xAI co-founder raises $1.1B for River AI to build an open AI stack
Tech Funding News Abhinaya Prabhu ● Covered by 3 sources
xAI co-founder Igor Babuschkin raised $1.1 billion for River AI. The startup wants companies to own and run custom AI models, not just rent them.
Based on reporting by Tech Funding News, Abhinaya Prabhu — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
River AI has landed $1.1 billion only two months after coming out of stealth, a staggering haul for a startup this young. The Palo Alto company is betting that businesses want more than access to AI; they want something they can train, own and run themselves.
The round was led by General Catalyst and included AMP PBC, NVIDIA, AMD Ventures, Y Combinator and Temasek. River declined to give a post-money valuation, though earlier reporting had put talks near $5 billion before the round closed at its final size.
Igor Babuschkin, who left xAI in August 2025, incorporated River in Nevada by April 2026. By June, the company was out of stealth. His background runs through Google DeepMind, OpenAI and xAI, and his co-founders also come from xAI and Tesla.
River’s product sits below the model itself. It offers an API that runs LoRA fine-tuning and reinforcement learning on top of existing open-weight models, while handling the infrastructure that would otherwise need a dedicated team. The company says enterprises can finish a reinforcement-learning run in 15 to 20 minutes, and do it at two to four times lower cost than closed-source alternatives by paying per token rather than keeping GPUs idle.
This is not a bid to outbuild OpenAI or Anthropic on model size. It’s an attempt to own the training, tuning, deployment and, eventually, hardware layer around personal AI. River says the new money will go into training infrastructure, personalisation software and later dedicated hardware. That’s an ambitious stack, and one that may be easier to pitch than to sell at scale.
General Catalyst cast the bet as a matter of American resilience, while River argues there is a gap between what AI can do and what companies actually get to use. The market around open models is already getting crowded, with Together AI, Featherless.ai and Moonshot AI all pushing on adjacent pieces. River is trying to go broader than any of them. Whether firms want to own that much of their AI stack is the real test.
My take — AI-written commentary, not fact-checked reporting
The open-model crowd keeps saying the future is ownership, and sometimes that sounds like strategy and sometimes like a nicer word for vendor lock-in with better PR. River’s pitch is smarter than “we host models” because it aims at the unglamorous middle: training, tuning, deployment, the stuff enterprises actually pay for. But the line between “your AI” and “our platform” gets blurry fast, which is exactly where these grand open stacks usually get very serious and very expensive.
Read more about this at: Tech Funding News