Prime Intellect announced that it raised $130 million in Series A funding to build what it describes as the open superintelligence stack. The round was led by Radical Ventures, with participation from NVIDIA Ventures, Intel Capital, Dell Technologies Capital, and existing investors.
The funding brings Prime Intellect’s total funding to over $150 million. The company is focused on building infrastructure that enables teams to train, deploy, and continuously improve AI models.
The company said reinforcement learning is changing who can build frontier AI. Prime Intellect argues that companies can now own their model optimization loop by training models directly on their own products, workflows, and production environments.
Prime Intellect’s open superintelligence stack spans compute, large-scale reinforcement learning, environments, sandboxes, evaluations, inference, and deployment. The company said it trains open frontier models and ships the same stack to customers.
Prime Intellect said more than 6,000 customers are using its stack across compute, RL and post-training, sandboxes, inference, environments, and evaluations. The company also said demand has scaled to more than $100 million in annualized revenue in under a year.
The company pointed to Ramp as an example of how customers are using its platform. According to Prime Intellect, Ramp trained a 35-billion-parameter model on its Lab platform that outperformed Opus on spreadsheet search while running faster and at a lower cost than Haiku.
Prime Intellect plans to scale every layer of its stack, including larger compute clusters, larger RL runs, agentic training, inference, and continual learning. The company is also investing in long-horizon agents, Recursive Language Models, automated AI research, and models that learn in production.
The company said its goal is to help teams train frontier models they own using their own data, workflows, and products. Prime Intellect also said it is hiring across reinforcement learning, inference, distributed systems, and compute.
