One point one billion for a company two months old

River AI came out of stealth on 10 June this year. Two months later it announced it had raised 1.1 billion dollars across its seed and Series A rounds. General Catalyst and Amp Pbc are leading the deal; participants include Nvidia, Amd Ventures, Y Combinator and the Singaporean sovereign fund Temasek. The estimated post-money valuation is around five billion dollars.

The founder is Igor Babuschkin, who left xAI after co-founding it and who had previously worked at DeepMind and OpenAI. The company says it wants to rebuild the technological chain of artificial intelligence from scratch, starting with how models are trained, with the aim of making bespoke training faster, cheaper and reachable for developers and firms that cannot afford it today. The stated positioning is deliberately different from that of the large labs: not assistants meant to replace human work, but models that an individual user or company can train for themselves.

The sum and the calendar

What stands out is not the sum itself, by now routine in the sector, but its relationship to the calendar. One point one billion raised by a company that in June had no public product means investors are pricing the founders' reputation and the scarcity of the skills, not revenue. It is the same dynamic that has produced billion-dollar valuations for companies without turnover over the past two years, and that feeds the argument about how much of this market is sustainable.

It should be said that not all the investors are financial. Nvidia and Amd, both in the round, sell the processors on which the models are trained: funding future buyers of your own chips is an industrial move before it is a speculative one, and the pattern recurs across most of the sector's large rounds. That structure makes it harder to read the scale of invested capital as an independent market signal.

Meanwhile the money keeps arriving. In the same week that River AI announced its round, Anthropic signed a twenty-year lease worth 9.1 billion dollars for computing capacity in Texas. Orders of magnitude that a few years ago marked an exception have become the sector's ordinary tempo.