Igor Babuschkin exits xAI to launch Babuschkin Ventures for AI-safety
Former DeepMind/OpenAI engineer led xAI’s model-training and infrastructure efforts
Venture will fund startups and research focused on safe, agentic AI systems
Departure highlights talent churn at xAI amid intense AI hiring competition
Igor Babuschkin, a co-founder of Elon Musk’s AI startup xAI, said Wednesday he has left the company and will launch Babuschkin Ventures, an investment vehicle to back AI-safety research and startups working on safe, trustworthy systems. Babuschkin announced the move on X and in subsequent media posts.
Babuschkin, who earlier held roles at DeepMind and OpenAI, was instrumental at xAI in building core infrastructure and tooling for large-scale model training, later overseeing engineering across infrastructure, product and applied AI projects. In his posts he described helping tackle urgent technical problems during xAI’s early push to launch models and data-centre efforts.
His exit follows recent departures from the broader Musk ecosystem, including xAI’s head of legal and X’s former CEO, and arrives as rivals such as OpenAI, Google and Anthropic double down on AI research and talent. Observers say the move highlights both the intense hiring and retention pressures in the race to build next-generation AI, and a growing ecosystem for safety-focused capital.
Babuschkin’s Next Act
Babuschkin said Babuschkin Ventures will prioritise backing teams and projects focused on AI safety, agentic systems and research that aims to make advanced AI systems safer and more robust. The new fund signals a common pattern among senior AI engineers shifting from product roles into mission-driven investment and research support.
Industry watchers will look for details on Babuschkin’s investment focus, initial partners and portfolio companies, and whether his move spurs similar exits of senior engineers into safety-oriented ventures. xAI has not publicly detailed a replacement; Babuschkin said he left to “become a founder again.”
