Adobe Is Buying One of the Last Good Things in Photo Editing

Adobe announced on June 25 that it has agreed to acquire Topaz Labs, the Dallas company whose denoising, sharpening, and upscaling tools quietly became part of how a huge number of photographers finish their work. Neither side put a number on the deal. Closing is targeted for the back half of 2026, assuming regulators sign off. Adobe says Topaz CEO Eric Yang will stay on, the standalone apps will keep running, and the underlying models will eventually flow into Firefly, Firefly Services, and Creative Cloud apps.

Read the press release and everything sounds reassuring. That is exactly the problem. The company doing the reassuring is the same one that spent the last decade teaching photographers that "continued support" is a phrase with an expiration date.

Let me be precise about what is and is not happening, because the easy version of this story is wrong and a wrong version is easy to dismiss. Adobe has not announced a price increase. It has not announced that Topaz tools are moving behind Creative Cloud. It has not said the standalone apps are dying. Anyone telling you otherwise is guessing, and the trade press urging everyone to calm down is, narrowly, correct: nothing changes the day the deal is announced. What Adobe has actually done is buy the leading independent AI enhancement company in the industry and fold it into a company that already controls the software most of us are required to use. The harm here is not a line item on next month's invoice. It is structural, and it is the kind of harm that is invisible until the moment you have no alternatives left.

What Topaz Actually Was

Topaz was bootstrapped, profitable, and family-founded, started in 2005 by Albert Yang and later led by his son Eric, and grown to more than a million customers and, by reported figures, roughly a $48 million business, all of it built as a bootstrapped company rather than through the raise-and-blitzscale playbook that defines most of its peers. That detail carries the whole argument. In an industry where the standard move is to take a fortune in funding, buy growth at a loss, and either get acquired or go public, Topaz built a real company by selling software people wanted at a price they would pay. It won a Technology and Engineering Emmy in December for its restoration work. Its tools are trusted by working filmmakers and studios, by enterprises, and by a long tail of professional photographers, video editors, and archival teams. None of that happened because Topaz locked anyone into an ecosystem. It happened because the software solved problems Adobe's own apps did not, and it did so as a focused specialist rather than a platform trying to own your entire workflow.

That independence was the asset. In a lot of photographers' workflows, Topaz was simply the specialist tool you reached for when Lightroom's built-in denoise or upscaling was not good enough, and you reached for it knowing it lived outside the subscription you were already paying. The company's newer work pushed in a direction Adobe should find genuinely uncomfortable. Its NeuroStream inference technology, which Topaz says cuts the memory a model needs by up to 95 percent, lets large enhancement models run locally on an ordinary consumer graphics card, with the company framing the goal in plain terms as no cloud needed and no additional usage costs. Adobe's own acquisition announcement says the quiet part out loud: it credits NeuroStream with running advanced models that were previously limited to high-end systems or cloud-only usage directly on consumer devices. That matters because it points away from the rented-cloud, metered-credit future Adobe is building toward, and toward one where your own machine does the work and nobody bills you per image. Topaz was the counterweight, the proof that a small company could out-execute the giant on the specific things photographers cared about, and on the specific architecture that threatens Adobe's model. Adobe is buying the counterweight. To be clear about what that does and does not mean: nothing in the announcement says Adobe intends to kill local processing. The point is narrower and worse. Adobe now controls a technology that could have pressured its own cloud-credit strategy, and it gets to decide how hard that pressure ever gets applied.

Even Topaz's AI pitch was usually closer to enhancement than replacement. Eric Yang has described the company's approach as the least offensive version of AI, and its strongest identity was never text-to-image generation from nothing but improving an image or piece of footage you already had: denoising, sharpening, restoring, upscaling, and more recently creatively enhancing AI-generated or digital source material. In a moment when "AI photo editing" increasingly means a machine inventing a picture from a prompt, that distinction was part of the brand. It is now owned by the company whose Firefly generative models are the headline act.

