Adobe
The Death Narrative
The narrative behind Adobe right goes something like this.
Generative AI has commoditized the hardest part of creative work. Anyone with a text prompt can produce a professional-looking image in four seconds. The next generation of designers, video editors, and content creators is growing up inside ChatGPT, Gemini, and Midjourney — not Photoshop. By the time today’s casual AI user becomes tomorrow’s marketing director, Adobe will feel like what WordPerfect feels like today. A thing your parents used. The company had a good run. The creative TAM is about to compress and Adobe is standing in front of it.
It is a clean narrative. It is internally consistent. And the market clearly believes it — Adobe’s stock is down roughly 39% over the past year and trades at a forward PE of around 10.5 times earnings, one of the lowest multiples the company has seen in a decade. That is closer to what you assign a business you believe is in structural decline or stagnation.
I want to take this narrative seriously before I dig into it. Because the version of Adobe’s future where the bears are right is worth understanding precisely — not to scare you, but because an investment thesis you haven’t genuinely stress-tested is not a thesis. It is a hope.
How Adobe Actually Dies
The bear case, stated at its most honest, does not require AI to be smarter than a designer. It requires AI to be good enough, and already open in the tab the user lives in.
Here is the ten-year sequence that ends badly for Adobe.
It starts with habit formation. Right now, hundreds of millions of people are building daily routines inside ChatGPT, Claude, and Gemini. These are not creative tools. They are general intelligence platforms that people use to write emails, summarize documents, answer questions, and plan their days. Creative work is incidental. But increasingly, when someone in that workflow needs a quick image — a social post visual, a presentation graphic, a product mockup — they just do it there. They never think to open a separate application. The thought doesn’t occur to them.
This is not a quality problem. It is a habit problem. And habit problems are the hardest kind to solve in software because they don’t require anyone to make a conscious decision to switch away from Adobe. They just require a generation of users to never form the habit of going to Adobe in the first place.
The second stage is the interactive canvas. Right now, refining an AI-generated image through natural language prompts is genuinely frustrating. Describing spatial, visual, precise changes in text is a terrible interface. “The shadow on the left side of his face, not the right, and only in the lower third” takes twenty iterations in a chat window and forty-five seconds in Photoshop. That friction is Adobe’s current moat.
But the moat is not permanent. The moment any major LLM provider ships a mouse-driven, layer-aware, spatially precise creative environment natively inside their platform — where you click, lasso, and brush rather than describe — the interface advantage Adobe has built over thirty years begins to erode for the casual and mid-market user. It doesn’t have to be as good as Photoshop. It has to be good enough, inside the platform where the user already lives.
The third stage is the generational succession problem. Adobe’s professional base is sticky in a way that is almost impossible to dislodge. A video editor who has spent fifteen years building muscle memory in Premiere Pro, custom workspace layouts, and specific plugin integrations has made a personal capital investment in that tool. That switching cost is real and large. Adobe does not have to earn the retention of that user. Inertia does it for them.
But that user retires. And the question of who replaces them is entirely open. If the next generation of people who would have become Adobe power users grows up forming deep habits inside LLM ecosystems, their escalation path — the natural graduation from casual tool to professional tool that has historically deposited new users into Adobe’s ecosystem — gets interrupted. They don’t switch away from Adobe. They never arrive.
Compress this into a ten-year picture and the bear case becomes coherent. Not a collapse. A gradual narrowing. Adobe’s professional core holds but ages. The casual and prosumer market — the territory Adobe needed to expand into to justify growth expectations — goes to whoever owns the LLM interfaces. ARR growth decelerates from 11% to 8% to 6% over successive years, each time with a plausible explanation, each time just a little below what the market expected. The stock re-rates lower. Not bankrupt. Not even bad. Just a business that was once a compounder and became a value trap.
That is how Adobe dies slowly. And it is not an unreasonable scenario.
