There’s a particular kind of chaos that happens when a billionaire gets frustrated with something. At xAI, that chaos looks like constant layoffs, cofounders being shown the door, and SpaceX “fixers” parachuting in to tell people their work isn’t good enough. Staff are saying the upheaval is destroying morale, and honestly, it’s not hard to see why.
Elon Musk ordered another round of job cuts at xAI this week after growing impatient with the startup’s coding product. The problem is straightforward enough: Anthropic’s Claude Code and OpenAI’s Codex have actually resonated with users. xAI’s offerings haven’t. So instead of letting the company iterate and learn, Musk decided to blow it up and rebuild it from scratch.
When Ambition Meets Reality
The technology world loves a good comeback story, but they’re far rarer than people think. What usually happens instead is what we’re seeing at xAI: talented people get burned out, leave for competitors offering better terms, and the organization becomes increasingly unstable.
Two cofounders departed this week alone. Zihang Dai, a senior technical member who had actually been honest about xAI’s lag in coding capabilities, is gone. Guodong Zhang, who ran pre-training for the Grok models, left after being blamed for the coding product’s failures and stripped of his responsibilities. That leaves only two of the original 11 cofounders still standing.
Managers from SpaceX and Tesla have been sent in to audit employees and fire people deemed inadequate. This is the “extremely hardcore” Musk management style in full effect. But there’s a difference between pushing people hard and systematically dismantling the organization you built.
The Data Center and the Dream
Musk merged SpaceX with xAI in a $1.25 billion deal and is now racing toward a June deadline for what could be history’s biggest IPO. He’s got massive ambitions: launching AI data centers into space, building factories on the Moon, colonizing Mars. The Memphis data center alone has over 200,000 specialized AI chips with plans to eventually reach 1 million GPUs.
That’s genuinely impressive infrastructure. But infrastructure alone doesn’t build great products. People do. And when those people are terrified of getting fired for not meeting impossible timelines, they tend to leave.
Researchers have been quietly quitting because of burnout or better offers from rivals. Musk even admitted on X that xAI had incorrectly rejected promising candidates in previous hiring rounds. He’s now reaching back out to offer them jobs on better financial terms. It’s a messy acknowledgment that something went wrong.
The Talent Grab
This week xAI poached Andrew Milich and Jason Ginsberg from Cursor, a popular AI coding app, to boost the “Grok Code Fast” product. Musk welcomed them with posts about orbital space centers and mass drivers on the Moon. Classic move: sell the dream when the reality is unstable.
The irony is thick here. Musk has the ability and resources to recruit top talent. The problem isn’t whether he can poach people from other companies. It’s whether those people will actually stay long enough to build something lasting when they see how quickly things can fall apart.
Business moves this aggressive often work in the short term. You cut costs, fire underperformers, bring in structure. But there’s usually a cost nobody wants to talk about: the people you didn’t need to lose, the institutional knowledge that walks out the door, the culture that never quite recovers.
Musk keeps saying xAI “was not built right first time around, so is being rebuilt from the foundations up.” He compares it to what happened with Tesla. But here’s the thing about that comparison: Tesla actually survived its chaos. Most startups don’t. Most people don’t stick around to watch their company get dismantled, no matter how compelling the vision of Mars colonization sounds.
At some point, even the best talent asks whether working 80-hour weeks for someone who fires your colleagues on a whim is actually worth the equity and the promises about the future.