The Inconvenient Part of This Story

Unfortunately, Topaz had already walked away from the thing photographers loved most about it.

For years, Topaz sold perpetual licenses. You paid once, you owned the version, you bought upgrades when you wanted them. It was the philosophical opposite of Adobe's rental model, and it was a large part of why photographers trusted the company. In September 2025, Topaz effectively ended it. With the launch of Topaz Studio, the company stopped selling new perpetual licenses and moved future versions to subscription, with a final purchase window that closed in early October. Existing customers kept the versions they already owned; nobody's license was switched off. But the direction was unmistakable, and the photographers who left over it called the move exactly what it was at the time. By the time Adobe came along, Topaz had already become the thing its fans used it to escape.

So no, Adobe is not riding in to destroy a perpetual-license sanctuary. That sanctuary closed last fall. Pretending otherwise hands Adobe an easy rebuttal. The honest version is more nuanced. The independent alternative had already drifted toward the subscription model on its own, and now the last bit of independence, the part where it answered to its own customers instead of a large public incumbent, is gone too. You can read the sequence two ways. The charitable reading is that a small company made a normal business decision, so good for them. The cynical reading is that a company grooming itself for sale tends to look more like Adobe first. I cannot prove the second reading. I can tell you the timeline does not argue against it. 

Adobe Already Showed You the Meter

Before we get to history, here is the single fact that should end any argument about whether the metering fear is paranoid. Adobe was already charging credits for Topaz models inside its own products before it bought the company.

When Adobe integrated Topaz tools into Photoshop and Firefly as partner models, it did not make them free. It made them metered. Adobe's own partner-model pricing already lists Topaz Gigapixel, Sharpen, Denoise, Bloom, and Astra as credit-consuming features, with the photo models costing generative credits per use depending on size, Bloom costing more, and the Astra video model billed per second of footage. So the future I am worried about is not speculation about what Adobe might someday do. It is a description of what Adobe was already doing to Topaz's technology while Topaz was still independent. The acquisition does not introduce the meter. It removes the last independent party with an incentive to keep a direct standalone and local route alive outside Adobe's own credit economy. When the partner becomes owned infrastructure, the only remaining check on how aggressively those tools get metered is Adobe's own restraint, and we are about to spend several paragraphs on Adobe's record of restraint.

The broader direction points the same way. Just days before the Topaz news, Adobe introduced Firefly Graph, a node-based canvas for building reusable AI workflows, whose own documentation says its AI-model and external-API nodes consume generative credits. That does not prove Adobe bought Topaz to feed Graph, and I am not going to claim it did. What it shows is the shape of where Adobe is heading: workflow-based, model-based, and meter-based, with credits as the currency at nearly every node. Topaz's technology is now another asset inside that architecture.

'Adobe Will Support It' Is Not Comforting

The reason to distrust Adobe's reassurances is that Adobe has a record, and the record is not subtle.

In 2013, Adobe moved Creative Cloud to subscription only and took perpetual licenses off the table. Photographers were furious then and adapted because there was nowhere else to go, which is the entire lesson Adobe seems to have drawn from the episode: anger is survivable when the customer has no exit. The entry-level Photography Plan that hooked everyone at $9.99 a month has been pared back; Adobe raised the 20 GB plan from $9.99 to $14.99 in January 2025 and discontinued it for new subscribers the same day, and the current Photography plan is listed at $19.99 a month. Mid-year, Adobe rebranded its all-apps tier to Creative Cloud Pro at $69.99 a month and cut the monthly generative credit allotment for new single-app and entry Photography subscribers to just 25. What's happening is segmentation: standard AI gets bundled into the higher tiers, premium AI becomes credit-metered, and the lower-entry plans get a much smaller monthly allowance. Now apply that logic to a tool you currently run as many times as you like for a flat price, a tool Adobe, as we just established, already meters inside its own products.

