Models
River AI's $1.1 Billion Round Bets Enterprises Will Want Agents They Can Train
River AI, founded by xAI co-founder Igor Babuschkin, has raised $1.1 billion to build a stack around personally trainable agents and open models, bringing a large capital bet to the question of who should own AI behavior.
By Elvin C ยท
River AI, a startup founded by xAI co-founder Igor Babuschkin, has raised $1.1 billion in a seed and Series A round led by General Catalyst and AMP PBC, with participation from Nvidia, AMD Ventures, Y Combinator and Temasek. The scale of the financing is striking for a company that emerged from stealth only in June.
River's argument is that agents should be personally trainable rather than merely prompted. Its initial API lets developers use reinforcement learning and LoRA fine-tuning with open models, positioning the company as infrastructure for organizations that want to shape model behavior around their own data, preferences and workflows.
That pitch responds to a real enterprise tension. Hosted frontier models are powerful and convenient, but companies do not always want their most important workflows to depend on a model they cannot inspect or improve. Open weights and post-training offer more control, although they also move more responsibility for safety, cost and operations onto the buyer.
The funding provides time to build, but it also sets a high bar. River must show that fast fine-tuning produces reliable business value and that its underlying infrastructure can compete with cloud providers, model hosts and a growing group of AI platform companies.
The lasting idea is larger than one startup: the next AI market may divide less cleanly between open and closed models than between systems people can meaningfully shape and systems they simply rent.
Topics: River AI, open models, AI funding