Cat Wu Leaves Anthropic: What the Claude Code Exit Signals Before the IPO

When Cat Wu leaves Anthropic weeks before a widely anticipated public offering, the departure is more than a personnel footnote. Wu, who led product for Claude Code, left the company on Friday and announced it on X on Tuesday. She was one of two public faces of what has become one of the most consequential products in the AI industry. This analysis separates what is confirmed from what is speculation, and asks what the exit means for Anthropic, its customers, and its prospective investors.

The confirmed facts are thin. Anthropic has confirmed the departure. Wu has not said why she left. She did not immediately respond to requests for comment. Everything beyond that is interpretation, and this article labels it as such.

What Is Confirmed About the Departure

According to Business Insider’s reporting, Wu joined Anthropic in the summer of 2024. She tested an early prototype of Claude Code and, in her own words, was hooked and joined the team. Anthropic released the tool in research preview in February 2025, with a brief demo video featuring Wu and its original creator, Boris Cherny.

The Information reported that Wu and Cherny left for the AI coding startup Cursor that summer and returned to Anthropic a couple of weeks later. Wu then oversaw the tool’s rise, which accelerated at the end of 2025. She also oversaw Cowork, Anthropic’s product for non-coding knowledge work built on the same models.

Her exit follows that of Thariq Shihipar, another Claude Code team member, who announced his departure last month. Anthropic is preparing what the report calls a gargantuan initial public offering later this month.

Why Claude Code Matters to Anthropic’s Position

To judge the significance of a product leader’s exit, you have to understand what the product did for the company. Claude had been a respected model family, but Anthropic lacked a breakout consumer-visible product. Claude Code supplied one. It began as an automatic coder for software engineers and could work through entire coding tasks rather than suggest single lines.

The reported commercial effect was direct. Large technology companies’ AI bills with Anthropic climbed as more developers consumed model tokens. That revenue, in turn, helped fund access to the compute that frontier model training demands, and it made Anthropic a more credible rival to OpenAI. The causal chain runs from a product, to usage-based revenue, to compute funding, to competitive standing. Any threat to the first link therefore deserves scrutiny, even if the evidence of a threat is currently absent.

The Rise of Agentic Coding

Claude Code belongs to a broader shift toward agentic tools, in which a large language model plans, executes, and iterates on multi-step work instead of answering one prompt at a time. This shift also fed the cultural phenomenon often called vibe coding. Claude Code’s popularity reflects demand for that workflow, but it also exposes Anthropic to a crowded field where rivals compete on the same developer attention.

What a Head of Product Actually Does

Outsiders often treat the product manager role as secondary to the engineer who builds the technology. Wu’s own description of her work argues against that view. On Lenny’s Podcast in April, she described her job as charting the path to the ‘AGI-pilled’ version of Claude Code that Cherny envisioned, a reference to artificial general intelligence.

In practice, according to the report, that meant working across marketing, sales, finance, and other teams. It also meant hiring some of Anthropic’s product managers and responding to user requests on social media. That is a wide remit. It combines product management, talent building, and public advocacy in one person.

The Concentration Risk

The breadth of that remit is exactly what makes a departure analytically interesting. When one leader is the connective tissue between engineering, go-to-market, and the user community, the organization’s knowledge of how those functions interact is partly informal. Formal org charts do not capture it. Whether Anthropic has distributed that knowledge is unknowable from the outside, and the company has not said who will take over her responsibilities.

Reading the Timing: Signal or Noise?

The timing invites two competing readings, and neither is supported by hard evidence.

The Benign Reading

Senior employees at fast-growing startups frequently leave around liquidity events, when equity has been valued at a high level and the founding-era work feels complete. Wu’s joking reply to a commenter who asked whether she would take a six-month vacation, ‘post agi!’, suggests a light tone rather than conflict. A departure after shipping a defining product and before a public listing is, in this view, ordinary career timing.

The Concerning Reading

The skeptical reading points to the cluster. Shihipar left last month, Wu followed, and Cherny remains as the creator. Two exits from a small team in about a month, just before an IPO, could indicate friction over strategy, workload, or the transition from startup to public company. That interpretation is plausible but unproven. Two data points do not establish a trend, and Wu’s silence on her reasons means any motive attributed to her is conjecture.

