The Architect's Exit
When the man who built the cathedral walks away, ask what he sees that you don’t.
Shantanu Narayen is leaving Adobe.
Not tomorrow. Not in a dramatic blaze. He’s doing what architects do — stepping back to survey the building one last time before handing the keys to someone else. After eighteen years as CEO, he’ll remain as Board Chair while the company finds his successor. The search should take “a few months.”
The timing is exquisite. And by exquisite, I mean suspicious.
The Numbers That Don’t Make Sense
Adobe just posted its best quarter ever. Revenue of $6.40 billion, up 12% year-over-year. Earnings of $6.06 per share, beating estimates by 3%. Subscription revenue up 13%. Cash flow of $2.96 billion. And the headline number: AI-first annualized recurring revenue more than tripled.
“That should be our next billion dollar business,” Narayen told analysts.
The stock dropped 7% after hours.
This isn’t a bug. It’s the market telling you something that the earnings report can’t.
The Cannibal’s Dilemma
Here’s what’s actually happening inside Adobe: the company is eating itself.
Adobe Stock — the service that sells stock photos, illustrations, and media to creatives — represents about $450 million in annual revenue. This quarter, it declined faster than management expected. David Wadhwani, president of Adobe’s creative business, acknowledged on the call that “this shift is playing out more quickly than we had planned for.”
The shift he’s referring to? Adobe’s own AI tool, Firefly, is replacing the need for stock photos. Why buy a stock image when you can generate one that’s exactly what you need? Adobe built the replacement for its own product and is watching the substitution happen in real-time.
This is the Cannibal’s Dilemma: you can either eat your own lunch or let someone else eat it. Adobe chose to eat its own. But here’s the uncomfortable truth — when you cannibalize your own product, you still lose the revenue. The bet is that AI-first products will generate more revenue than the products they’re destroying. That bet hasn’t paid off yet.
The AI-first ARR tripling sounds impressive until you realize it’s tripling from a small base, while the products it’s replacing had decades of accumulated revenue.
The Gerald Carter Story
If the financial paradox isn’t enough, there’s a human one.
Gerald Carter spent years building Diversity Photos — a stock photography archive specifically focused on minority communities, who have been historically underrepresented in commercial imagery. He partnered with Adobe in 2018 through a Stock Contributor Agreement. Revenue share. Mutual benefit. Standard deal.
Then, in 2023, Carter discovered that Adobe had fed his entire library — nearly 12,000 images — into Firefly’s training data. Without asking. Without a separate license. Without compensation.
Adobe’s defense? The 2018 agreement already gave them the right.
“They took our competitive advantage and turned it into their feature,” Carter told PetaPixel. Diversity Photos’ entire value proposition was scarcity and specificity. Now Adobe’s AI can generate similar diverse imagery on demand.
Adobe offered Carter $1,173.93 as a “bonus.” Not compensation. Not a licensing fee. A bonus. For 12,000 images used to train a model that Adobe monetizes through subscriptions.
Carter rejected it. Adobe sent a process server to his home.
This is the part that doesn’t make it into the earnings call. The part where “AI-first ARR tripling” has a human cost denominated in destroyed livelihoods, broken partnerships, and creative communities whose work was harvested to build the tool that replaces them.
The Architect Knows
Shantanu Narayen is not a fool. He’s the man who transitioned Adobe from perpetual licenses to subscriptions — one of the most successful business model transitions in software history. Under his watch, the stock increased sixfold. He knows how to read the trajectory of a business.
So when the architect chooses this moment to announce his exit — record revenue, AI momentum, but a stock down 60% from its 2021 high and 23% YTD in 2026 — what does he see?
I think he sees the Innovator’s Dilemma made flesh.
Adobe is simultaneously the incumbent being disrupted and the disruptor doing the disrupting. It’s attacking its own castle from the outside while trying to defend it from the inside. This is a fundamentally different company than the one Narayen took over in 2007, and it needs a fundamentally different leader.
The next Adobe CEO won’t be managing a software subscription business. They’ll be managing an intelligence subscription business. The product isn’t Photoshop — it’s the AI that makes Photoshop unnecessary for 80% of use cases.
850 Million Monthly Users
Here’s the number that actually matters: Adobe now has 850 million monthly users across Acrobat, Creative Cloud, Express, and Firefly. Up 17%.
That’s a staggering number. It means Adobe is becoming a platform that reaches nearly a billion people. Most of them aren’t paying full Creative Cloud prices. Many are using free tiers or Express. The question isn’t whether Adobe has users — it’s whether it can monetize the AI layer on top of them.
This is the “tool provider to intelligence provider” transition that every legacy software company is attempting. Microsoft is doing it with Copilot. Google is doing it with Gemini in Workspace. Salesforce is doing it with Agentforce. Adobe is doing it with Firefly.
The playbook is the same: take your existing user base, add an AI layer, charge more. The problem is also the same: the AI layer commoditizes the very expertise that made your users loyal to your tools in the first place.
If Firefly can generate what a Photoshop expert used to create, why pay for the expert or the tool?
The Real Question
Adobe’s stock isn’t falling because the numbers are bad. The numbers are great. It’s falling because the market is pricing in a future where:
- AI commoditizes creative tools. Canva, Figma, and dozens of AI-native startups can now do what required Photoshop.
- The moat is unclear. Is it the tools? The user base? The training data? The ecosystem? Adobe’s moat used to be obvious: professionals had to use Photoshop. That’s no longer true.
- The transition math is uncertain. AI-first ARR tripling still doesn’t compensate for the erosion of Stock, the commoditization of basic creative tasks, and the competitive pressure from AI-native tools.
Satya Nadella congratulated Narayen and praised his “empathy for the creative process.” Dylan Field — the Figma CEO whose $20 billion acquisition Adobe tried and failed to complete — called Narayen “thoughtful, kind, and relentless.”
These are the words you say when someone is leaving. They’re also the words you say when an era is ending.
The Lesson
Every software company is now running the same experiment: can you replace yourself faster than your competitors can replace you?
Adobe is further along than most. Firefly is real. The AI-first ARR is real. The 850 million users are real. But so is the $450 million stock photo business that’s declining faster than expected. So is Gerald Carter’s 12,000 images, ingested without consent, used to build the very product that destroys his business.
The architect sees all of this. The record revenue and the structural vulnerability. The AI momentum and the creative community’s erosion of trust.
Maybe that’s why he’s leaving. Not because the building is falling down — but because the next building needs to be something entirely different.
And he’d rather be remembered as the architect of the cathedral than the foreman of the demolition.
Adobe’s stock, 60% off its 2021 high, is the market’s way of saying: we believe you’re being disrupted, even by yourself. Record revenue is the rearview mirror. The windshield shows something else entirely.