River AI raises $1.1 billion to build a personally owned AI stack

River AI, a startup founded two months ago by xAI co-founder Igor Babuschkin, has raised $1.1 billion in seed and Series A funding led by General Catalyst and AMP PBC, with participation from NVIDIA, AMD Ventures, Y Combinator, and Temasek.

What's new

Per TechCrunch's report: "River AI, an AI startup founded by xAI co-founder Igor Babuschkin, has secured $1.1 billion in funding in a seed/Series A round led by General Catalyst and AMP PBC, with participation from Nvidia, AMD Ventures, Y Combinator, and Temasek."

Key details:

• The round combines a seed and a Series A raised together, an unusually large amount for a company incorporated in April 2026.
• River AI's stated ambition is to rebuild the AI stack "end to end: training, models, the product layer, and new hardware," rather than build a thin wrapper on top of existing frontier-lab models.
• Its first product, the River API, lets developers and enterprises run reinforcement-learning training and LoRA fine-tuning on open-weight models, billed per million tokens.
• The company says its infrastructure can complete complex reinforcement-learning training runs in roughly 15 to 20 minutes without a dedicated infrastructure team, at costs it claims are two to four times lower than closed-source alternatives.
• Babuschkin previously worked on generative modeling and reinforcement learning at Google DeepMind, led large-scale training efforts at OpenAI, and co-founded xAI before starting River AI.

Context

River AI's pitch runs counter to the dominant enterprise AI pattern of the last few years, where companies mostly consume general-purpose models from a handful of frontier labs (OpenAI, Anthropic, Google, xAI) through an API. River is betting instead that enterprises increasingly want to own and customize their own models on their own data, using open-weight models as the base rather than a closed frontier model they don't control. That thesis has been gaining ground alongside the broader push toward open-weight releases from Meta, Mistral, DeepSeek, and others over the past year, and it puts River in more direct competition with infrastructure and fine-tuning players like Together AI and Fireworks than with OpenAI or Anthropic directly.

The raise also continues a trend of senior departures from the top frontier labs going on to start well-funded, well-connected new ventures almost immediately — Babuschkin left xAI to found River AI, and the new company drew investment from both NVIDIA and AMD simultaneously, an unusual pairing given the two chipmakers' rivalry.

Why it matters

$1.1 billion for a two-month-old company is an extreme valuation-to-age ratio even by current AI-funding standards, and it signals how much capital is chasing the idea that the next phase of enterprise AI adoption will be about ownership and customization rather than API access to someone else's model. If River AI's cost and speed claims for training and fine-tuning hold up, it could pressure existing fine-tuning and inference infrastructure providers on price. It's also a data point on founder mobility at the top of the industry: a co-founder can leave one of the best-funded AI labs in the world and immediately raise nine figures for an unproven, two-month-old company, purely on reputation and thesis.