OpenAI poaches top talent from Tesla and xAI

The AI talent war just took its most personal turn yet.

OpenAI has poached four high-profile engineers from its fiercest rivals to bulk up its scaling team, according to an internal Slack message sent Tuesday by cofounder Greg Brockman, first reported by WIRED.

Raiding Musk's supercomputer bench

The list of recruits reads like a direct hit on Elon Musk's engineering core. OpenAI brought on David Lau, former vice president of software engineering at Tesla, alongside Uday Ruddarraju, former head of infrastructure engineering at xAI and X, and Mike Dalton, an infrastructure engineer from xAI. Rounding out the group is Meta AI researcher Angela Fan.

What makes the move especially stinging is that both Ruddarraju and Dalton, who previously worked together at Robinhood, were instrumental in building xAI's Colossus—a supercomputer powered by more than 200,000 GPUs.

At OpenAI, the recruits will join the scaling team led by Brockman, working on the backend hardware, data centers, and the ambitious Stargate project that underpins future frontier models. Ruddarraju called Stargate an "infrastructure moonshot" in a statement to WIRED, while Lau pointed to safe AGI as the most rewarding mission for his next chapter.

Escalating tit-for-tat poaching

OpenAI's offensive comes as top labs pull out all the stops to recruit and retain talent. Meta CEO Mark Zuckerberg has been aggressively poaching, luring at least seven employees away from OpenAI with massive pay packages and vast compute allocations. Zuckerberg has also reportedly targeted staff at Thinking Machines Lab, the new startup from former OpenAI executives Mira Murati and John Schulman.

That pressure recently pushed OpenAI CEO Sam Altman to tell staff that the company plans to recalibrate researcher compensation to stay competitive.

Taking key figures from Tesla and xAI also pours gasoline on an already bitter rivalry. Musk and OpenAI are currently suing each other over the company's shift toward a for-profit structure, and stripping xAI of its core infrastructure builders will only raise tensions.

Why it matters

Frontier AI is no longer just a research challenge; it is a massive distributed computing bottleneck. Very few people in the world have hands-on experience orchestrating clusters of 200,000 GPUs without hardware meltdowns and network bottlenecks.

In our view, OpenAI made a brilliant, calculated move. By poaching the exact engineers who wired xAI's Colossus, OpenAI instantly accelerates its own Stargate infrastructure while slowing down Musk's momentum. But the talent arms race is reaching fever pitch: when even OpenAI has to hike its compensation packages to fend off Meta and poach from rivals, the engineering cost of frontier AI is climbing to heights only a handful of trillion-dollar balance sheets can stomach.