AI trims early cancer drug discovery to 3-4 months from 8 years, Mukherjee says

Artificial intelligence is collapsing the front end of cancer drug discovery, Dr Siddhartha Mukherjee said on Saturday at the Hindustan Times Leadership Summit 2026 in New Delhi. The oncologist, who co-founded and runs the US drug discovery startup Manas AI, put the timeline for early-stage discovery at three to four months, against roughly eight years before. He called healthcare the single most critical proving ground for AI.

Manas AI's approach pairs machine learning with the basic rules of physics and chemistry. Rather than screening billions of existing chemicals one at a time, the software hunts for weak spots on disease-causing proteins, Mukherjee explained, comparing it to finding the keyhole on a lock. It then assembles new drug molecules from scratch, atom by atom, to jam them. Locating a target spot on a pancreatic cancer protein takes the company's algorithms two hours, a job that used to take human researchers a decade. Combining AI molecular design with robotic lab testing gets each design-and-test cycle done in three weeks, and Mukherjee said the resulting experimental cancer candidates are up to 200 times stronger than competitor drugs already on the market.

He tempered the wins with a blunt caveat: cancer will not fall in one swoop because every cancer is biologically distinct. Breast cancer, leukemia and lymphoma have seen remarkable progress, he said, while pancreatic and stomach cancers remain formidable. The backdrop is Eroom's Law, the decades-long pattern of rising pharmaceutical development costs paired with fewer new drug approvals.

Speeding up lab discovery is only half the fight, Mukherjee said, because human clinical trials, manufacturing and government approvals still stretch across years and pile on cost. He urged India and China to streamline testing rules without loosening safety checks. According to Mukherjee, no country is better placed to flip Eroom's Law than China and India, provided both push regulatory and technological reform.

He pointed to Immuneel Therapeutics, the Bengaluru cell therapy company he co-founded with Kiran Mazumdar-Shaw, as proof the model travels. The team put an automated manufacturing facility next to doctors in a hospital and used AI-designed binders to build customized blood cancer treatments that cost roughly $35,000 in India against $500,000 in the United States. The work cured a young boy from Kolkata who had run out of standard leukemia therapies.

The stakes are enormous. According to the World Health Organization and its cancer research agency IARC's 2026 report, the world records an estimated 20.6 million new cancer cases and nearly 10 million deaths every year. India alone accounts for 1.56 million new cases and more than 900,000 deaths annually, with 3.32 million people living with the disease over a five-year window, the same report estimates. Separate Indian Council of Medical Research data projected 1.57 million cases for 2025 and noted that one in nine Indians carries a lifetime cancer risk.