$1.1B Before the First Birthday

Two months is not a long runway — but it was long enough for River AI to close one of the most striking debut financings in recent memory. Founded by former xAI co-founder Igor Babuschkin, River has raised $1.1 billion across its seed and Series A rounds to pursue an ambitious idea: AI that belongs to the people and companies using it, rather than the labs that created it.

The funding round was led by General Catalyst and AMP PBC, with participation from NVIDIA, AMD Ventures, Y Combinator, and Temasek.

The company declined to disclose its post-money valuation, though an earlier Forbes report had pegged the pre-close target at up to $5 billion.

Who Is Igor Babuschkin?

The founder's résumé is doing a lot of work here — and investors clearly think it earns the price of admission.

Babuschkin is a co-founder of xAI who previously worked on generative modeling and reinforcement learning at Google DeepMind and led large-scale training efforts at OpenAI.

He departed xAI in August 2025, launched an entity called Babuschkin Ventures, and within months had assembled a team and a corporate structure for River AI.

The founding team also includes engineers with experience at xAI and Tesla across deep learning, reinforcement learning, and AI infrastructure.

"The way AI is built today is not how it will be built in the future. 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. We started River to allow people and companies to own their intelligence."

What River Is Actually Building

River is not a model lab in the traditional sense — it's positioning itself as the infrastructure layer for the open-weight era.

The platform delivers LoRA fine-tuning and reinforcement learning for frontier open-weight models via an API, allowing developers and enterprises to train, tune, deploy, and serve custom models without dedicated infrastructure teams or expensive hardware.

The performance claims are aggressive. The company says enterprises can complete complex reinforcement-learning training runs in as little as 15 to 20 minutes without maintaining an infrastructure team, while achieving two to four times the cost savings of closed-source alternatives.

River's core argument is that general-purpose models can be powerful but are rarely tailored to the specific data and workflows of individual organizations — and that until now, creating a custom model typically required specialized hardware, an infrastructure team, and months of development.

Longer term, the ambition scales well beyond an API. The company is also developing hardware and consumer-facing products alongside its training infrastructure, with a long-term strategy to build a vertically integrated stack in which personal AI can operate close to users, continually learn from them, and remain under their control.

Why General Catalyst Wrote the Check

General Catalyst CEO Hemant Taneja framed the bet in explicitly geopolitical terms:

"American leadership in AI urgently requires leadership in open-weight models, while maintaining a lead in closed frontier models. Igor and the River AI team have the experience to make this happen, and we view their agenda as a priority for American resilience."

General Catalyst said the growth of open-weight AI models is strategically important alongside continued development of closed frontier models, and sees River's approach as addressing the gap between the capabilities of advanced AI and the customized systems most enterprises are currently able to deploy.

The strategic chip investors — NVIDIA and AMD Ventures — signal something else: whoever wins the custom model training market will be a significant compute customer. Their presence is less about financial return on this round and more about securing a seat at the table with whoever becomes the dominant fine-tuning layer.

What This Means for Founders and Builders

River's pitch maps directly onto a frustration that most enterprise software teams know well. The startup is betting that the next wave of enterprise AI adoption will not be driven by one-size-fits-all models from major labs, but by businesses that want to train and own models tailored to their proprietary data.

For startup founders, that shift has real product implications. If fine-tuning a frontier open-weight model truly becomes a 15-minute API call rather than a months-long infrastructure project, the barrier to shipping differentiated AI features collapses — but so does one of the moats that well-resourced incumbents currently enjoy. Companies that have quietly built defensibility by owning the fine-tuning process will need to find their edge elsewhere.

This is part of a broader pattern we've tracked this month: capital is flowing aggressively toward the picks-and-shovels layer of AI, whether that's Salesforce's $50M bet on ISV distribution via AgentExchange or River's $1.1B wager that the future of enterprise AI is open, owned, and continuously trained. River's round sits at the extreme end of that trend — but the direction of travel is consistent.