Sam Altman told developers Monday that OpenAI plans to significantly reduce its hiring pace because artificial intelligence now handles work that once required human employees. The announcement came during a livestreamed builder town hall that OpenAI billed as an experiment in developer engagement.
The hiring freeze that isn't
Altman was careful with his framing. OpenAI isn't freezing hiring, he said, just growing more slowly.
The distinction matters when you're burning through roughly $17 billion annually. OpenAI's compute footprint exploded from 0.2 gigawatts in 2023 to 1.9 gigawatts last year, according to Benzinga's reporting, and that infrastructure doesn't maintain itself. But Altman suggested the company's AI tools have changed the math on headcount.
He warned other tech companies to avoid hiring aggressively only to realize later that AI can do the work. OpenAI's approach, as he described it, is to grow headcount slowly rather than face "very uncomfortable conversations" about cuts down the road.
What builders actually wanted to know
The town hall format, which Altman called "a first pass" at a new developer communication channel, surfaced tensions that have been simmering since late 2025. Some developers pushed back on OpenAI's model deprecation practices. One pointed out that OpenAI had said it had "no plans to sunset GPT-4o" just weeks before announcing exactly that.
Questions about autonomous agents dominated. Developers want clarity on how to build reliable agentic workflows when the underlying models keep shifting. The trust issue here is real: if you're building a business on OpenAI's API, you need some confidence that your foundation won't disappear.
Altman didn't offer binding commitments. He rarely does.
The memory play
Altman has been on a consistent message lately: the next AI breakthrough won't come from smarter reasoning but from better memory. He's described current memory capabilities as being in "the GPT-2 era," which is OpenAI-speak for primitive but foundational.
The vision is an AI that remembers your entire interaction history, your preferences, your quirks. As Altman put it on a recent Big Technology Podcast appearance, no human assistant can remember every word you've said or every document you've written. AI could.
This is partly about product stickiness. Once your AI has years of context about you, switching providers means starting over. It's also a deflection from the increasingly crowded reasoning benchmark race, where Google's Gemini 3 has been making OpenAI nervous enough to trigger internal "code red" alerts.
The uncomfortable subtext
The timing of Altman's hiring comments lands awkwardly. The U.S. job market has been stuck in what some economists call "the Great Freeze," with unemployment for 20-to-24-year-olds hitting 9.2% in recent months. Young workers are already struggling, and here's the CEO of the world's most prominent AI company essentially saying his technology will let companies do more with fewer people.
Altman framed this as responsible management: plan for AI's impact now rather than scramble later. But the developers in that town hall are building the tools that will enable exactly this kind of workforce reduction across industries. That's the bargain they've made, whether or not Altman says it out loud.
OpenAI reportedly aims to raise $100 billion at a valuation around $830 billion. When Altman was asked about an IPO timeline, he made clear he's in no rush. Being a public company CEO doesn't appeal to him, he said, though he acknowledged OpenAI will probably go public eventually.
For now, the company keeps burning cash, building models, and apparently, hiring more slowly than before.
