Cat Wu’s exit from Claude Code is the latest in a string of departures that raises an uncomfortable question: why are AI’s best builders walking away just as the money arrives?
Cat Wu, the product lead behind Anthropic’s Claude Code, left the company on Friday, Oct. 30, and announced it publicly on Nov. 3, just as Anthropic prepares for one of the largest IPOs in history. Anthropic confirmed the departure. Wu thanked her team and users in her farewell post but gave no reason for leaving.
The timing is what turns this into news rather than noise. Anthropic filed its S-1 on Oct. 26, plans to list on Nasdaq under the ticker ANT, and could begin trading as early as the week of Nov. 16. Reports suggest a valuation between $1.5 trillion and $2 trillion.
Why Wu Was a Big Deal
Wu joined Anthropic in summer 2024 after testing an early Claude Code prototype. In her post, she described asking the tool to build a reinforcement learning environment without writing a single line of code herself. “It worked, I was hooked, and I joined the team,” she wrote.
Alongside Boris Cherny, the tool’s original creator, Wu became one of the two public faces of Claude Code. The pair appeared in Anthropic’s launch video in February 2025. They briefly left for Cursor in July 2025, then returned two weeks later, a detour that now reads like an early warning sign.
Wu also helped lead Cowork, Anthropic’s push beyond coding into general knowledge work. Her exit follows Thariq Shihipar and Anthony Morris, two other Claude Code contributors who announced departures on Oct. 24. Three key people from one product team, gone in under two weeks.
Why Leave Before the Payday
The obvious question is why anyone would quit now. Anthropic reported $16.3 billion in revenue for the first half of 2026, up from $4.6 billion in all of 2025. The growth is staggering.
But the same filing shows net losses of $80.7 billion in the first half of 2026, driven by enormous spending on chips and data centers. The money is real, and so is the burn rate.
For people like Wu, the calculus may be simpler than stock tables suggest. She spent two years building a product that went from internal experiment to industry defining tool. At some point, the job shifts from making something new to defending something big. Some builders find that shift draining. Others just want to build the next thing before the current thing becomes a compliance department with a chatbot attached.
Think of it like a chef who opens a beloved restaurant, then leaves once it becomes a chain. The food is still good. The checks still clear. But the chef misses the kitchen, not the franchise meetings.
Why This Keeps Happening Across AI
Wu is not alone. Across the AI industry this year, senior researchers, product leads, and safety experts have stepped away from the biggest labs even as valuations soar. The pattern suggests something deeper than burnout or a better offer. People who joined AI companies to push the frontier are watching those companies turn into public market machines, complete with reporting lines, risk committees, and shareholder expectations.
Anthropic’s S-1 notes that its seven co founders will retain majority voting control through a special share structure, a sign the company knows its identity is tied to a small group of people, not just its balance sheet. When identity is that concentrated, every visible departure gets read as a signal, whether or not it should be.
There is also the simple fact that AI talent has never had more options. Startups, research labs, and even non AI companies are hiring. Someone with Wu’s track record does not need to wait for an IPO to cash out. She can walk away with reputation, experience, and a network, then choose what comes next on her own terms.
Why Anthropic Should Worry Anyway
For Anthropic, the timing is awkward but not fatal. The company still has Cherny, its leadership team, and a product that clearly resonates with developers. The S-1 shows revenue growth most public companies would trade a limb for.
But IPOs are as much about narrative as numbers. Investors buying into ANT are betting on the people as much as the product. Losing the face of your fastest growing tool weeks before listing is the kind of detail that makes analysts pause mid sentence.
Wu signed off her post with “post agi!”, a joke that lands differently depending on whether you think artificial general intelligence is around the corner or just a very expensive group project. Either way, she is leaving before the confetti drops. In a race where everyone is supposed to be chasing the prize, some of the fastest runners are deciding the finish line is not where they want to be.
