Lilian Weng’s post on X was startlingly candid, not for its corporate polish, but for its raw human honesty. The cofounder of Thinking Machines Lab – the AI startup she launched with OpenAI alum Mira Murati – announced she was stepping down. Her reason wasn’t a new venture or a strategic pivot, but something far more fundamental: her health. “The amount of consistent stress and workload have pushed me beyond what my health can sustain physically,” she wrote. This wasn’t just a resignation letter; it was a stark memo from the front lines of the AI boom, highlighting a toll that’s becoming impossible to ignore.
Weng’s trajectory reads like a blueprint for modern AI success. From Peking University to a PhD, through stints at Dropbox and Affirm, she joined OpenAI in 2018. There, she helped pioneer work in robotics and safety systems, becoming a vice president of research and safety. In 2025, she leapt into the startup fray with Thinking Machines Lab, known internally as “Thinky.” The pace was relentless, a “sprint” to release their first model, Inkling. But in her post, she described a personal cost measured in constant illness and a gnawing sense of guilt for trying to rest. “If I cannot give it 100% for what I should be responsible for, I fail my job,” she concluded, deciding a scaled-back role wasn’t the answer. She’s leaving the cofounder title behind, seeking a more predictable position, but staying in AI.
Her story is not an isolated incident. It’s part of a worrying pattern emerging at the very top of the AI industry, where the pressure to innovate at light-speed is colliding with human limitations. Just weeks ago, Fidji Simo stepped down from her full-time role as OpenAI’s CEO of AGI due to a flare-up of postural orthostatic tachycardia syndrome (POTS), a chronic condition. In January, xAI cofounder Greg Yang moved to an advisory role after a Lyme disease diagnosis, vowing to “go founder mode on my health.” These are not mere anecdotes; they are data points in a growing crisis of sustainability. The culture of “hardcore” dedication, glorified in tech lore, is showing its cracks. When leaders tasked with shaping our technological future are buckling under the weight of it, we must ask what this model is truly building.
The stress isn’t solely personal; it’s systemic, fueled by a vicious war for talent. Thinking Machines Lab itself is a case study. A Business Insider review in May found that 13 of its 42 founding members had already left, lured away by rivals like Meta and OpenAI brandishing monumental compensation packages. This relentless churn and burn environment creates a pressure cooker where downtime is seen as a luxury and vulnerability as a liability. Weng’s mention of feeling guilt for taking sick leave is a telling symptom of this pervasive mindset. The machinery of innovation, it seems, has little built-in tolerance for human maintenance.
What does this mean for the technology being built? We rightly worry about AI safety and alignment – technical challenges that Lilian Weng herself worked on. But we rarely consider the human infrastructure behind the code. Exhaustion, chronic stress, and ill health are not conducive to clear, ethical, or creative thinking. The “move fast and break things” mantra breaks more than software; it risks breaking the people whose judgment we rely on to steer these powerful systems. If the architects are operating beyond sustainable limits, what unforeseen flaws are being baked into the foundations of our AI-driven future?
Weng’s decision, and those of Simo and Yang, represent a quiet but profound rebellion. They are prioritizing personal well-being over the relentless grind, setting a new kind of precedent. Mira Murati’s supportive reply – “I’m glad that you’re putting your health first” – signals a necessary shift in values. The next frontier in tech might not be a smarter model, but a smarter way of working. It may involve redefining productivity, embracing asynchronous work, and fundamentally rejecting the idea that burnout is a badge of honor. The most innovative move an AI leader can make right now might be to build a sustainable career.
As Weng moves to a new chapter, her departure serves as a critical stress test for the industry’s culture. The race for artificial general intelligence is a marathon, not a sprint, yet we’re running it like a series of desperate dashes. The human body and mind are not infinitely upgradable hardware. Lilian Weng’s story is a powerful reminder that before we can build machines that think, we must protect the people who build them. The future of AI may depend less on compute power and more on our collective ability to find a healthier, more humane pace.
- Pressure to innovate
- Relentless work culture
- High turnover rates
- Impact on health
- Shift in values
- Need for sustainable practices
| Name | Position | Reason for Leaving |
|---|---|---|
| Lilian Weng | Cofounder, Thinking Machines Lab | Health issues |
| Fidji Simo | CEO of AGI, OpenAI | Chronic condition flaring |
| Greg Yang | Cofounder, xAI | Lyme disease diagnosis |
