Prime Intellect, a startup that lets companies train and run their own AI models instead of relying entirely on frontier labs, has raised $130 million in a Series A round led by Radical Ventures, the company announced July 8. The round values Prime Intellect at $1 billion and pushes its total funding past $150 million.
NVIDIA Ventures, Intel Capital and Dell Technologies Capital also joined the round, according to Prime Intellect, alongside a group of angel investors that includes OpenAI co-founder John Schulman, Perplexity CEO Aravind Srinivas and Box CEO Aaron Levie.
What Prime Intellect Sells
Founded in 2024 and led by chief executive Vincent Weisser, Prime Intellect offers what it calls an “Open Superintelligence Stack” — compute access, reinforcement-learning tools, evaluation environments and deployment infrastructure bundled together so businesses can train and continuously improve their own machine learning models rather than depend entirely on closed systems from labs like OpenAI or Anthropic.
The company says it has surpassed $100 million in annualized revenue in under a year of operation and now serves more than 6,000 customers. Among the examples it cites is fintech company Ramp, which Prime Intellect says trained a 35-billion-parameter model on its platform that outperformed Anthropic’s Claude Opus on a spreadsheet-search task while running 27% faster.
Riding the Open-Infrastructure Wave
The raise lands as more enterprises look for alternatives to paying per token for proprietary frontier models, weighing concerns about cost, data privacy and vendor lock-in. Radical Ventures partner David Katz said the firm backed Prime Intellect because it has assembled compute, training and evaluation tools that are typically scattered across multiple vendors into what he called a “one-stop shop” for building AI agents.
Prime Intellect said the new funding will go toward expanding compute capacity and its research efforts, including work on recursive language models and autonomous model training.