Prime Intellect Raises $130M Series A to Build the Open Superintelligence Stack

Today, we're announcing that we've raised $130M, led by Radical Ventures, with participation from NVIDIA Ventures, Intel Capital, Dell Technologies Capital, and our existing investors. Bringing our total funding to over $150M to build the open superintelligence stack.

We're also joined by angels who are building the frontier themselves:
Karim Atiyeh (Ramp), John Schulman (Thinking Machines), Aravind Srinivas (Perplexity), Aaron Levie (Box), Dwarkesh Patel, Milan Kovac (Tesla), Winston Weinberg (Harvey), Mike Knoop (Zapier, Ndea), Asher Spector (Flapping Airplanes), Jeff Wang (Cognition), Rohan Anil (CoreAuto), Matthew Prince (Cloudflare), Brendan Foody (Mercor) and many more.

The products that feel magical today, like the leading coding agents, weren't built on APIs alone. They came from teams with full control over the model↔product optimization loop: scaling RL on the right harness, shipping, and feeding real-world signal back into the next run. That loop is why these products work, and why nobody outside a handful of labs has shipped anything close.

That capability has been locked inside frontier labs. We're opening it up.

Why now: RL changes who can build frontier AI

The pre-training era concentrated model development in a handful of labs. Reinforcement learning breaks that structure open. For the first time, it makes sense for companies to own their model optimization loop — to train directly on their own product surface, optimize for their specific workflows, and build agents that improve continuously in production.

Training and inference start to collapse into a single system: models deploy, interact with real tasks, generate feedback, and improve in near real-time. Models no longer have a fixed final state — they learn in production and compound performance with usage. The company that owns that loop owns the model.

Companies are already post-training open models against their own product as the RL environment, owning their optimization loop instead of renting it from closed labs. The shift is underway: every company will become an AI company, and the ones that own their learning loop will win.

But the infrastructure to do this hasn't existed outside frontier labs. That's the stack we've built — and the stack we validate by training our own frontier models on it.

The Open Superintelligence Stack

We train frontier models and ship the exact same stack to our customers. It's one system, spanning the full stack — from compute to large-scale reinforcement learning, environments, sandboxes, evals, and deployment — so any team can run frontier-scale agentic training and inference.

Who's building on Prime Intellect

A growing community of AI-native companies, research labs, and enterprises already run on our stack — across compute, post-training, environments, and evaluations.

In under a year, that demand has scaled to over $100m annualized revenue.

We are grateful to 1000s of customers for working with us including:

• Ramp
• NVIDIA
• Zapier
• Character.AI
• Goodfire
• Inception
• Arcee
• Browserbase
• Flapping Airplanes
• Standard Intelligence

What's next

We're scaling every layer of the stack — ever-larger compute clusters, larger RL runs, and the tooling to make frontier agentic training and inference usable, then continuous, then autonomous.

Beyond that, we are placing ambitious bets at the frontier of where the puck is going and build infrastructure for the problems we believe are most consequential, such as:

• Long-horizon agents and Recursive Language Models (RLMs). Today's models break down over long contexts; RLMs manage their own context and coordinate sub-agents. We've been scaling RLM training over the last months — we believe it will be the scaling paradigm for agents that work for days, not minutes.
• Automated AI research and science.
• Continual learning. The future is models that learn in production, where training and inference collapse into a single continuous loop. Our stack was built for this world — a tight integration between RL rollouts, training, and serving.

Join us

We're a small team building the infrastructure for open superintelligence — and we're racing the most well-funded labs in the world to do it in the open. The same stack that trains our frontier models is now in the hands of thousands of teams, and we're just getting started.

We're hiring engineers and researchers across RL, distributed systems and compute infrastructure.

If that's you, we'd love to hear from you. See open roles →