Adobe closes the Topaz Labs acquisition and gains technology for running heavy AI models locally

Adobe has completed its acquisition of Topaz Labs, a developer of neural models for image and video enhancement. The parties signed the agreement in late June; the closing took place on September 23, after regulatory procedures were cleared. Adobe did not disclose the value of the deal.

Topaz models are already available in Adobe Firefly and Photoshop — for upscaling and enhancing images and video while preserving the original's natural detail. The Topaz Labs brand is being kept, and its apps and models continue to exist as standalone products. The company's head, Eric Yang, has moved to Adobe's Digital Video and Audio division.

Heavy models move from the cloud to local hardware

Topaz's own technology is called Neurostream: it allows complex AI models to be processed on the user's device or in the Topaz cloud. According to the deal announcement, Neurostream makes it possible to run large models locally on consumer devices — previously such models were available only on high-performance systems or in the cloud.

Adobe pointed to this directly in its description of the deal: Topaz brings experience in optimizing large and complex AI models to run directly on the device, which should deliver faster response and make advanced AI more accessible and cheaper. For studios and freelancers the difference is practical: cloud processing requires uploading source files to an outside party, while local computation does not.

What already works and what Adobe is only promising

Among the company's most in-demand solutions are Astra for video upscaling and Wonder for image editing. Adobe separately highlights the task of modernizing content at industrial scale: studios with film archives, broadcasters with libraries of old recordings and marketing teams can bring legacy material up to modern resolutions.

Going forward, Adobe promises to bring several more scenarios into existing pipelines:

• restoring archival film and old video catalogs;
• rescuing footage shot in insufficient light;
• lifting the quality of material shot on a phone;
• preparing content for the big screen;
• creative upscaling of AI-generated video to professional level.

That last point probably explains the logic of the purchase: an upscaler becomes the link between a generative model and the editing timeline, and Adobe is closing that stretch inside its own tools.

What happens to Topaz users

Topaz Labs and Adobe customers can count on the same level of support, and Topaz products remain available as separate offerings through the company's website — a Creative Cloud subscription is not implied for them.

Topaz's technology was recognized with a 2025 Emmy for video technology. Its solutions are used by millions of customers, including 20 of the world's 50 largest companies; according to Eric Yang, the team has been working on image and video enhancement technologies for more than twenty years.

Why Adobe went for this deal

The purchase looks like a continuation of the Firefly pivot: the service launched as a showcase for Adobe's own models and then turned into a platform where third-party models are available alongside them. Upscaling and detail restoration are the link without which hybrid pipelines of shot and generated material do not come together, and Adobe apparently preferred to own that technology rather than rent it.

The backdrop for the decision is not an easy one. Amid pressure from generative services, investors doubt that the classic creative software suite will hold its former positions, and Adobe itself has in recent years leaned toward targeted technology acquisitions rather than large platform deals — after the Figma purchase fell through because of regulators.

Competition is pressing too. Blackmagic Design has built upscaling into DaVinci Resolve; ByteDance's CapCut hands its own video enhancer to a mass audience. The feature Topaz charged separately for is getting cheaper across the market, and independent upscaler developers are finding it ever harder to stay independent: Magnific went to Freepik, Topaz to Adobe.

The main point for artists remains open: will Adobe carry Neurostream's local computation logic over to its own Firefly models, or will credit-based billing stay the primary model for everything computed in the cloud.

Community reaction

Under Topaz Labs' official announcement of the deal's completion, opinions split. Some creators and regular users congratulate the team and see the union with Adobe as natural: those who have sat in Photoshop for years are glad the resolution-boosting tools will be right there, and hope for the next generation of models inside Creative Cloud. Alongside that runs steady skepticism and open irritation: in the critics' view, Adobe cannot keep up in the AI race on its own and is buying itself a seat at the table, then hollowing out products people had grown used to. The main practical argument is not about running heavy models locally — that barely comes up in the discussion — but about the fate of the standalone programs. Some are grateful for the promise to keep the apps and their own subscription, yet no longer expect proper updates for either branch. Others ask outright whether Topaz will be folded into the Creative Cloud subscription and whether the separate payment can be canceled. The overall tone is wary: there is gladness, but it is outweighed by distrust of Adobe as a buyer.

"Well, there goes another company Adobe will gut and ruin products people liked."

— @MinistryOvTrvth (X)

"Adobe on its own is not competitive in AI, so it just buys itself a seat at the table through acquisitions."

— @WyldeChyldeRec (X)

"I've used Photoshop constantly for eighteen years and worked in Topaz every day for over a year, so this merger makes so much sense to me it's as if I made the decision myself. Congratulations!"

— @ibexdream (X)

What lies behind the Topaz purchase: the race for local inference, savings on cloud computation, fear of losing the upscaling audience?