Babuschkin's River AI closes $1.1B at ~$5B for user-owned AI

Another senior name from Elon Musk's xAI has walked out the door and immediately raised nine figures on the way. Igor Babuschkin, an xAI co-founder who spent time at DeepMind and OpenAI before that, has closed $1.1 billion for a new startup called River AI at roughly a $5 billion valuation, The New York Times reported. General Catalyst and AMP PBC led the round, with NVIDIA and AMD Ventures, Y Combinator and Temasek among the participants. Babuschkin himself is putting up to $100 million of his own money in.

The pitch is a swipe at the frontier-lab model. Rather than one giant hosted model that everyone rents access to, River wants to sell personalized AI that users own and continually retrain on their own data, workflows and preferences. The team includes former xAI and Tesla staffers, and Babuschkin's stated long-term plan is a vertically integrated stack that eventually reaches into hardware and physical infrastructure, so personal AI can operate close to the user.

That is a lot of capital for a newly launched company, and River is walking into the same expensive fight that OpenAI, Anthropic and Babuschkin's own former employer are already spending tens of billions on. Applied Compute is separately in talks for a $3 billion round led by Elad Gil, so late-stage AI checks are still landing at scale. Neither the round announcement nor the reporting names a first paying customer, a launch date, or how much of the $1.1 billion is earmarked for chips versus consumer hardware, and the 'you fully own it' framing is easier to say than to price when frontier training economics still favor centralized providers. This is the sort of round where investors are backing the founder and the network first and hoping the product story catches up.

For General Catalyst and AMP PBC, it locks in exposure to one of the more decorated researchers to leave a frontier lab this cycle. For NVIDIA and AMD, it is another well-funded buyer at the top of the chip stack. And for developers looking for something other than a closed API from a handful of hyperscalers, River joins a busy tape of AI funding rounds pitching that alternative.