Then there is the Federal Trade Commission. In June 2024, the government sued Adobe over allegedly hidden early-termination fees and a cancellation process the complaint described as deliberately difficult, naming two executives, including David Wadhwani, the same executive whose name appears on Adobe's Topaz announcement. In March 2026, the Department of Justice announced a settlement: $75 million in civil penalties and $75 million in free services, $150 million in total, with a federal judge signing the order in April. Adobe neither admitted nor denied the allegations. This is not ancient history. This is the company paying nine figures to resolve government allegations that its subscription design hid fees and made canceling hard. When that company tells you it will take good care of a product you love, you are entitled to ask how it defines care.

Remember Pantone, too. In 2022, the licensing relationship between Adobe and Pantone broke down, and the fallout landed on users. Pantone color libraries were pulled from Photoshop, Illustrator, and InDesign, certain colors in existing files rendered as black, and accessing some libraries going forward required a separate Pantone Connect subscription. The two companies spent the episode disputing who was responsible, and a later Photoshop update softened the worst of the file-breakage. Assign the blame however you like; it does not change the experience. The customer was the one who opened a finished file and found it broken, then learned the fix was a new monthly charge. That is the recurring shape of these stories: a capability you thought you owned becomes a capability you rent, and the transition is someone else's business dispute that lands on your work.

And in 2024, vague new terms of service language convinced users Adobe was granting itself the right to train AI on their content. The backlash was loud enough that Adobe rewrote the terms and publicly stated it had never trained generative AI on customer content, did not claim ownership of customer work, and would never use customer content to train generative AI tools. The relevant fact is not whether the original fear was justified. It is that Adobe wrote terms broad enough to provoke the fear, and only narrowed them when enough people threatened to leave. The defaults tell you what a company would do if it could. The walkbacks tell you what its customers will not tolerate. Topaz's customers are about to inherit Adobe's defaults.

None of these were emergencies the day they were announced, either. They were reassurances, right up until they were invoices.

The Plugin Problem Nobody Is Naming

There is a quieter mechanism in this deal that deserves its own paragraph, because it is where the squeeze actually happens. For years, Topaz did not just sell standalone apps. It sold plugins and panels that ran inside Photoshop, Lightroom, and more recently Premiere, which meant Adobe's own platform was the distribution channel for a competitor's enhancement tools. Every time you sent an image from Lightroom to Topaz and back, you were using Adobe's ecosystem to reach a product Adobe did not own.

Adobe has now bought the most important supplier in that arrangement. To be careful about the claim: Adobe has not said the plugin model is going away. What it has said is that it intends to integrate Topaz's models into Firefly, Firefly Services, and the Creative Cloud apps. But integration changes the incentive structure even if nobody ever bans anything. The host platform now owns the most important enhancement tool that used to sit beside it, which means the company deciding how well third-party tools plug into Photoshop is the same company now selling the leading one. The independent developers who still build enhancement tools for Adobe's platforms, the DxOs and ON1s and Skylums of the world, now compete against a rival that also controls the doorway they have to walk through to reach customers. Adobe does not have to ban them. It just has to bundle a good-enough version of what they sell into a subscription tens of millions of people already pay for, and let the math do the rest. If you're as old as I am, this should feel familiar. It is roughly what Microsoft did to Netscape in the 1990s, bundling Internet Explorer into Windows until the standalone browser could not compete, and it took a landmark antitrust case to call it what it was.

The Part Regulators Will Probably Miss

If you remember Adobe's attempt to buy Figma, you might assume the Topaz deal faces the same gauntlet. It almost certainly does not, and that is the worrying part.

Figma was a direct competitor to an Adobe product, a $20 billion deal that European and British regulators treated as a clear horizontal threat. After the UK competition authority escalated its inquiry and provisionally found concerns, Adobe and Figma terminated the deal in December 2023, concluding there was no realistic path to approval, and Adobe paid a $1 billion breakup fee. Topaz is almost certainly a different regulatory problem: adjacent, complementary, and undisclosed in value rather than a $20 billion head-on challenge to a flagship product. That is the kind of deal that tends to draw a much harder-to-prove antitrust theory, which is precisely why this one may pass quietly while Figma collapsed in public. You can already see Adobe playing to that scoreboard. The announcement does talk about integrating Topaz across Adobe's portfolio, but it wraps that in repeated promises of continued support and standalone availability, and after the Figma humiliation that framing is not an accident. The company learned exactly which adjectives keep regulators calm.

