River AI, a Palo Alto startup barely four months old, has secured $1.1 billion in combined seed and Series A funding, according to TechCrunch, which first reported the deal. The round was co-led by venture firm General Catalyst and AMP PBC, a new AI-focused investment firm founded this year by former Andreessen Horowitz partner Anjney Midha.
Chipmakers Nvidia and AMD Ventures took strategic stakes, while Y Combinator and Temasek, Singapore's state investment fund, also joined. The company has not disclosed a valuation.
A founder with an unusual exit story
River AI was founded by Igor Babuschkin, who co-founded Elon Musk's xAI in 2023 and previously worked on generative modelling and reinforcement learning at Google DeepMind, later leading large-scale training efforts at OpenAI. He announced his departure from xAI in August 2025, saying at the time that he planned to start a venture firm to back AI safety research and agentic systems, following what Forbes described as a broader exodus of founding staff from the company.
That venture vehicle evolved into River AI, which TheNextWeb reports was incorporated in Nevada in April this year and has raised its enormous first round in roughly four months, an unusually fast and large jump even by the standards of a sector accustomed to outsized funding rounds.
Ownership over rental
River's pitch is that most companies today rely on general-purpose AI models trained on the broad internet and designed for the widest possible audience. Building a custom model, by contrast, has typically required a dedicated infrastructure team, specialised chips and months of engineering work, putting it out of reach for most organisations, according to the company's own announcement carried by Business Wire.
River says it wants to close that gap by giving developers and enterprises tools to train, tune and serve their own models on proprietary data. The company's longer-term ambition, it says, is to extend that same control down to individual users, effectively letting people own a personal AI shaped by their own information rather than depending entirely on a handful of large labs.
“"American leadership in AI urgently requires leadership in open weight models, while maintaining a lead in closed frontier models," said Hemant Taneja, chief executive of General Catalyst.”
Marc Bhargava, a managing director at General Catalyst, framed the bet in similar terms: "There is a gap between what AI can do and what most companies actually experience... River closes this gap, helping any company build models on their own data, tailored to how they actually work."
Why it matters beyond Silicon Valley
The debate River is stepping into has particular resonance in Europe, where policymakers and businesses have spent recent years pushing for greater control over the AI systems they depend on, often described as digital or AI sovereignty. French lab Mistral AI has built its business partly around offering open-weight models and European data hosting as an alternative to the mostly American proprietary systems that dominate the market.
River's model, giving companies the ability to train and own bespoke systems on their own data, addresses a related concern that has surfaced repeatedly in EU discussions on AI regulation: that businesses relying on outside providers for critical AI infrastructure may have little visibility into how those models work or how their data is used. Whether River can deliver a genuinely open and portable alternative to the large closed labs, at the scale its backers are betting on, remains to be tested; the company has so far said more about its ambitions than about the specific products it will ship first.
For now, River AI's headline achievement is the size and speed of its fundraising rather than any released product. As TechStartups reported, the round leaves the young company with serious financial backing to pursue what it calls a fundamental question in the AI industry: who ultimately owns the intelligence that increasingly has access to people's work, preferences and data.
This article is free to read. It always will be — no paywall, no account, no tracking.
