River AI, founded by xAI co-founder Igor Babuschkin, raised $1.1 billion across seed and Series A rounds led by General Catalyst and AMP PBC, with strategic investment from NVIDIA and AMD Ventures, plus Y Combinator and Temasek. Founded earlier this year and emerging from stealth just two months prior, the Palo Alto startup aims to let enterprises train, tune, and deploy custom open-weight AI models without dedicated ML infrastructure teams or specialized hardware.
River's core product is an API enabling reinforcement learning and LoRA fine-tuning on open-weight models ranging from 35 billion to 1 trillion parameters. The company claims enterprises can complete complex reinforcement learning runs in 15 to 20 minutes at two to four times lower cost than closed-source alternatives, with token-metered billing and zero idle compute charges. The platform abstracts infrastructure complexity—fast weight transfers, sampling consistency, elastic compute—so enterprises can focus on model improvement rather than DevOps.
Babuschkin's pedigree matters: he led large-scale training at OpenAI, worked on generative modeling and RL at Google DeepMind, then co-founded xAI. The investor slate—General Catalyst framing open models as a 'national priority', NVIDIA and AMD backing the chip economics—signals this is not just a software bet but an infrastructure play on who owns the training toolchain as enterprises shift from off-the-shelf models to in-house open-weight personalization.