares. Adobe repurchased 9.5 million in the third quarter alone and still has $24.55 billion of authorization unused.
That’s the bar. A stock at 9.8 times earnings yields a little over 10%, so $340 million of buybacks hands continuing holders about $35 million a year of earnings (roughly $42 million before tax at Adobe’s 18% non-GAAP rate). Topaz, family-founded and bootstrapped out of a guest bedroom in Texas, has never published revenue. The most specific outside estimate puts it around $48 million. Matching the buyback on that base would take an operating margin near 90%. Double the revenue and the margin needed is still about 44%, from a company that committed in 2025 to a much bigger headquarters and 130 new jobs by the end of 2027. On its own income statement, Topaz loses to the buyback.
So the case has to rest on a different world. Buying back stock at 10 times earnings is a bet that Creative Cloud keeps its economics through the generative shift; the market has spent 2026 declining to make that bet. Topaz pays in the other branch, where more footage comes out of a model than a camera and the model belongs to Google or OpenAI. Adobe has half-conceded that branch already, since the Firefly app sells access to both companies’ generators. Topaz sits downstream of all of them. Whoever renders the clip, somebody has to finish it.
Finishing persists because of arithmetic. A 4K frame holds four times the pixels of a 1080p frame. A diffusion transformer that attends across the whole frame pays for attention in proportion to the square of its token count, so that part of the bill rises toward sixteen-fold, and it’s paid again at every denoising step. The cheap path is to render below delivery resolution and let a smaller model finish. Topaz already sells Astra as that last step for clips out of Veo, Sora, Runway and Higgsfield; its pitch to AI creators is to start in any app and finish in Topaz. The rest of the demand owes nothing to generation: archived film, legacy game footage, marketing libraries, low-light and phone clips. Neurostream, the on-device inference engine Adobe highlighted in June, matters less to this math than where Topaz sits in the pipeline.
Adobe won’t win a spending race on frontier video models against Google and OpenAI. Its moat is position: the timeline where the edit happens, plus the enterprise contract that governs what ships. Enhancement is one of the last operations before export, inside that timeline, and one of the few AI features whose usage rises when a rival’s generator wins. Topaz brings quality at the professional end (detail recovery and frame-to-frame consistency, the property that makes upscaled video watchable) along with a 2025 Emmy for video technology. The weak flank is the low end, where generators keep adding an upscale button and a good-enough result often costs nothing. DaVinci Resolve has an upscaler built in, yet Topaz still found it worth shipping a Resolve plugin last December. Editors pay on top of a built-in tool only when the built-in result misses delivery spec. That gap is Topaz’s pricing power; each model generation narrows it.
Inside Adobe the deal has two ways to beat the buyback. The first is small and fairly certain. Topaz’s upscalers have been in Firefly since July 2025 and in Photoshop as a third-party option, so Adobe was paying a partner for output it now owns. The second is large and unproven. Topaz mostly sells subscriptions to individual photographers and editors, with an API on the side; Adobe sells Firefly Services to enterprises on consumption and has the sales force to push it. Remastering a studio’s back catalog or a brand’s video archive is priced by volume, which seat pricing never captures, and Topaz already counts 20 of the world’s 50 largest companies as users. Moving those models onto Adobe’s enterprise meter is where $42 million a year stops looking like a stretch. The closing statement gave no timetable for the Firefly Services and Creative Cloud integrations named in June.
Two problems sit between here and there. The first is timing. Topaz only became a subscription business last autumn: Topaz Studio launched on September 16, 2025, and perpetual licenses were gone by October 3. Its first annual cohort therefore comes up for renewal in the same weeks Adobe takes ownership, and some of those buyers picked Topaz as a way to stay outside Adobe’s subscription model. Adobe settled with the Justice Department in March over its cancellation practices, for $150 million split between civil penalties and free services. Keeping the Topaz name on the invoice is retention engineering as much as brand respect. It also keeps the Resolve plugin selling to editors who never open Premiere.
The second problem is people. David Wadhwani, president of the creativity and productivity business and the executive named on the June announcement, said in early September that he’s leaving. Dan Durn left the CFO job for Marvell in June. Shantanu Narayen hands the chief executive role to Anil Chakravarthy on December 1 and becomes executive chair. Topaz’s own chief executive, Eric Yang, joins the Digital Video and Audio team while the layer above it changes. One thing cuts the other way: Chakravarthy has run Digital Experience and worldwide field operations since 2020. He has spent six years selling to the kind of enterprises that would buy back-catalog remastering, which is the route this deal needs.
The stock has already priced a good share of this. ADBE closed near $285 on September 3, before the succession news, and has lost about 16% since, including a 4.5% drop on September 22. Narayen sold 125,000 shares on September 16 and 17 at $246 to $254. Consensus is Hold, with an average target near $280. Morgan Stanley’s Underweight and $240 target, set in July, now sits on top of the share price; the firm calls fiscal 2026 a reset year. Mizuho went into the print at Neutral and $260, while Barclays (Equal-Weight, $295) and RBC (Outperform, $315) raised targets in the first days of September, before the succession announcement. The base case is $225 to $270, nine to eleven times guided earnings, into the December print; that’s Chakravarthy’s first as chief executive and the one that sets fiscal 2027 ARR growth. The bull case is $300 to $330, about 12 to 13.5 times, if the fiscal 2027 guide holds ending ARR growth at 10% or better and puts AI-first ARR on a path past $1 billion. The bear case is cohort derating, with software selling off again on AI displacement while the guide shows ARR growth slipping under 9%. That sends ADBE back to its $190.12 52-week low at about eight times earnings, where the Street’s lowest target, $195, already sits.
The number that settles the Topaz question arrives with that print. Against $27.5 billion of total ARR, Topaz is noise; even the most specific outside estimate of its book moves Adobe’s 10.2% growth target by about a fifth of a point. The AI-first line is a different matter. It went from more than $500 million at the end of the second quarter to more than $650 million at the end of the third, roughly $150 million in three months, and most of the bull case leans on it. Topaz sells nothing but AI enhancement, so its subscriptions fit that category naturally. At half the outside estimate, Topaz’s book would still equal about a sixth of last quarter’s AI-first gain. If the fourth-quarter ARR bridge separates acquired from organic, investors can price the line. If it doesn’t, the metric Adobe most needs believed starts carrying bought revenue in the new chief executive’s first quarter.