The xAI founder exodus: why Musk's team keeps losing its key people

When Elon Musk assembled a team of twelve brilliant researchers and engineers in the summer of 2023 to found xAI, it looked like a serious bid to reshuffle the artificial intelligence market. Two and a half years later, little of that team remains: Toby Pohlen is the seventh co-founder to announce his departure.

Pohlen, one of xAI's key developers, said he is leaving the startup without giving details about his reasons or future plans. The context of his exit speaks for itself, though. The wave of departures among xAI's founders picked up markedly after Musk decided to merge the company with SpaceX, a move many in the industry viewed with mixed feelings. Combining an AI startup with an aerospace giant changed not only the corporate structure but the philosophy of a project that was originally positioned as an independent research organization.

To grasp the scale of the problem, it helps to recall how it all began. xAI was founded with an ambitious mission: to build artificial intelligence capable of "understanding the true nature of the Universe." Musk recruited specialists from DeepMind, Google Research, Microsoft Research and other leading labs worldwide. Each co-founder brought distinct expertise, from large language model architecture to AI safety systems. That team built Grok, the chatbot integrated into the X ecosystem (formerly Twitter), which became one of ChatGPT's more visible competitors.

Now the picture looks worrying. Seven of the roughly twelve original co-founders have left. This is not mere staff turnover; it is a systemic loss of institutional memory and strategic vision. In AI, where a company's success is determined above all by the quality of its research team, such a talent drain can have far-reaching consequences. By comparison, OpenAI has also endured painful departures of key figures, including Ilya Sutskever, but xAI's losses are proportionally far larger.

The SpaceX merger appears to have been a catalyst rather than the cause. Combining with an aerospace company inevitably shifts priorities: instead of fundamental AI research, applied tasks move to the front, such as optimizing rocket systems, processing Starlink satellite data and automating manufacturing. For scientists who joined xAI for pure science and ambitious research goals, that turn may have been a fundamental disappointment. Building AGI is one thing; optimizing the logistics of space launches is another.

There is also a more prosaic explanation. The market for AI talent is red-hot. Leading labs, including Anthropic, Google DeepMind and Meta AI*, along with dozens of generously funded startups, are ready to offer former xAI co-founders not only competitive compensation but also what many researchers value more than money: the freedom to set the direction of their work and access to cutting-edge compute. For specialists at this level, leaving xAI is not the end of a career but rather the start of a new, possibly more interesting, chapter.

For xAI itself, the situation is not necessarily catastrophic. The company has substantial computing power, including the Colossus supercluster, and continues to develop its Grok model line. SpaceX's resources and Musk's personal brand can still attract new talent. But in the race for artificial general intelligence, the winner is not whoever has more GPUs but whoever has the stronger team and the clearer vision. When more than half the founders vote with their feet, that is a signal that cannot be ignored.

The story of xAI increasingly resembles a pattern typical of many Musk projects: a grand start, recruitment of top minds, and then a collision between research ambitions and a rigid management hierarchy with unpredictable strategic turns. The question now is not whether xAI can survive these losses, but what kind of company it will become without the people who created it.

*Meta is designated an extremist organization and banned in the Russian Federation.