A disciplined observer should hold both readings open. The honest conclusion is that the available facts cannot distinguish between them.

The Cursor Detour as Context

One detail deserves more attention than it received. The Information reported that Wu and Cherny briefly left for Cursor in summer 2025 and returned within weeks. That episode shows the market value of the two leaders and the fluidity of talent in AI coding. It also shows that a leader’s return is not guaranteed to be permanent, and that competitors treat the Claude Code team as a recruiting target. Wu has not said where she is going next, and no source in the report suggests a destination.

What This Means for Investors and Customers

The IPO context changes who needs to pay attention. A public offering invites scrutiny of key-person dependency, meaning how much of a company’s value depends on specific individuals. Prospective shareholders will weigh this alongside revenue growth, compute costs, and competition. Departures of public-facing product leaders do not by themselves reduce revenue, but they feed the narrative around execution risk.

For Enterprise Buyers

Companies that have integrated Claude Code into engineering workflows should focus on continuity, not personalities. The practical questions are these:

• Does the product roadmap and release cadence remain stable over the next two quarters?
• Is support responsiveness, which Wu personally contributed to on social media, maintained?
• Do pricing and token-consumption terms stay predictable as the company becomes publicly accountable?
• Is there a named successor for the product function?

For Developers

Individual developers should track changelogs and the official Anthropic website for announcements rather than inferring product direction from staff movements. Avoid architecting workflows that depend on one vendor’s undocumented behavior, since leadership change is one of several sources of product drift.

For Investors

Watch the offering documents for how the company describes dependence on key personnel and on Claude Code’s revenue share. Those disclosures will say more than any social media post.

Common Mistakes in Interpreting Executive Departures

1. Treating timing as causation. A departure before an IPO does not mean the departing person lacks confidence in the offering.
2. Over-weighting public visibility. The most visible leader is not always the most operationally critical one.
3. Ignoring the silence. When a person does not explain a departure, the absence of explanation is not evidence of conflict.
4. Extrapolating from two exits. Small samples mislead, especially in a company with rapidly growing headcount.

Where Claude Code and Cowork Go From Here

The next developments are reasonable to watch but not confirmed. Anthropic could name a new head of product for Claude Code, or fold the role into a broader structure that also covers Cowork. Both tools rely on the same underlying models, which argues for tighter integration. How the company handles that integration will show whether Wu’s departure reflects a planned handoff or an unplanned gap.

Competition will also shape the outcome. OpenAI and AI-native coding startups are all pursuing agentic software development, and a public company faces quarterly pressure to defend its lead. Talent retention across the Claude Code team is the leading indicator to monitor.

Frequently Asked Questions

Who is Cat Wu?

Cat Wu was head of product for Claude Code at Anthropic and one of its two top leaders and evangelists alongside creator Boris Cherny. She joined in summer 2024 and also oversaw Cowork.

Why did Cat Wu leave Anthropic?

She has not said. Anthropic confirmed her departure but neither the company nor Wu has given a reason, so any explanation is speculation.

When did Cat Wu leave Anthropic?

She left on Friday and announced it on X the following Tuesday, weeks ahead of Anthropic’s planned IPO.

What is Claude Code?

Claude Code is Anthropic’s AI coding tool, released in research preview in February 2025. It can carry out entire coding tasks for software engineers.

What did Cat Wu mean by ‘post agi’?

It was her reply to a commenter asking if she would take a long vacation. It is a joke implying she would rest after artificial general intelligence arrives. It is not a statement of her plans.

The Question the IPO Documents Must Answer

The most important insight is structural, not personal. Anthropic’s commercial momentum was built on a product led by a very small group, and two members of that group have now gone in about a month. That does not prove fragility. It does mean the company must show that Claude Code’s success is institutional rather than dependent on a few individuals.

Watch three things: who inherits Wu’s responsibilities, whether Cherny stays, and how the offering documents describe key-person risk. If the product cadence holds through the first public quarters, this exit will look like routine turnover at a fast-growing company. If it does not, Wu’s departure may be remembered as the first sign that the Claude Code era had passed its founding chapter. The evidence does not yet say which, and that uncertainty is the real story.