OpenAI前科学负责人Kevin Weil创办AI药物研发公司Warp Sciences，携2.15亿美元融资出隐身，主打回流药明系主导的分子优化外包环节

Kevin Weil, the former OpenAI executive who led the ChatGPT maker’s AI-for-science efforts, is launching a new startup aimed at using artificial intelligence to speed up a key step of drug development.

Warp Sciences is launching out of stealth with $215 million in funding, the company plans to announce Thursday. Lightspeed Venture Partners led the fundraising effort, with other firms including IVP, Premji Invest and Lux Capital also investing. The round values Warp at $865 million after the new money, according to the company.

A slew of angel investors joined the round, too. The long list includes many of Weil’s peers at OpenAI, such as Chief Financial Officer Sarah Friar and former executive Fidji Simo, as well as other tech personas including Coinbase Chief Executive Officer Brian Armstrong, Palantir Chief Technology Officer Shyam Sankar and LinkedIn co-founder Reid Hoffman.

AI for drug development has become a hot topic among venture-backed companies — including giants such as OpenAI. Alphabet Inc.’s Isomorphic Labs, which is using AI models to discover drugs, has held early talks for a new round of funding at a valuation of at least $40 billion, Bloomberg previously reported. And earlier this week, OpenAI acquired Orchestra Bio, a platform that uses AI agents to help pharmaceutical companies manage their drug development processes.

Warp, rather than doing its own drug discovery, is targeting a specific, labor-intensive step that pharmaceutical companies have historically outsourced to Chinese firms. After researchers identify a promising molecule, they often generate hundreds or even thousands of variants to increase the molecule’s likely effectiveness and reduce harmful side effects in humans.

Chinese companies, particularly WuXi AppTec Co., have come to dominate that market by offering cheaper and faster results than US counterparts. That reliance has become increasingly fraught as tensions between the US and China spill into pharmaceutical supply chains.

In June, the Pentagon named WuXi AppTec on a list of firms allegedly linked to the Chinese military, which prompted some of the company’s customers to cut ties. (A US judge then blocked the enforcement of that designation in August after WuXi sued the government.)

Some pharma companies also have concerns about sharing their sensitive data and intellectual property with Chinese contractors, Weil said.

Warp’s goal is to reshore this process, Weil said, and in order to compete with the efficiency of Chinese companies, the startup wants to build a fully automated laboratory.

The company is starting with a team of about 20 people, and has already amassed hardware, including robotic arms and AI chips. Warp’s chemists will use wearable cameras, gloves with pressure sensors, and other devices while running experiments to generate training data for custom AI models, Weil said.

Weil, the former vice president of OpenAI for Science, is serving as Warp’s CEO. His co-founders are Manohar Paluri, a former vice president at Meta Platforms Inc.’s Superintelligence Labs; DJ Patil, the White House’s first chief data scientist; and Solomon Reisberg, a former executive at Terremoto Biosciences.

Weil said OpenAI has seen its biggest scientific research successes in mathematics, pointing to the AI company’s breakthroughs with previously unsolved math problems. But traditional large language models run into problems in physical sciences, Weil said.

“It's still an area where frontier AI models are not yet changing the world in the way models are changing the world in software engineering and math. That’s because the field is so much more experimentally driven. The models need to see that data,” he said.

Guru Chahal, a partner at Lightspeed who is joining Warp’s board of directors, said his firm saw a need for the startup’s technology among its own life sciences portfolio.

“It was almost as if — forget about the investment, please go build this and we will bring the customers, because our portfolio companies will actually pay more for this,” Chahal said. He noted that Warp could expand to serve a variety of disciplines beyond chemistry in the long term.

Warp plans to use a mix of frontier models and its own customized open-weight models. While those efforts are early days, Weil said Warp would ideally use US-based open-weight models. Many of the highest-performing open-weight models on the market today are made in China.