River AI Raises $1.1B in Massive General Catalyst-Led Round

Two months after emerging publicly, River AI has already joined the billion-dollar funding club.

The AI startup founded by former xAI co-founder Igor Babuschkin has raised $1.1 billion across its seed and Series A rounds, giving the young company an unusually large war chest to pursue a different vision of artificial intelligence: models that companies and eventually individuals can train, customize and own rather than simply rent through an API.

General Catalyst and newly formed AI investment firm AMP PBC led the financing. Nvidia and AMD Ventures joined as strategic investors, while Y Combinator and Singapore investment company Temasek also participated. River did not disclose the valuation attached to the completed financing.

Earlier reports had suggested River was seeking as much as $1 billion at a valuation of up to $5 billion, with Babuschkin potentially contributing as much as $100 million himself. Those figures were discussed while the round was still being negotiated in May, and River has not confirmed that the final valuation reached that level.

A Billion-Dollar Bet on Owning AI Models

River is taking aim at one of the dominant business models in generative AI.

Companies today commonly send data and prompts to general-purpose models offered by companies such as OpenAI, Anthropic and other model providers. River argues that larger businesses will increasingly want models built around their own data, workflows and requirements instead.

Its first product, the River API, is designed to let developers fine-tune open-weight models and run reinforcement-learning jobs without building the underlying training infrastructure themselves. River says enterprises can complete complex reinforcement-learning runs in 15 to 20 minutes without a dedicated infrastructure team and claims its system can cost two to four times less than closed-source alternatives. Those performance and cost comparisons are River's own claims and have not been independently verified.

The API is currently labeled a v0.1 preview and supports LoRA fine-tuning and reinforcement learning across open models ranging from roughly 35 billion to 1 trillion parameters. River meters both training and inference by tokens rather than charging customers directly for reserved GPU hours.

For one of its published reinforcement-learning examples, River says a training run using roughly 500 million completion tokens and 250 million training tokens can be completed for under $1,000.

That offers a clearer picture of River's immediate business than the company's much broader long-term ambitions. For now, it is selling infrastructure for organizations that want more control over open models.

Babuschkin Is Building Beyond Another AI API

The size of the round is also closely tied to who is behind River.

Babuschkin previously worked on generative modeling and reinforcement learning at Google DeepMind, contributed to large-scale model training at OpenAI and later co-founded xAI with Elon Musk. River says its founding team also includes people with experience at xAI and Tesla.

River was publicly introduced on June 10, when Babuschkin laid out a much broader plan than building a model-training service. The company eventually wants to create AI systems that continually learn from individual users and adapt their underlying model weights to a person's preferences, goals and working style.

The roadmap could eventually extend all the way to hardware. River says reaching its vision will require rebuilding multiple layers of the AI stack, including training infrastructure, models, products and new hardware that could allow personal AI systems to operate closer to users rather than entirely inside remote data centers.

That is a far bigger undertaking than its current API product, and the $1.1 billion financing gives River the capital to attempt it.

Silicon Valley Is Funding AI Labs Earlier Than Ever

River's financing also highlights how dramatically traditional startup funding rules have changed during the AI boom.

For decades, a billion-dollar financing would typically have been associated with a mature private company that already had substantial customers, revenue and several previous funding rounds. Frontier AI has compressed that cycle.

Earlier in 2026, former DeepMind researcher David Silver's Ineffable Intelligence raised $1.1 billion at a reported $5.1 billion valuation. Richard Socher's Recursive Intelligence had also raised $650 million at a $4.65 billion valuation, according to Forbes.

These companies are part of a growing category sometimes described as AI “neolabs”: research-heavy startups built around prominent AI scientists and engineers that can attract enormous amounts of capital before developing businesses of comparable scale.

River is slightly different because it already has a developer product available, but the same investor logic is visible. The bet is being made not only on current revenue, but on the possibility that a small group of experienced AI researchers could build infrastructure or models capable of becoming foundational platforms.

River Is Betting Open Models Will Win More Enterprise Work

The larger strategic wager behind River is that open-weight AI will continue closing the gap with proprietary frontier models.

Babuschkin argues that companies will become increasingly reluctant to hand their most valuable internal data and intelligence workflows entirely to external model providers. Instead, River expects businesses to mix models and customize open systems around their own operations.

General Catalyst is making essentially the same bet. CEO Hemant Taneja said maintaining U.S. leadership in open-weight models is important alongside continued leadership in closed frontier systems, while River's software is intended to make that model ownership practical for enterprises without massive AI infrastructure teams.

The $1.1 billion round gives River enormous resources for a company that was barely visible a few months ago. It does not guarantee that personalized, user-owned AI will replace today's API-driven model economy.

But it does show how valuable investors believe that possibility could become.

River is no longer being funded like a conventional two-month-old startup. It is being financed as though the next major AI platform could be built before the market has even decided exactly what that platform should look like.