xAI co-founder Igor Babuschkin is raising up to $1 billion for a new AI startup, River AI, at a valuation of up to $5 billion, according to Forbes reporting circulating this week. General Catalyst is in talks to lead the round, and Babuschkin himself is committing up to $100 million of his own capital.
The specifics
River AI was incorporated in Nevada on April 20, 2026. Neither Babuschkin nor General Catalyst responded to Forbes for comment, and the article explicitly notes that River AI's technical roadmap and product plans have not been disclosed. The funding target — $1 billion at a $5 billion pre-money — would put River AI in the upper tier of seed-stage frontier-lab rounds, on par with the early raises that launched Thinking Machines Lab and Safe Superintelligence.
Babuschkin announced his exit from xAI in August 2025 and initially positioned his next act as Babuschkin Ventures, an AI safety–focused VC firm that Elon Musk pledged $200 million toward as a limited partner. River AI appears to be a separate operating company rather than a portfolio bet, signalling that Babuschkin intends to ship models, not just write checks.
Context: the xAI co-founder exodus
River AI's emergence caps a near-complete co-founder turnover at xAI. By multiple counts, every one of the eleven (some sources say twelve) original xAI co-founders has now departed, with the most recent exits accelerating after SpaceX's acquisition consolidated control under Musk's umbrella.
Babuschkin's pedigree matters for the raise. As chief engineer, he owned the technical architecture that connected the Colossus supercluster's 200,000-plus GPU array to xAI's training stack — the work that put Grok on a competitive curve in 2024–2025. That operational track record is what underwrites a $5 billion mark on a company with no public product.
What changes for builders
The neolab pattern — small, researcher-led entities raising nine- to ten-figure seeds at multibillion-dollar valuations — is now the dominant funding modality for ex-frontier-lab talent. Babuschkin joins Mira Murati's Thinking Machines, Ilya Sutskever's SSI, and David Silver's Ineffable Intelligence in compressing the gap between "resignation announcement" and "hyperscaler-adjacent compute commitment" to under twelve months.
For enterprise buyers and infrastructure teams, the practical signal is supply-side: another lab with $1 billion of dry powder will compete for the same GB300 allocation, the same data-center power, and the same senior researcher headcount the four incumbents are already bidding on. Expect downstream pressure on training-cluster lead times and senior IC compensation through the back half of 2026.
General Catalyst's lead, if it closes, also continues the firm's pattern of writing the largest checks for solo-founder labs rather than diversifying across the agentic and applied layers.
