A 2-Month-Old Startup Just Raised $1.1 Billion to Rebuild AI

A company that’s only two months old just raised $1.1 billion. That’s River AI, founded by xAI co-founder Igor Babuschkin, who launched it in June 2026. Barely two months later, General Catalyst and AMP PBC co-led a massive funding round, with Nvidia, AMD Ventures, Y Combinator, and Temasek also participating.

Not another chatbot wrapper

Babuschkin’s background is in core AI research, having previously worked at both DeepMind and OpenAI. River AI’s stated goal is to “rebuild the AI stack” — training, the models themselves, the product layer, and hardware, redone from the ground up. That’s a fundamentally different bet than most AI startups today, which wrap an interface around an existing model’s API; River AI is positioning itself as heavy infrastructure investment starting from the foundation up.

The goal isn’t replacing workers — it’s “guardian angels”

Babuschkin’s framing of the product vision is worth sitting with: he’s described capable AI agents becoming a normal part of daily life, but not as replacements for human workers — more like guardian angels, quietly present and on the user’s side. That’s a notable contrast to the “automate away the headcount” pitch a lot of AI startups lean on; River AI is aiming at personalized, trainable assistants rather than enterprise process-automation tools.

There’s already a product, not just a vision statement

River AI isn’t running on a pitch deck alone. It already has an API live, billed per million tokens, supporting reinforcement learning and LoRA (low-rank adaptation) fine-tuning. The company claims complex reinforcement-learning computations can finish in 15 to 20 minutes, at 2 to 4 times lower cost than closed-source alternatives. That a two-month-old company can put out specifics like that — real technical specs and real pricing — says something about what this $1.1 billion is actually betting on: not just Babuschkin’s résumé, but a product that already has something to point to.