AI cuts early cancer drug discovery from 8 years to 3 months: Dr Siddhartha Mukherjee

New Delhi: Artificial intelligence (AI) is compressing early-stage cancer drug discovery from nearly eight years down to three or four months, according to Dr. Siddhartha Mukherjee, an oncologist and co-founder and CEO of Manas AI, an American drug discovery startup.

Speaking at the Hindustan Times Leadership Summit 2026 in New Delhi on Saturday, Dr. Mukherjee said healthcare is the single most critical proving ground for AI. For decades, the pharmaceutical sector has faced rising development costs alongside fewer new drug approvals—a trend known as Eroom’s Law.

To reverse rising drug costs, Manas AI combines machine learning with basic rules of physics and chemistry. Instead of testing billions of existing chemicals one by one to see if they work, the software finds weak spots on disease-causing proteins—much like identifying a keyhole on a lock—and builds new drug molecules from scratch out of atoms to block them, said Dr. Mukherjee, also an associate professor of medicine at Columbia University.

Finding a target spot on a pancreatic cancer protein, a task that previously took human researchers a decade, now takes Manas AI’s algorithms just two hours. By pairing AI molecular design with automated robotic lab testing, the company completes each design-and-test cycle in three weeks, yielding experimental cancer candidates up to 200 times stronger than existing competitor drugs on the market.

No quick fix for cancer yet

Despite these breakthroughs, Dr. Mukherjee cautioned that cancer cannot be cured in one fell swoop because every cancer is biologically distinct. While remarkable progress has been achieved against specific cancers such as breast cancer, leukemia, and lymphoma, complex diseases like pancreatic and stomach cancers remain formidable challenges, he said.

Dr. Mukherjee also emphasized that speeding up early lab discovery at Manas AI is only half the battle, noting that human clinical trials, manufacturing, and government approvals still take years and drive up costs.

“AI buys time and cost in the front end, but what it does not change yet is the cost of running a trial and the regulatory clock,” he said, urging India and China to streamline testing rules without compromising safety checks. “Of all countries in the world, China and—I truly believe—India, with appropriate regulatory and technological reform, can reverse Eroom’s Law.”

Highlighting real-world impact, Dr. Mukherjee pointed to Immuneel Therapeutics, the Bengaluru-based cell therapy company he co-founded with Kiran Mazumdar-Shaw. By placing an automated, hospital-adjacent manufacturing facility right next to doctors and using AI-designed binders, the team developed customized blood cancer treatments costing roughly $35,000 in India compared to $500,000 in the United States, successfully curing patients like a young boy from Kolkata who had exhausted all standard leukemia therapies.

According to the World Health Organization (WHO) and its cancer research agency (IARC) 2026, the world sees an estimated 20.6 million new cancer cases and nearly 10 million deaths every year.

In India, the same global report estimates 1.56 million new cases and over 900,000 deaths annually, with 3.32 million people living with the disease over a five-year period. Separate data from the Indian Council of Medical Research (ICMR) projected 1.57 million cases for 2025, noting that one in nine Indians faces a lifetime cancer risk.