The xAI co-founder who helped raise Grok from nothing and built the Memphis supercluster in 122 days has launched River AI, a startup seeking up to $1 billion at a $5 billion valuation to put personalized artificial intelligence under individual ownership rather than corporate control.
When one of the architects of your competitor's AI stack turns around and starts a company explicitly designed to undercut your model, that is probably the most credible signal the market can send. Igor Babuschkin spent two years at xAI building the infrastructure that powers Grok, overseeing engineering across infrastructure, product, and applied AI, and personally leading the effort to construct the Memphis supercomputer cluster at a pace that stunned the industry. Now he has walked away from that and incorporated River AI in Nevada on April 20, 2026, with a pitch that is essentially the photo negative of everything centralized AI companies are selling.
River AI's stated mission is to deliver intelligence that users fully own: the hardware it runs on, the data it learns from, and the AI itself. That framing puts it squarely in opposition not just to xAI but to every major model provider, including OpenAI, Google, and Anthropic, whose revenue models depend entirely on users renting access to capabilities rather than owning them. Babuschkin is betting that a large enough segment of the market is uncomfortable with that arrangement and is willing to pay, or at least invest, in an alternative.
The numbers being floated are striking for a company that has not yet shipped a single product. Babuschkin is seeking up to $1 billion in initial funding at a valuation of up to $5 billion, with General Catalyst in active talks to lead the round. He is also committing up to $100 million of his own capital, a figure that signals genuine conviction and gives institutional investors something concrete to anchor their own confidence around. In the current AI funding environment, founder skin-in-the-game has become a meaningful differentiator as VCs try to separate genuine builders from credential-collectors raising on reputation alone.
The $5 billion pre-product valuation is audacious, but it fits a pattern that has become routine in this cycle. Pedigree is being priced like a moat right now, and Babuschkin's resume is genuinely unusual. He started as a particle physics researcher at CERN, moved into AI research at Google DeepMind where he was part of the team that built AlphaStar, the system that defeated top-ranked human players at StarCraft in 2019, then joined OpenAI before helping Elon Musk stand up xAI from scratch in 2023. The arc runs through every major inflection point in modern AI development. If any individual's technical judgment carries a premium, his does.
General Catalyst's reported interest deserves its own attention. The firm has backed AI infrastructure plays before, but positioning itself as a lead investor in a user-sovereignty startup would mark a meaningful shift in how it is narrating its AI thesis. The firm has been deliberate about associating itself with what it calls resilience-oriented investments. River AI, framed as a hedge against corporate AI concentration, maps neatly onto that story, especially at a moment when regulatory scrutiny of large AI providers is increasing in both the US and Europe.
What Babuschkin is actually building
The honest answer, as of now, is that specifics are thin. River AI's website promises to create personal AI that is owned and shaped by you, describing a new personal stack that works entirely for you, shares your values, and operates on your terms. Beyond that, there is a careers page, a note that the team is assembling top researchers and engineers, and a line promising that first offerings arrive soon. No product architecture, no training approach, no hardware roadmap. As Forbes recently pointed out, the funding talks surfaced before the company had even formally launched, and neither the round's closing nor a final investor list has been confirmed.
The unanswered question is whether the economics can work. Frontier-quality AI currently demands exactly the kind of capital concentration Babuschkin pioneered in Memphis, and reconciling that with hardware and data that individuals actually own is the hard engineering and business problem at the center of this bet. Watch two things in the coming months: whether General Catalyst formally closes the round, and what River AI's first technical disclosures reveal about how user-owned intelligence would actually run. If Babuschkin can show even a credible prototype, the rented-access model underpinning the entire industry gets its first serious counter-narrative from someone who knows precisely how the incumbents are built.
Also read: Oracle raised $48 billion this fiscal year and plans to raise $40 billion more, and the market still sent the stock down 7% • Ramp's June AI Index reveals a 680-fold spending gap between AI leaders and everyone else • Amazon's $17.5 billion no-covenant loan tells you everything about how Wall Street is betting on AI