If AI Helps People Work Less, Why Are AI Lab Employees Working So Much?

A BBC report reveals that employees at AI labs like OpenAI and Anthropic work 50 to 90-hour weeks despite public promises that AI would enable shorter workweeks. OpenAI President Greg Brockman has documented his own 90-hour work weeks, and Cerebras CEO Andrew Feldman dismissed the idea of working 38 hours a week with work-life balance as 'mind-boggling.' Research from Harvard Business School and the AI Work Institute indicates AI intensifies work rather than reducing it, with employees saving up to 11 hours per week but spending 6.5 of those hours maintaining AI workflows.

Publicly, AI labs like OpenAI and Anthropic are urging other companies to prepare for the AI labor apocalypse, promising productivity boosts so big that businesses will be able to shift to four-day https://www.businessinsider.com/openai-superintelligence-ai-upheaval-tax-shorter-workweek-public-wealth-fund-2026-4 —or even three-day https://www.windowscentral.com/artificial-intelligence/zoom-ceo-ai-could-enable-3-day-workweek —workweeks. These same companies are the earliest of early adopters, openly bragging about how their workforce uses AI tools to tackle upwards of 100% of their normal work tasks https://fortune.com/2026/01/29/100-percent-of-code-at-anthropic-and-openai-is-now-ai-written-boris-cherny-roon/ . So why, according to a report from the BBC https://www.bbc.com/news/articles/cvgx4yd1gl2o , do former employees of these companies claim that they’re working 90-hour weeks, working and taking meetings during weekends, and just generally experiencing the same old Silicon Valley grindset mentality? Part of it almost certainly is the ongoing arms race that the AI labs find themselves in, constantly looking to churn out new versions of their models to top their competitors. Per the BBC’s reporting https://www.bbc.com/news/articles/cvgx4yd1gl2o , code sprints remain a staple of the tech company even in the AI era, with teams tasked with working day and night until they finalize a feature or product in an effort to push it out faster. But even outside of crunch time, the publication said that employees of OpenAI reported working 50 to 60 hours a week as the standard. And then there are the executives who seem to think these types of work conditions are acceptable—necessary even—to reach their goals. It’s not hard to find examples of tech execs lionizing themselves about how much time they put in to get to where they are—OpenAI President Greg Brockman has documented his own 90-hour work weeks https://x.com/gdb/status/726099677716713472 in the past. And then you get guys like Andrew Feldman, cofounder and CEO of AI chip company Cerebras, who say things https://fortune.com/article/ceo-of-49-billion-ai-company-says-its-mind-boggling-people-think-you-can-work-38-hours-a-week-have-work-life-balance-and-be-successful/ like, “This notion that somehow you can achieve greatness, you can build something extraordinary by working 38 hours a week and having work-life balance, that is mind-boggling to me.” Cerebras seemingly assumes there is only one way to be great, and it is through work, and that working more hours guarantees productivity and is not just an act of performative busyness. Reality is, whatever productivity “gains” have been made via the availability of AI, it has not shortened workdays. If anything, it has extended them. Research published earlier this year by Harvard Business School https://hbr.org/2026/02/ai-doesnt-reduce-work-it-intensifies-it found that employees who end up working at a faster rate because of their AI usage simply end up working more hours and broadening the scope of their work rather than reducing their hours. A report from the AI Work Institute https://www.glean.com/work-ai-institute/reports/work-ai-index found that while some employees report saving as much as 11 hours per week thanks to AI, they spend about 6.5 of those gained hours maintaining their AI workflows and fixing mistakes. What seems more likely: that companies will use realized gains from AI deployment to lessen their employees’ workloads, or that they’ll pile more on their plates to fill the newly “free” hours? You know the answer. Don’t go planning that 3-day weekend vacation any time soon.