The Narrative Has a History Problem
We have seen this movie before. Many times. And the ending is almost always the same.
In 1979, Dan Bricklin shipped VisiCalc for the Apple II, the world’s first electronic spreadsheet. The people whose jobs it most threatened understood immediately what it meant. Show it to a working accountant and, as Bricklin recalled, he’d start shaking and say “I spent all week doing that.” The conventional prediction was demand for accountants would collapse as the machine took over the arithmetic.
It was completely wrong. US Bureau of Labor Statistics data put total accountants and accounting clerks at roughly 339,000 in 1980. By 2022, there were approximately 1.4 million accountants and auditors in the United States. The spreadsheet didn’t compress the profession. It created an entirely new one — the financial analyst, the FP&A team, the LBO modeler — that had not existed as a meaningful occupation in 1978 and today numbers in the hundreds of thousands.
In January 1985, Aldus shipped PageMaker for the Macintosh and Paul Brainerd coined the term “desktop publishing.” The International Typographical Union — which had existed since 1852 — watched its membership halve between 1984 and 1987 and dissolved itself into the Communications Workers of America in 1986. Massimo Vignelli, perhaps the most celebrated designer of his generation, described desktop publishing as “a disaster of mega proportions. A cultural pollution of incomparable dimension. As I said at the time, if all the people doing desktop publishing were doctors we would all be dead.”
American graphic designers numbered roughly 133,000 in 1982. The Bureau of Labor Statistics Occupational Outlook Handbook reports 265,900 graphic design jobs in 2024. The profession roughly doubled in the four decades since the machine that was supposed to destroy it arrived. Vignelli’s pollution turned into a multi-hundred-billion dollar global creative industry.
In 2016, Geoffrey Hinton — who would later win the Nobel Prize in Physics for his contributions to deep learning — delivered perhaps the most quotable doom prediction in the history of AI. “People should stop training radiologists now,” he said at a Toronto machine learning conference. “It’s just completely obvious that within five years deep learning is going to do better than radiologists.” He compared radiologists to the cartoon coyote who has already run off the cliff but hasn’t yet looked down.
Nine years later, US radiology residency programs offered a record 1,208 positions in 2025 — a 4% increase over the prior year. The Mayo Clinic’s radiology staff grew from roughly 260 in 2016 to over 400 in 2025. Radiologist compensation reached $520,000 to $571,000, among the highest of any American medical specialty, up roughly 48% from 2015. Hinton told the New York Times he “spoke too broadly.”
The pattern across every case is the same. When a tool lowers the floor on who can participate in a creative or knowledge discipline, latent demand floods in. Tasks that were uneconomical at $10,000 become routine at $100. The professionals don’t disappear. They move up the value chain into judgment, taste, integration, and orchestration, while a much larger layer of new participants forms underneath them. The total addressable market doesn’t get sliced. It gets re-platformed onto a bigger substrate.
There is a name in economics for why this happens. William Stanley Jevons observed in the 19th century that James Watt’s improvements to the steam engine did not reduce British coal consumption — they dramatically increased it, because more efficient engines made steam power economical for uses it had never been applied to before. When you make something dramatically cheaper, you don’t just get the same demand at a lower price. You get entirely new demand from use cases that weren’t economical before. Creative work is no different.
The world is on pace to take approximately 2.1 trillion photographs in 2025 — at least a twenty-fold expansion over the late film era. Professional photographers grew 18% in the United States between 2014 and 2024, during the smartphone era that was supposed to have eliminated them. GarageBand shipped on over a billion iOS devices and Lil Nas X made “Old Town Road” — the longest-running number one in Billboard history — for $30 of beats off the internet, while professional sound engineering technician employment hit an all-time high. The App Store went from zero to 2.4 million jobs in the United States in fourteen years.
The through line is consistent enough that it deserves weight as a base rate. The default expectation when a powerful new creative tool democratizes access to a discipline is not compression. It is explosion.