Why Tech’s Already-Rich Are Grinding Again at AI Labs

A clear pattern is forming in Silicon Valley: founders and executives who already built and cashed out from major tech companies are going back to work, specifically at or around AI frontier labs. Tom Blomfield, who co-founded both GoCardless and Monzo and spent 4.5 years as a Y Combinator partner, announced on Monday he is joining Anthropic's compute team as a member of technical staff. He is joining a list that includes Andrej Karpathy, Mike Krieger, Chamath Palihapitiya, and Peter Bailis.

What happened

Person	Previous role	New role
Tom Blomfield	YC Group Partner (4.5 years), co-founder of Monzo and GoCardless	Member of technical staff, Anthropic compute team
Andrej Karpathy	Founding member of OpenAI, head of AI at Tesla, founder of Eureka Labs	Anthropic pre-training team (joined May 2026)
Mike Krieger	Co-founder of Instagram	Chief Product Officer, Anthropic (joined 2024)
Chamath Palihapitiya	Facebook exec, investor, “SPAC King”	CEO, 8090 Labs (first operating role since leaving Facebook in 2011)
Eric Wu	Ran Opendoor for a decade, stepped back in 2023	Founder and CEO, NavigateAI ($25M seed)
Peter Bailis	CTO at Workday, overseeing AI strategy across an $8B-revenue business	Member of technical staff, Anthropic (joined March 2026)

Blomfield announced his move on Monday, framing it as a leave of absence from Y Combinator. He is not joining as a senior executive. His title at Anthropic is “member of technical staff,” the deliberately flat label that both Anthropic and OpenAI use for nearly everyone on their technical teams regardless of seniority.

Karpathy made almost the same framing when he joined Anthropic’s pre-training team in May, writing that “the next few years at the frontier of LLMs will be especially formative.” Bailis made the sharpest career reversal: he left Workday’s CTO seat, a role covering AI strategy across a company doing $8 billion in annual revenue, in under a year to take that same flat title at Anthropic.

Not everyone is joining a lab

Palihapitiya, known mostly for SPAC deals and co-hosting the “All In” podcast after leaving Facebook in 2011, took his first full-time operating role in over a decade as CEO of 8090 Labs, an enterprise AI coding startup. The company announced a $135 million Series A led by Salesforce Ventures. He wrote on X: “I am convinced that what we are building now is even more important, so there was no decision to make except to be all in.”

Eric Wu, who stepped back from Opendoor in 2023 after running it for roughly a decade, launched NavigateAI, an AI assistant (“copilot”) for construction workers, with $25 million in seed funding. Wu told TechCrunch: “I knew if I looked back in 10 years and didn’t do something related to it, I would probably regret that.”

Why it matters

When people who are already wealthy and operationally proven choose to grind again, it tells you something about where they think value is being created. These are not people chasing a salary. They are signaling, with their time, that the current period of AI development is worth more than the comfort of a board seat or a venture portfolio.

The “member of technical staff” title is worth paying attention to. It is not a courtesy title. Both Anthropic and OpenAI use it as their standard designation for hands-on technical contributors, whatever their biography. That Blomfield, Karpathy, and Bailis are all taking it suggests they want to be close to the actual work, not above it. For anyone evaluating how to integrate AI into their own business, the fact that this level of talent is concentrating at the frontier labs should inform how seriously you take the capabilities those labs are building.

Our take

The FOMO framing is real, but it is probably too simple. These people are not impulsive. Blomfield spent 4.5 years mentoring founders before this. Bailis spent months in a CTO role at one of the largest enterprise software companies in the world before walking away in under a year. The pattern looks less like panic and more like a considered bet that the next two to three years at the frontier will produce results that are much harder to access from the outside.

For businesses trying to figure out which AI tools and vendors to trust, this concentration of serious operators at Anthropic in particular is a signal worth tracking. As we covered in our comparison of Claude Sonnet 5 versus earlier models, Anthropic’s output quality has been improving fast. That improvement does not happen without the kind of talent that is now walking in the door.

The practical risk for companies like Workday or any firm that loses a key AI leader to a frontier lab is real: institutional knowledge and strategic momentum walk out with them.

What to do about it

1. Watch who is joining which lab. Talent concentration is a leading indicator of where model capabilities will improve fastest.
2. Prioritise building familiarity with the models coming out of those labs now, before the capability gap widens further.
3. If you have AI projects sitting on the backburner, move them forward. The people closest to the frontier are treating urgency as a given.
4. If you need help scoping what AI actually makes sense for your business, talk to a team that ships this work daily rather than one that theorises about it.

The smartest operators in tech are voting with their calendars. That is probably the most honest benchmark you will find.

Source: TechCrunch · AI