Anthropic's product rockstar just walked across the street to OpenAI three weeks before a $60B IPO, taking institutional knowledge of what made Claude Code work with her.

The Summary

The Signal

Wu joined Anthropic in summer 2024 and quickly became half of the two-person leadership team behind Claude Code, alongside original creator Boris Cherny. In roughly two years, she helped turn an internal experiment into what the sources call "one of the most important products in the AI industry's history." That's not hyperbole when you consider what Claude Code did for Anthropic's position. It gave the company viral distribution, boosted revenues enough to fund more compute, and made them a credible OpenAI competitor instead of a well-funded also-ran.

Now that institutional knowledge walks into OpenAI's office. Wu cited OpenAI's reach of over 1 billion users as part of her reason for joining. She'll lead product management for the core products team, which likely means touching everything from ChatGPT to Codex, OpenAI's own coding tool.

"The poach is a win for OpenAI ahead of its rival's hotly anticipated IPO in a few weeks."

The timing matters more than usual startup musical chairs. Anthropic is weeks away from its IPO, reportedly targeting a $60B valuation. Losing a top product leader this close to going public sends a signal, even if Wu's departure was amicable. Investors will ask questions. Early employees cashing out before an IPO is normal. Key product leaders jumping to your main competitor is not.

This is also the second Claude Code leader to leave recently. Thariq Shihipar, who also worked on the tool, departed last month. Two exits from the same product team in quick succession suggests either the work is done or the team knows something about what comes after IPO that made leaving attractive.

Key context on what Wu built:

  • Turned Claude Code from prototype to commercial product in under two years
  • Became one of the most visible product evangelists in AI on social media
  • Helped establish coding ability as table stakes for frontier AI models

What OpenAI gets is someone who knows how to ship agent-adjacent products that developers actually use. AI models' coding abilities and their capacity for long-term tasks are now central to the industry's growing ambitions. Claude Code proved people will pay real money for AI that writes, debugs, and ships code. Wu knows which product decisions made that happen and which ones were theater.

The Implication

Watch what Wu ships at OpenAI in the next six months. If she's leading core products, that probably means ChatGPT and Codex get rethought through the lens of what worked at Claude Code. OpenAI already has the users. They now have someone who knows how to turn coding tools into revenue generators at scale.

For Anthropic, this is the IPO tax. Going public means early employees get liquidity and options. It also means your best people get recruited harder. Losing Wu doesn't kill Claude Code, but it does mean whoever takes over needs to learn the product intuition she built while Anthropic is also dealing with being a public company. That's a harder transition than it sounds.

Sources

Business Insider Tech