Tom Blomfield built two fintech unicorns. He spent years guiding the next batch of founders at Y Combinator. Now he reports for duty as a rank-and-file engineer at Anthropic.
The return of the restless winners
Blomfield announced his leave of absence from YC on July 13 to join Anthropic’s compute team. Not as a vice president. Not as head of something. As a member of technical staff. The move caps a visible trend. Seasoned operators who cashed out handsomely during the last tech boom now grind side by side with fresh talent at the labs shaping artificial intelligence. Short hours. Long odds. But the pull proves too strong to ignore.
Instagram co-founder Mike Krieger joined Anthropic as chief product officer in 2024 before shifting to technical staff on its coding efforts, according to a Business Insider report. Andrej Karpathy left his own startup, Eureka Labs, to focus on pre-training at the same company in May. He posted that the next few years at the frontier of large language models “will be especially formative.”
These aren’t desperate career pivots. The participants already sit on fortunes from prior exits. Monzo and GoCardless made Blomfield wealthy. Krieger helped sell Instagram to Facebook for $1 billion. Their decisions signal something deeper about the current moment in technology. AI feels unfinished. The upside looks asymmetric. And sitting on the sidelines carries its own regret.
Chamath Palihapitiya offers the clearest case of full commitment. The former Facebook executive and prominent SPAC investor hadn’t held an operating role in more than a decade. In late June he stepped in as CEO of 8090 Labs, his new enterprise AI coding company. The firm closed a $135 million Series A led by Salesforce Ventures. Palihapitiya wrote on X that he felt convinced what the team builds now “is even more important” and that there was “no decision to make except to be all in.” The TechCrunch article that first highlighted this pattern captured the sentiment perfectly.
Eric Wu followed a similar script. He steered Opendoor through its public listing and a decade of leadership. In 2023 he stepped back. This year he launched NavigateAI, an AI copilot aimed at construction workers. The startup landed $25 million in seed funding. Wu told TechCrunch’s Connie Loizos that he would regret not pursuing something in AI if he looked back in ten years.
But the title change may speak loudest. Peter Bailis spent less than a year as Workday’s chief technology officer. The role involved steering AI efforts across an $8 billion revenue operation. In March he traded the corner office for a member of technical staff position at Anthropic, focusing on reinforcement learning. Workday framed the exit as part of its “next chapter of AI innovation.” The same Business Insider report noted that both Anthropic and OpenAI deploy the flat title to blur lines between research and engineering. Everyone contributes across functions. Status takes a back seat.
And yet compensation doesn’t. H-1B visa data shows Anthropic pays MTS roles between $300,000 and $405,000 base. OpenAI ranges from $210,000 to $530,000. Equity packages at companies valued in the hundreds of billions can still mint new millionaires many times over. The financial incentive exists. But for these veterans the draw runs deeper.
They see the technology at an inflection. Recursive self-improvement. Compute constraints that will define access and capability. Tools already disrupting software markets. One Anthropic blog post reportedly triggered a widespread sell-off in SaaS stocks earlier this year. The Medium analysis from July 13 described how frontier labs now influence public markets in real time. Smart founders in Y Combinator’s recent batches have pivoted from building AI products to supplying the infrastructure those products need. Identity systems for agents. Payment rails. Persistent memory. Sandboxed environments. Insurance for autonomous actions.
The pattern echoes earlier waves. Operators who succeeded in mobile or cloud often sat out the next cycle. This time feels different. The window appears narrow. Talent concentration in a handful of labs creates its own gravity. Recent X discussions following Blomfield’s announcement highlighted the “YC founder to MTS at OpenAI pipeline” now extending to partners and seasoned CEOs. One post noted Anthropic has “humbled everyone from CEOs and CTOs to staff engineers.” Everyone becomes technical staff.
Investors take note. Forbes updated its AI 50 list this year and observed three prior honorees acquired or absorbed by larger players. Scale AI’s CEO left to start Meta’s superintelligence lab. Google paid $2.4 billion for an AI coding startup’s cofounders and technology. The battles for people with proven track records intensify. A Forbes AI 50 feature published July 14 captured how the list reflects rapid consolidation and shifting power.
Construction. Coding. Compute. The applications vary. The motivation holds. These founders and executives don’t need the money. They chase impact. They chase relevance. They chase the rare chance to shape what comes after the current models. Palihapitiya called it more important. Wu feared future regret. Karpathy pointed to formative years ahead.
So the pattern spreads. More veterans will likely announce similar moves in coming months. Some will join existing labs. Others will launch their own bets on agent infrastructure or industry-specific tools. The last wave of winners refuses to watch from the bench. They pick up the ball again. The game has changed. They want to play.
Recent coverage reinforces the momentum. A CRN roundup of the hottest AI startups as of June 30 lists Anthropic, Cognition and others raising billions while pushing agentic systems and coding agents. The CRN article shows capital still flows to teams that combine technical depth with domain focus. Construction copilots fit that mold. So do enterprise coding platforms.
Yet risks remain. High valuations invite scrutiny. Rapid progress in one lab can obsolete work elsewhere overnight. The same Medium piece warned that two AI startup models already face extinction. Pure wrappers without defensibility won’t survive. Execution on hard technical problems matters more than ever.
Blomfield’s compute focus addresses one of those hard problems directly. Availability of resources could limit or accelerate recursive improvement. Karpathy’s emphasis on pre-training targets the foundation. Krieger’s product lens shapes how users eventually interact with these systems. Each brings hard-won experience from building at scale in earlier eras.
The broader market notices. LinkedIn conversations and X threads from the past week buzz with reactions to these moves. Private equity firms still chase status and titles while AI labs flatten hierarchies to attract the best operators regardless of pedigree. The contrast stands out.
In the end the decision comes down to conviction. These leaders believe the next chapter of AI will reshape industries, economies and daily life more profoundly than previous shifts. Missing it would sting more than any board seat or speaking gig could soothe. So they grind. They ship. They accept the MTS label and the long hours that come with it.
The rest of the industry watches closely. If the pattern holds, expect more announcements before summer ends. The winners are back at work.
WebProNews is an iEntry Publication