Weng’s Departure Follows Seven Months of Mounting Health Issues

Lilian Weng, co-founder of AI startup Thinking Machines Lab, announced on Tuesday that she is leaving the company, describing a seven‑month period of repeated illness and unsustainable pressure. Weng, who previously served as OpenAI’s vice president of research and safety, said the pace and stress of building a startup had pushed her “beyond what my health can sustain physically.”

In a message shared on X and with colleagues at the firm — known internally as “Thinky” — Weng wrote that despite the laughter and challenges, she could not continue to give “100% for what I should be responsible for.” She noted that even attempting to rest properly during the sprint to release the company’s first open‑weight model, Inkling, left her with guilt. Ultimately, she concluded that a narrower, less anxiety‑inducing role was not a fit, and she would seek a more predictable position outside the co‑founder spotlight while staying in AI.

Thinking Machines Lab, co‑founded with former OpenAI CTO Mira Murati, has already seen significant attrition. A Business Insider review in May found that 13 of the company’s 42 founding members had departed, including three of its six co‑founders, as rivals such as Meta and OpenAI offered large compensation packages.

The Broader Exodus at Thinking Machines Lab—and the AI Industry’s Health Reckoning

Thinking Machines Lab’s Founding Team in Flux

Weng’s exit is the latest in a string of departures that have hollowed out the startup’s original leadership. Losing a sixth of the founding team — and now another co‑founder — raises questions about the company’s ability to maintain momentum in a market where execution speed is everything. While Murati’s public reply was supportive, the steady outflow of founding members to larger rivals signals a fierce talent war in which well‑funded startups struggle to hold onto top people. The 13 departures uncovered by BI suggest that the very conditions Weng blamed — relentless pace, high stress — may be endemic at the firm.

A Growing Pattern of Health‑Related Departures in AI

Weng’s decision does not stand alone. Earlier in July, Fidji Simo stepped down from her full‑time role as OpenAI’s CEO of AGI after a flare‑up of postural orthostatic tachycardia syndrome (POTS), moving to a part‑time advisory position to focus on recovery. In January, xAI co‑founder Greg Yang shifted to an informal advisory role following a diagnosis of Lyme disease, saying months of unexplained fatigue had sapped his creativity and that he needed to “go founder mode on my health.” These cases — spanning different AI labs — point to an industry‑wide reckoning with burnout. The AI arms race demands extraordinary hours and mental bandwidth, and even well‑compensated, highly esteemed leaders are stepping away when the physical toll becomes too great.

What Weng’s Exit Signals for Startups and Founders

  • For startup boards and investors: Weng’s description of guilt during an official leave period while the company was “sprinting” reveals a culture that makes recovery difficult. Boards should examine whether workload expectations are compatible with retaining key talent, especially co‑founders whose departure can reverberate across the organisation.
  • For AI talent competitors: The pattern of founding members leaving TML for Meta and OpenAI, combined with co‑founder exits, indicates that the industry’s compensation‑driven poaching is destabilising smaller labs. Recruiting may offer quick wins, but the broader industry risks a brain drain from the very startups that often pioneer new ideas.
  • For health‑sensitive risk management: With high‑profile health exits now occurring at OpenAI, xAI, and TML, investors should recognise burnout as a material operational risk in AI ventures. Diligence questions could address safeguards, sabbatical policies, and psychological support, rather than assuming a “grind” culture is sustainable.

Risk & Opportunity Assessment

Commercial RiskHighLosing a co‑founder and a sizeable chunk of the founding team undermines execution capacity at a critical stage, especially with the Inkling model launch recent and competitive pressure high.
Competitive RiskHigh13 of 42 founding members had already left, many to larger rivals like Meta and OpenAI offering huge compensation packages; further departures could accelerate competitive disadvantage.
Regulatory RiskLowNo regulatory issues are mentioned in the story; the departure is health‑ and culture‑driven.
Reputation RiskMediumOngoing exits and a public statement about unsustainable stress could damage the startup’s employer brand and fuel perceptions that its work culture is unhealthy, making it harder to attract and retain talent.
Technology DisruptionLowThere is no immediate technology threat from the departure; the core disruption is human capital, not a rival’s technological advance.
Commercial OpportunityLowNo specific commercial upsides arise from this event, though rivals may see an opportunity to recruit remaining TML talent.