Key Points
- Prime Intellect raised a $130 million Series A at a $1 billion valuation.
- Radical Ventures led the round; Nvidia, Intel, and Dell venture arms joined.
- The company reports a $100 million annualized revenue run rate.
- Customers include Ramp, Zapier, and Flapping Airplanes; it was founded in 2024.
Prime Intellect has raised a $130 million Series A at a $1 billion valuation, and the cap table is the story. Nvidia, Intel, and Dell — three companies whose fortunes are tied to the biggest AI labs buying the most hardware — just funded a startup whose entire pitch is that you should not have to depend on those labs to train a capable model.
The round was led by Radical Ventures, with Nvidia Ventures, Intel Capital, Dell Technologies Capital, and Iconiq joining alongside a roster of operator angels: Perplexity's Aravind Srinivas, Box's Aaron Levie, Harvey's Winston Weinberg, Cognition's Jeff Wang, and Mercor's Brendan Foody. It is the company's largest round to date, less than two years after it was founded in 2024.
What Prime Intellect sells
Prime Intellect runs a full-stack platform for building and training AI agents: compute access, reinforcement-learning frameworks, and evaluation tools, packaged in a modular marketplace. The point is to let a company train its own agent on its own data rather than renting intelligence from a frontier lab through an API. TechCrunch reported the company has reached a $100 million annualized revenue run rate, with customers including Ramp, Zapier, and Flapping Airplanes. Ramp, per the company, built an agent on the platform that beat frontier models on accuracy while running faster and at a fraction of the cost.
The read
"It shouldn't just be a few nerds in a glass tower in San Francisco that have the capability to train AI models," co-founder Vincent Weisser told TechCrunch — a line that doubles as the investment thesis. The tension worth watching is that the same infrastructure giants bankrolling the frontier labs are now hedging into the tools that would let everyone else route around them. That is what a $1 billion valuation on a two-year-old company buys a view of: the belief that the durable money in AI is moving down a layer, from who owns the largest model to who owns the pipes that let enterprises train their own. It rhymes with the rest of the 2026 raise cycle — Together AI's $800 million round at an $8.3 billion valuation and Baseten's $1.5 billion Series F both priced the same wager on the layer beneath the models. The open question is whether "train your own" stays a real alternative or becomes a feature the frontier labs absorb, the way they have already moved to standardize the agent stack around themselves.
Source: TechCrunch.
Comments