Prime Intellect Secures $130M Series A to Power Enterprise AI Agent Development

Prime Intellect, a startup focused on providing computing infrastructure and specialised software for AI development, has secured $130 million in Series A funding at a $1 billion valuation. The company aims to help enterprises build and train their own AI agents without depending entirely on major frontier AI providers.

The funding round was led by Radical Ventures and included participation from Nvidia Ventures, Intel Capital, Dell Technologies Capital, Iconiq, along with a wide range of angel investors. Among those backing the company are founders of several prominent technology businesses, including Aravind Srinivas of Perplexity, Aaron Levie of Box, Winston Weinberg of Harvey, Jeff Wang of Cognition, and Brendan Foody of Mercor.

Established in 2024, Prime Intellect was created to give organisations the tools needed to develop their own agentic AI systems independently of frontier AI laboratories. While that vision would have been difficult to achieve only a few years ago, recent advances in reinforcement learning have changed the landscape. By rewarding successful outcomes and penalising errors through repeated iterations, reinforcement learning allows businesses to refine AI models for highly specific operational tasks, effectively enabling them to function as their own AI research labs.

Although organisations now have more opportunities to reduce reliance on proprietary AI providers, assembling the underlying infrastructure remains highly complex. Many enterprises lack the technical expertise to integrate computing resources, training frameworks, and evaluation systems into a production-ready AI platform.

Prime Intellect aims to solve that challenge.

The company has developed what it describes as a complete AI development stack, combining access to computing resources, reinforcement learning frameworks, and model evaluation tools within a single platform.

Its system operates as a marketplace, allowing customers to select only the individual components they require rather than committing to a single, all-inclusive platform. This modular approach gives organisations greater flexibility while avoiding vendor lock-in.

David Katz, a partner at Radical Ventures, said Prime Intellect has successfully integrated these capabilities into a platform that operates at frontier-level performance while remaining cost-effective. He added that although other providers offer individual components of the AI development process, Prime Intellect stands out by delivering the capabilities of a leading AI laboratory through a single, unified development environment.

The company’s approach has already attracted customers including Ramp, Zapier, and Flapping Aeroplanes, all of which use hosted versions of Prime Intellect’s platform. According to the company, that growing customer base has helped it reach an annualised revenue run rate of $100 million.

Prime Intellect attributes much of that growth to measurable business results. One example involved the fintech company Ramp, which used the platform to develop an AI agent capable of extracting information from spreadsheets. In a statement, Ramp co-founder and co-CEO Karim Atiyeh said the resulting system delivered higher accuracy than frontier AI models while operating at greater speed and significantly lower cost.

Another important factor contributing to the company’s expansion is a growing shift in how businesses view dependence on frontier AI providers.

Many organisations are becoming increasingly reluctant to share proprietary business information with companies such as OpenAI and Anthropic because of concerns about maintaining control over sensitive data. Businesses are also wary of relying on AI models that could become unavailable without warning, following incidents such as Anthropic’s decision last month to discontinue access to its Fable model.

Katz said many enterprise customers are asking how they can avoid becoming dependent on AI providers that could eventually compete directly with them or expand into their industries. Those concerns, he explained, are encouraging organisations to explore ways to own and control their AI capabilities while reducing operational risks.

Prime Intellect co-founder and CEO Vincent Weisser believes this shift reflects a broader movement away from closed-source frontier models. He said the company’s mission is to provide the infrastructure enterprises need to make that transition successfully.

According to Weisser, the ability to train advanced AI models should not remain concentrated in the hands of a small number of technology companies. Instead, he believes every enterprise and every nation should have access to the tools required to build and develop their own artificial intelligence systems.