Igor Babuschkin left Elon Musk’s xAI to build a company of his own. On Monday, investors lined up to pay for it.
River AI, the startup founded by the xAI cofounder and former DeepMind researcher, said it had raised $1.1 billion in a round led by General Catalyst and AMP PBC. Nvidia and AMD participated, and Y Combinator and Singapore’s Temasek joined as strategic investors, according to a statement. The round ranks among the larger private AI financings of the past year and lands River AI in the top tier of young labs chasing the frontier-model business.
Babuschkin is one of the more closely watched figures in the field. He spent years at DeepMind, where his work included game-playing systems trained at frontier scale, before joining Musk to cofound xAI in 2023. At xAI he helped build the Grok chatbot that anchors the company’s consumer push. His departure was among the more consequential exits in a sector where talent is the scarce asset, and investors have been tracking what he built next. Monday’s round answers that question with a check.
The company has said little about its technology roadmap, staying quiet even by the standards of a field that prizes secrecy. What is clear from the investor list is where the market expects value. The presence of Nvidia and AMD is the most telling detail: the two chipmakers have made a habit of backing model developers whose demand for processors grows with their ambitions. A $1.1 billion round with both on the cap table suggests River AI plans to train models at a scale that requires serious computing, and the chipmakers want a seat at the table when it does.
Temasek’s participation adds a different kind of signal. The Singapore sovereign fund has been building a position in frontier AI across geographies, and its presence gives River AI a financial backer with a long horizon and government-adjacent credibility. Y Combinator, where Babuschkin’s ties to the startup ecosystem run deep, rounds out a group that spans venture capital, semiconductors and state capital — an unusually broad coalition for a company that has yet to describe its product.
The structure of the round matters as much as the size. Leading alongside a fund like AMP PBC, a public-benefit company that invests with a social mandate, suggests River AI is positioning itself to talk about safety and purpose as well as performance. The pairing of General Catalyst, one of the most aggressive growth-stage investors in AI, with a mission-oriented partner lets the company argue that it can move fast without abandoning the field’s preferred vocabulary of responsibility.
For Nvidia, the investment is one more thread in a strategy of attaching itself to every significant model developer. The chipmaker’s willingness to put capital into potential rivals’ customers has been a defining feature of the AI boom, and its presence in River AI’s round extends that pattern. AMD’s participation is a sharper statement: the company has been pressing to convert its data-center market share into an ecosystem, and backing developers early is one way to build one.
A round of this size buys three things: compute, people and time. Frontier-model training demands clusters that cost hundreds of millions of dollars, and the best researchers command packages that rival public-company compensation. The money also buys patience — the months of quiet engineering that separate a funded lab from a shipping product. River AI now has all three, which is more than most startups can say.
The round also lands at a delicate moment for the field. Capital is plentiful, but expectations have been reset by a year in which several heavily funded labs shipped less than they promised, and investors have grown pickier about whom they back. A founder with Babuschkin’s record clears that bar more easily than most, but the money comes with a clock attached: River AI will be expected to show a credible path to a frontier-class model within a defined window, and the investors who wrote these checks will be watching for it.
The round also says something about the funding climate. After a stretch in which AI capital concentrated in a small group of established labs, money is flowing again to new entrants with proven founders. Babuschkin’s track record at DeepMind and xAI is the collateral behind the check: investors are betting that the person who helped build Grok can build a frontier lab with a different culture and a clearer mandate.
For Babuschkin, the money is a second act on his own terms. He left xAI at a moment when its valuation had soared alongside Musk’s public profile, and he now runs a company whose shareholder list reads like a map of the industry’s centers of gravity. The work of justifying the round comes next, and it will be measured in the one thing the statement did not mention: results. River AI has its $1.1 billion, and the field will now watch what it does with it.
