Kevin Weil seeks $150M for AI startup to gather scientific data, four months after leaving OpenAI

Kevin Weil Targets $750 Million Valuation for Undisclosed AI Science Startup

Kevin Weil, who departed OpenAI in April 2025, is raising $150 million for a new AI startup at a valuation of at least $750 million2

2

. The 43-year-old former chief product officer has pitched the company as a vehicle for gathering scientific data to train AI models1

1

, though the startup's name and product remain publicly undisclosed. According to Business Insider, terms are not final and remain subject to revision1

1

. Weil did not respond when approached for comment.

The exact mechanisms for how Weil plans to gather scientific data for AI models remain unclear. Potential approaches include operating its own laboratories, forming partnerships, acquiring datasets, or embedding software in existing labs2

2

. This ambiguity raises questions about differentiation in an increasingly crowded field of AI-powered scientific research ventures.

From OpenAI for Science to Independent Venture

Weil's journey to launching an AI science startup follows a rapid pivot from consumer technology to scientific applications. He joined OpenAI as chief product officer in June 2024 after product leadership roles at Twitter and Instagram, plus a stint as president of product and business at satellite-data company Planet2

2

. Less than two years into his OpenAI tenure, Weil transitioned from the chief product officer role to lead science initiatives within the research organization.

During his time overseeing OpenAI for Science, Weil developed Prism, a platform designed to integrate cutting-edge AI models into researchers' day-to-day work1

1

. The AI workspace aimed to bring frontier models directly into scientists' research and writing processes2

2

. He departed OpenAI after the science initiative was decentralized across other research teams, suggesting a strategic shift within the company's approach to AI models in research workflows.

Targeting a Different Layer of the AI Stack

Weil's new venture appears focused on a different part of the AI stack than his previous work at OpenAI. Rather than building tools that apply AI models to scientific research, the startup targets data collection—the foundational layer that AI systems need to make AI-driven discoveries2

2

. This positions the company as infrastructure for the broader scientific AI ecosystem, potentially serving as a data supplier to other platforms and research institutions.

The focus on scientific data acquisition reflects growing recognition that quality training data represents a critical bottleneck for advancing AI capabilities in specialized domains. Scientific data presents unique challenges compared to text or image data scraped from the internet, requiring structured collection, validation, and domain expertise.

Competing in an Emerging Lab Race

Weil enters a competitive landscape where well-funded players are already pursuing similar visions. Periodic Labs, founded by former OpenAI researcher Liam Fedus and former Google DeepMind researcher Ekin Dogus Cubuk, raised a $300 million seed round last year to build AI scientists and the autonomous laboratories in which they operate2

2

. Lila Sciences has gone even further, raising a $350 million Series A to scale what it calls "AI Science Factories," bringing its total funding to $550 million2

2

.

These competitors are building end-to-end solutions that combine data generation with AI-powered analysis. Weil's approach of focusing specifically on scientific data collection could prove complementary or face challenges differentiating from more comprehensive offerings. The ability to secure $150 million before revealing a product suggests investors are betting heavily on Weil's track record and vision, despite the competitive dynamics.

The timing of Weil's fundraise coincides with intensifying interest in applying AI to accelerate scientific discovery across fields from drug development to materials science. Whether his focused approach to gathering scientific data will prove more effective than integrated lab-and-AI solutions remains to be seen, but the substantial valuation indicates confidence that data infrastructure represents a valuable position in the emerging AI science ecosystem.