Prime Intellect, a startup founded just two years ago, has secured $130 million in Series A funding, pushing its valuation to $1 billion. This substantial capital injection, led by Radical Ventures with participation from Nvidia Ventures, Intel Capital, Dell Technologies Capital, and Iconiq, aims to equip large corporations with the infrastructure to develop their own AI agents.
The company's stated goal is to allow organizations to train their own agentic systems, bypassing reliance on established frontier AI labs. This move represents a strategic shift for capital, enabling individual enterprises to refine models for specific business tasks, effectively becoming their own AI development centers. The underlying infrastructure for such systems remains complex, a barrier Prime Intellect claims to dismantle for its corporate clientele.
Capital's New Tools
Prime Intellect offers a full stack for AI agent development, including compute access, a reinforcement learning framework, and evaluation tools. Its platform operates as a modular marketplace, allowing corporate customers to select only the tools they require. David Katz, a partner at Radical Ventures, noted the company's ability to provide top-tier AI lab capabilities as a "one-stop shop" that's "affordable" for enterprises.
This approach has already attracted customers like Ramp, Zapier, and Flapping Airplanes, driving Prime Intellect to an annualized revenue run rate of $100 million. Ramp, a fintech firm, utilized Prime Intellect to develop an agent for analyzing spreadsheets. Karim Atiyeh, Ramp’s co-founder and co-CEO, reported that the agent surpassed frontier models in accuracy, operated at faster speeds, and did so at "a fraction of the cost."
Securing Corporate Control
The drive for corporate self-sufficiency in AI is rooted in capital's desire for control. Katz highlighted that companies are increasingly hesitant to share proprietary information with external AI providers like OpenAI and Anthropic, fearing loss of data control. He also pointed to the risk of models being abruptly deactivated, citing Anthropic’s Fable as a recent example. "How do I know that I’m not working with a company that is going to try to replace me and generalize to what I’m doing," Katz questioned, articulating capital's inherent distrust and competitive nature. This fear fuels the push for enterprises to "own [their] own enterprise intelligence" and mitigate such risks.
Vincent Weisser, Prime Intellect co-founder and CEO, echoed this sentiment, stating that enterprises seek to move away from closed-source frontier models. He asserted that his company provides the necessary infrastructure for this transition. Weisser declared, "It shouldn’t just be a few nerds in a glass tower in San Francisco that have the capability to train AI models. It should be every enterprise, every nation state." This vision, while presented as a democratization of technology, in fact represents a decentralization of advanced tools among the ruling class and state apparatus, further entrenching their power and control over productive forces.