River AI Raises $1.1B to Build Custom Enterprise AI Models on Proprietary Data

River AI, a startup founded by xAI co-founder Igor Babuschkin, has raised $1.1 billion to expand products that let enterprises train AI models on their own proprietary data.

The round was co-led by General Catalyst and AMP PBC, with strategic participation from Nvidia, AMD Ventures, Y Combinator, and Temasek. The company did not disclose a valuation.

Why enterprises want private models

River AI’s thesis is that large companies will move away from off-the-shelf offerings from major AI labs toward privately owned, open-weight models tailored to each organization’s needs and data.

According to the company, its API can complete reinforcement-learning training runs in as little as 15 to 20 minutes, without an in-house infrastructure team, at a cost it claims is two to four times lower than closed-source rivals.

“AI should be open, freely available, and affordable. It should feel like it is working for the person using it, not the lab that trained it.” Igor Babuschkin, founder, River AI

Founder pedigree

Before River AI, Babuschkin worked on generative modeling and reinforcement learning at Google DeepMind, led large-scale training efforts at OpenAI, and co-founded xAI.

The funding arrives amid a broader rush of capital into applied enterprise AI tooling, as companies look for ways to keep sensitive data in-house while still matching frontier-model performance.

Decoded Take

River AI is pitching a structural alternative to renting intelligence from closed labs: train on private data, own the weights, and keep costs closer to infrastructure than to API markups. Chip-maker participation from Nvidia and AMD underscores how tightly this thesis is tied to compute economics. If the speed and cost claims hold at scale, more enterprises will treat custom models as a default architecture choice. If they do not, River becomes another well-funded middle layer between hyperscalers and corporate IT.