River AI, founded by Igor Babuschkin, raised 1.1 billion dollars in a Series A led by General Catalyst
The startup River AI, founded by Igor Babuschkin, the founder of xAI, raised 1.1 billion dollars in a round led by General Catalyst and AMP PBC, with participation from Nvidia, AMD Ventures, Y Combinator and Temasek. It offers an API for RL and LoRA fine-tuning of open models.
The startup River AI, founded by Igor Babuschkin, a co-founder of xAI (previously also at DeepMind and OpenAI), raised 1.1 billion dollars in a seed/Series A round. The round was led by General Catalyst and AMP PBC, with Nvidia, AMD Ventures, Y Combinator and Temasek also participating. AMP PBC is an AI-focused investment firm founded in 2026 by Anjney Midha, a former general partner at Andreessen Horowitz. River AI emerged from stealth mode in June 2026, so the funding came approximately two months after its founding.
River AI describes its goal as rebuilding the AI technology stack from the ground up — training, models, the product layer and new hardware intended to enable personal AI to run close to the user. According to Babuschkin, the resulting agents should function more like “guardian angels”, meaning tools serving a specific person, in contrast to the direction other AI labs are taking toward replacing human workers.
The first product from the company is an API billed according to the number of tokens processed, with the rate depending on the open model used. The API allows developers to apply both reinforcement learning and LoRA fine-tuning to models, which the company says is intended as an alternative to prompting models that users do not own and cannot improve.
According to an announcement from River AI itself, any enterprise can complete a complex reinforcement learning run in 15 to 20 minutes without needing an in-house infrastructure team, with a two- to fourfold reduction in costs compared with closed-source alternatives. These figures come from the press announcement issued by the company and are not independently verified in the source.
Why it matters
A new offering is emerging for developers and companies working with open models, intended to replace manual prompt engineering with training their own closed version of a model through an API — if the claims from the company about speed and pricing are confirmed, this could potentially lower the barrier to post-training without an in-house infrastructure team. The amount of funding also signals that investors including Nvidia and AMD Ventures are betting on the segment of personalized AI agents and open models as an alternative to large closed labs.
Two audiences, two different impacts
What this means
For individuals
Developers working with open models are being offered another option for training their own version of a model instead of relying on prompt engineering, but for now it is a newly launched product without independently verified results.
For a business
A new player is entering the market, offering enterprise customers reinforcement learning and LoRA fine-tuning of open models as a service without the need for an in-house infrastructure team, which could influence how companies decide between building their own post-training capabilities and using an external provider.
DevelopmentCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.