xAI Co-Founder’s River AI Raises $1.1B To Bring AI Back Home

River AI has raised $1.1 billion, and the story is bigger than another large AI funding round. The company is trying to build AI that can be owned, customized and run closer to the user, instead of living entirely inside the giant cloud platforms.

The startup was co-founded by Igor Babuschkin, a former xAI co-founder and engineer who previously worked on frontier AI systems. River AI raised the new funding led by General Catalyst and AMP, with strategic investment from Nvidia and AMD. Reuters also reported the round as $1.1 billion for custom AI tools.

River’s own public message says the company wants to build AI that is owned and shaped by users. Its first product is the River API, which lets people build custom agents and large language model workflows. The New York Times framing highlighted another important part of the plan: computers for homes and small businesses that can run AI locally.

That local-AI angle is what makes this worth paying attention to. Most of today’s AI experience depends on cloud data centres. You type into a chatbot, the request travels to remote servers, and the answer comes back from infrastructure controlled by a few very large companies. That model is powerful, but it creates concerns around privacy, cost, latency, control and dependence.

Local AI tries to change that balance. If more capable models can run on a device, small server or business appliance, users may get faster responses, more control over private data and lower dependence on a single cloud provider. This is especially relevant for companies that cannot freely send sensitive documents, customer records, code or compliance material to external AI services.

There is also a strategic chip angle. Nvidia and AMD investing in River makes sense because the next AI market may not only be about giant hyperscale clusters. It may also include smaller AI servers, edge boxes and business-grade machines built for inference and agents. If that market grows, it could expand demand beyond the biggest model labs and cloud companies.

The idea fits into a broader trend we have been tracking. Open-weight models are making it easier for developers to experiment outside closed platforms, and we recently explained why open-weight AI models matter for accessibility and research. River is not simply an open-weight story, but it points in the same direction: more users want AI they can shape, inspect and run on their own terms.

The challenge is execution. Local AI sounds attractive, but running useful models privately is still difficult. Hardware costs, memory limits, model updates, security, energy use and maintenance can quickly become a burden for small businesses. A server that runs AI locally must be simple enough for non-experts and powerful enough to justify not using the cloud.

There is also the question of whether users really want to own AI infrastructure. Businesses say they want control, but they also like managed services because someone else handles uptime, patches and scaling. River will have to make local or custom AI feel less like a science project and more like ordinary business software.

Even with those questions, the funding round shows where investor imagination is moving. The AI market is no longer just about bigger chatbots. It is about who controls the stack, where the models run, who owns the data and whether businesses can build agents that feel like their own systems rather than rented intelligence. River AI is now one of the companies trying to answer that question with serious money behind it.