That is how dominant companies actually consolidate. Not through the one giant merger that draws every regulator's attention, but through the steady absorption of smaller, adjacent, complementary players that individually look harmless. Buy the enhancement specialist. Buy the on-device inference technology that lets large models run on a consumer GPU, the exact technology that could have pressured the metered-cloud model, so it can be absorbed rather than competed with. Build it into the subscription that tens of millions of people already pay for. Each step is defensible in isolation. The sum is a company that owns generative AI and enhancement AI, in the cloud and on your laptop, across the entire pipeline from capture to export, with the independent alternatives picked off one by one. Antitrust enforcement is much better at seeing the dramatic merger than the slow roll-up. Adobe appears to understand this better than the regulators do.

And the alternatives are being picked off. In 2024, Canva bought Affinity, the one serious one-time-purchase challenger to Adobe's suite, then bought the generative startup Leonardo a few months later. Adobe bought Frame.io before that. The Affinity story comes with a caution that actually sharpens the point rather than blunting it: Canva initially promised continuity and no forced subscriptions, then later folded Affinity into a new free, Canva-integrated model on its own timetable. The lesson is not that acquisition always means an immediate price hike. It is that acquisition always means the roadmap stops belonging to you. Strip the logos off and the trend is unmistakable: the independent middle of this industry, the companies that were neither tiny hobby projects nor members of a megacorp's portfolio, is being hollowed out from both ends. Topaz was one of the last names in that middle that photographers actually respected. There are fewer of them every year, and there is no mechanism currently working to replace them, because the moment a new one gets good enough to matter, the obvious exit is to sell to exactly the companies doing the consolidating.

What This Costs You, Specifically

Strip away the structural language and here is what it means at your desk. The value of a competitive market is not abstract. It is a large part of the reason Lightroom's denoise got meaningfully better in recent years, which is to say, it improved once Topaz had spent years proving photographers would pay for the good version. Competitors discipline each other. It's the reason why two gas stations at the same intersection will have the same prices 99% of the time. They force the price down and the quality up, and they give you somewhere to go when one of them does something you hate. Every name that disappears from that list removes a little of that discipline. The prices and meters that follow are not a conspiracy; they are just what happens when the pressure that held them down gets bought out. You will not feel it as a single bad day. You will feel it as a slow narrowing of options, a creep in your monthly total, and a growing sense that the tools serve the company's revenue targets more than your photographs.

What Would Change My Mind

I am not interested in outrage for its own sake, so here is the honest test, stated plainly enough that you can hold me to it later. If Adobe keeps the Topaz standalone apps fully updated and fairly priced after the deal closes, integrates the enhancement tools into Creative Cloud without burning generative credits for them, and keeps the plugins working inside competing editors like Capture One and Affinity, then the damage is limited and I will write that column too. That is a real possibility, not a rhetorical concession. Adobe went out of its way to promise continuity, and the Figma collapse gave it concrete reasons to behave.

But watch what happens, not what was promised, and watch on a timeline longer than the news cycle. Given that Adobe was already metering Topaz's models before it owned them, the credit question is the one to watch hardest. If the standalone apps quietly stop getting meaningful updates, if the plugins stop working in rival editors, if the per-use credit costs climb, if the price creeps the way almost every other Adobe price has crept, then we will know this was the pattern all along, and we will know it a year too late to do anything but switch to whatever independent alternative is still standing. Given the decade of evidence, I know which way I would bet.

The tools are still good today. Eric Yang is still there today. The apps still run today. Hold onto that word. With Adobe, "today" has always been the load-bearing part of the sentence.