Prime Intellect Raises $130 Million to Help Companies Build Their Own AI Agents
Prime Intellect Raises $130 Million to Help Companies Build Their Own AI Agents
Prime Intellect has raised $130 million in Series A funding at a $1 billion valuation, led by Radical Ventures and including investments from Nvidia Ventures, Intel Capital, Dell Technologies Capital, and Iconiq TechCrunch. The company also attracted a notable group of angel investors, including founders from Perplexity, Box, Harvey, Cognition, and Mercor.
Founded in 2024, Prime Intellect sells a platform that bundles three things: access to computing power, tools for teaching AI agents through reinforcement learning (a technique where AI systems improve by trial and feedback), and systems to measure how well those agents perform. The company targets enterprises that want to train AI agents specifically for their own workflows rather than simply relying on off-the-shelf AI models from OpenAI, Anthropic, or Google.
The company has reached $100 million in annualized revenue run rate, according to TechCrunch, and counts Ramp, Zapier, and Flapping Airplanes among its customers.
The jump in valuation is significant. Prime Intellect's earlier announced funding round, in February 2025, was $15 million, bringing cumulative funding to just over $20 million. In roughly eighteen months, the company has moved from that modest total to a $1 billion valuation — a trajectory that places it among a small group of infrastructure-focused AI startups whose valuations are scaling quickly alongside revenue.
The Build-Your-Own AI Approach
Co-founder and CEO Vincent Weisser told TechCrunch that Prime Intellect's core pitch is straightforward: give enterprises the tools to train and tune AI agents on their specific tasks, so they don't have to depend entirely on general-purpose models. This sits at the center of a bigger tension in enterprise AI over the past two years: should companies pay-per-use for off-the-shelf AI APIs, or invest in building custom agents tuned to their own workflows.
The clearest example of this value proposition is Ramp, the corporate card and spend-management platform. Ramp used Prime Intellect's platform to build an agent that answers questions about data in spreadsheets. According to TechCrunch, the resulting agent was more accurate than frontier models while running faster and costing less to operate — a combination that, if it holds across different business problems, points to a real advantage narrow, specialized agents have over general-purpose models designed to handle any task.
What the Investor List Signals
The array of investors tells its own story. Hardware vendors Nvidia, Intel, and Dell all participating in the same round suggests they view Prime Intellect's bundled compute-and-tooling offering as complementary to their own infrastructure products, not a threat to them. The angel investors — founders of Perplexity (an AI search engine), Harvey (an AI-powered legal tool), and Cognition (an AI coding assistant), alongside executives from Box and other enterprise software — represent both AI-native companies and traditional software firms now under pressure to adopt agentic AI at scale. That mix is a data point in itself about where informed investors think this market is heading.
The Broader Shift in Enterprise AI Spending
Through 2024 and into 2025, most enterprise AI spending went toward paying per-token for frontier models — essentially renting access to GPT-style or Gemini-style systems with little to no customization. Prime Intellect's reported traction suggests a meaningful share of that spending is now flowing toward companies that want to own the training process: building their own reinforcement learning pipelines, creating evaluation systems, and running fine-tuned AI checkpoints on rented or dedicated compute infrastructure.
Whether this shift will prove lasting or reflects temporary enthusiasm for anything labeled "agentic AI" is a fair question. The $100 million revenue run rate is a snapshot at one moment, not a proven trend, and a two-year-old company valued at $1 billion naturally invites the kind of scrutiny that has, in previous technology cycles, separated enduring platform businesses from shorter-lived enthusiasm plays.
In my view, the technical foundation of Prime Intellect's bet — that narrow, specialized agents optimized through reinforcement learning can outperform general-purpose models on cost and accuracy for well-defined enterprise tasks — aligns with what applied machine learning practitioners have observed for years. Whether Prime Intellect itself becomes the company that scales this opportunity widely is a separate question, and one the investors are betting several hundred million dollars will be answered affirmatively.
What matters next is whether more customers can achieve what Ramp did across different kinds of business problems, and whether the cost savings hold up as the company moves beyond early adopters to the broader enterprise market.
