Google has been working on it for years with DeepMind and Isomorphic Labs, Sanofi has bet on Insilico Medicine, and more recently Amgen and Moderna have chosen OpenAI to discover and develop drugs. Now Bristol Myers Squibb also relies on artificial intelligence to accelerate research on new treatments and to do so has formed a partnership with Anthropic, which recently launched a $200 million plan with the Gates Foundation to bring AI to healthcare, education, and research in low- and middle-income countries.
The agreement comes at a time when the global pharmaceutical industry is investing billions of dollars in artificial intelligence, not only to discover new molecules but also to transform manufacturing processes, clinical trials, and regulatory activities. At the same time, China is emerging as a new global epicenter of biotech and AI-based pharmaceutical research, attracting investments and partnerships from major Western multinationals.
CLAUDE ENTERS THE HEART OF BRISTOL MYERS SQUIBB
Bristol Myers Squibb (BMS) has announced the large-scale implementation of Claude, the model developed by Anthropic, in research, clinical development, manufacturing, and commercial and corporate functions. The company will make Claude’s capabilities available to over 30,000 employees, aiming to integrate AI directly into corporate workflows and systems managing scientific, clinical, and regulatory data.
The agreement also includes the use of Claude Code, Anthropic’s software development tool, to accelerate the work of engineering and data science teams. Bristol Myers wants to use AI not just as a chatbot but as an “agentic layer” capable of connecting people, data distributed throughout the company, and serving as a centralized digital archive.
“Most enterprise AI stops at the chatbot. The real goal is the untapped value hidden behind decades of data silos – said Greg Meyers, Chief Digital and Technology Officer of BMS. ‘Anthropic’s Claude offers us the agentic capabilities, innovation pace, and security needed to connect our systems and put this collective knowledge into the hands of every BMS employee.’”
For Eric Kauderer-Abrams, head of life sciences at Anthropic, the goal is to create “a single layer of intelligence” capable of generating clinical reports from trial data, retrieving the correct scientific context from decades of internal research, or identifying in real time the causes of production deviations.
FROM DRUG DISCOVERY TO MANUFACTURING
But the use of AI in the pharmaceutical industry is extending far beyond the discovery of new molecules. According to several industry executives who spoke earlier this year at the JPMorgan Healthcare Conference, artificial intelligence is already being used to identify clinical trial participants, select trial sites, and prepare regulatory documentation, reducing traditionally very burdensome processes by weeks.
Novartis, for example, used AI during a cardiovascular study on 14,000 people for the cholesterol-lowering drug Leqvio. Chief Medical Officer Shreeram Aradhye reported that the clinical site selection process went from 4-6 weeks to a 2-hour meeting thanks to automated data analysis: “AI becomes augmented intelligence, not artificial intelligence.”
As Reuters notes, GSK is also using digital and AI tools to reduce manual data collection and speed up clinical studies by 15%, while Danish biotech Genmab has announced the adoption of Claude’s agentic AI to automate post-trial data analysis and clinical report production.
ALL THE BIG PHARMA COMPANIES BETTING ON AI FIRMS
In recent months, large pharmaceutical companies have multiplied partnerships and investments in AI. Eli Lilly has struck a deal with Nvidia to build what it calls the most powerful supercomputer in the pharmaceutical sector and subsequently expanded collaboration with a joint lab in the Bay Area dedicated to drug discovery through AI.
Roche, adds Fierce Pharma, has also announced building a supercomputer with Nvidia, while Merck has chosen Google and its Gemini ecosystem in an enterprise deal potentially worth $1 billion to integrate AI into research, manufacturing, and corporate functions.
Meanwhile, Novo Nordisk has signed a partnership with OpenAI to integrate AI tools across all its activities, from drug discovery to commercial operations. After that deal, OpenAI launched GPT-Rosalind, a model dedicated to biology, translational medicine, and pharmaceutical research.
According to McKinsey & Company, agentic AI could increase clinical development productivity between 35% and 45% over the next five years.
BETWEEN ENTHUSIASM AND SKEPTICISM
Despite the billion-dollar investments, the most ambitious promise of AI – discovering revolutionary drugs – has not yet fully materialized. “There was a promise to see a drastic improvement” in clinical trial success rates thanks to AI, explained Trung Huynh of RBC Capital Markets, who added: “I don’t think it has happened yet. There is no definitive evidence that AI really improves outcomes.”
However, many companies claim that the first concrete results are emerging. The Wall Street Journal writes that Recursion Pharmaceuticals stated its AI platform identified a protein involved in a rare hereditary colorectal disease, contributing to the development of an experimental treatment that in early studies significantly reduced polyps in patients.
AI is also accelerating the design of experimental oncology drugs. According to Recursion, the time needed to design a new drug candidate has dropped from about 4 years to 18 months thanks to the use of machine learning models and “in silico” simulations.
AI IN PHARMACEUTICAL FACTORIES
Bristol Myers Squibb has also become a case study in advanced pharmaceutical manufacturing. The New York Times reports that the company’s Devens facility was the only U.S. plant included this year in the World Economic Forum’s Global Lighthouse Network, which honors the world’s most innovative factories.
At the manufacturing site, AI monitors variables such as oxygen, temperature, and pH levels in real time during cell cultivation needed to produce biologic drugs and immunotherapies. Systems analyze data from previous batches and suggest corrective actions to avoid production errors.
According to the company, these innovations have increased the volume of drugs produced for clinical studies and commercial use by 40%. “Now we can intervene on batches during the manufacturing process and not wait until the end,” explained Karin Shanahan, executive vice president, head of supply chain and corporate operations.
Eli Lilly has also applied AI to the production of weight-loss drugs Mounjaro and Zepbound. The company created a “digital twin” to simulate the manufacturing process of tirzepatide and use machine learning algorithms to optimize parameters such as plant pressure and temperature.
CHINA AS A NEW BIOTECH HUB
While Western big pharma invests in AI, China is emerging as one of the main global hubs for biotech research. According to McKinsey, the country already accounted for 30% of the global pipeline of experimental drugs last year.
Pfizer paid $1.25 billion to Chinese 3SBio to obtain rights to an oncology drug candidate, while Gilead Sciences invested $120 million in Chinese biotech Pregene Biopharma, specialized in CAR-T therapies. “China is mobilizing its innovation at levels we have never seen before,” said Pfizer CEO Albert Bourla.
For Paul Zhang of Bluestar BioAdvisors, “most American and European pharmaceutical companies are actively seeking new Chinese drugs” because “things are cheaper and faster.”
Not surprisingly, cities like Shanghai and Suzhou have become hubs for biotech startups and advanced labs, supported by public investments, accelerated regulatory procedures, and a growing availability of researchers trained in the United States. Moreover, according to several industry companies cited by the WSJ, starting a clinical trial in China can take 2 to 4 months, compared to 6-9 months needed in the United States.
