Ex-OpenAI Chief Launches Warp Sciences With $215M Robot Lab Bet

Former OpenAI science lead Kevin Weil has launched Warp Sciences, a startup that wants to move frontier AI out of the browser and into the chemistry lab. The company came out of stealth on October 22 with a $215 million Series A round led by Lightspeed Venture Partners, valuing Warp Sciences at $865 million, TechCrunch reports. The announcement, made by Weil on X, marks one of the largest recent bets on physical, robot-driven science.

Warp Sciences is not trying to discover drugs itself. Instead, it is going after the most labor-intensive step that comes after a promising molecule is found: synthesizing hundreds or thousands of variations of it so chemists can improve efficacy and cut side effects. By training robots to make and test those compounds around the clock, Warp Sciences aims to turn wet-lab chemistry into an automated, data-generating engine for drug discovery.

According to Weil, the Series A will fund the build-out of U.S.-based chemistry labs where artificial intelligence, robotics, and laboratory services are combined for biotechnology and pharmaceutical customers, as reported by citybiz. The company has assembled a team of about 20 and has already purchased robotic arms and AI chips for its first automated facilities.

What Warp Sciences plans to automate

The core idea behind Warp Sciences is simple: AI models are only as good as the real-world data they train on, and chemistry generates far too little of it today. Human chemists at Warp Sciences will wear cameras and pressure-sensing gloves while running experiments, and that footage becomes the training data for the company's own models. Over time, robots take over progressively more of the physical work, running synthesis tasks continuously while the models decide which molecules to make and how to produce them.

The company's initial offering centers on lead optimization analog synthesis, the process of creating variations of a promising molecule so researchers can compare their properties and identify candidates for further development. Warp Sciences intends to compete on speed, cost, and intellectual property protection, while the stream of experimental data steadily improves the performance of its AI systems.

That data flywheel is the long-term prize. Each physical experiment teaches the models something a simulation cannot, and automated labs can run far more experiments per week than human teams. As the company puts it, chemistry and biology advance when something is made, tested, and learned from — and Warp Sciences wants to generate that real-world data at industrial scale.

The reshoring bet behind Warp Sciences

Warp Sciences is also a geopolitical play. For years, drugmakers have outsourced the slog of synthesizing molecular variations to contract firms in China, most notably WuXi AppTec. Weil wants to bring that work back to the United States, insulating the drug development pipeline from supply-chain risks and political friction.

The timing is pointed. In June, the Pentagon added WuXi AppTec to a list of companies it says are linked to China's military, and although a U.S. judge blocked enforcement of that designation in August after the company sued, the episode left the industry uneasy, according to reporting on the launch. Warp Sciences is positioning its automated domestic labs as the secure alternative: the chemistry gets done on American soil, by robots, with the intellectual property never leaving the country.

The round's investor list reads like a who's who of frontier-tech capital. Beyond Lightspeed, participants include IVP, Mayfield, Lux Capital, Premji Invest, Menlo Ventures, Conviction, Lowercarbon Capital, SignalFire, and others, according to company materials. Palantir's Shyam Sankar is also reported among the backers.

A growing race to physicalize AI

Warp Sciences lands in the middle of a broader shift: AI talent leaving the big labs to build things that touch the real world. Weil ran OpenAI's science push as chief product officer from 2024 to 2026 before founding the startup, and his pitch — that AI progress in science is bottlenecked by physical experimentation, not by models — echoes a growing consensus among researchers.

The capital flooding into adjacent AI infrastructure tells the same story. OpenAI's record $30 billion raise aims at pre-IPO scale, while initiatives like Google's Agent Compute Engine are giving AI agents their own computers to work from. Even open collaboration is accelerating, with Anthropic and OpenAI donating agent skills to a shared foundation. Warp Sciences is the physical-world counterpart to that digital buildout: agents that don't just reason about chemistry, but run the experiments.

Questions remain. Warp Sciences has not shipped a product, has not said where its labs will be, and has not named a timeline for its first customers. Automating wet chemistry is notoriously harder than automating a data pipeline — liquids spill, reactions surprise, and robots break. But with $215 million in the bank and a veteran AI product leader at the helm, Warp Sciences has bought itself a serious runway to find out whether robots can learn the oldest job in science.