Indian scientists and biopharmaceutical firms are using artificial intelligence to identify promising drug candidates faster and at lower cost. But industry experts tell CNA the technology can address only part of a lengthy and risky development process.

NEW DELHI: A team of about a dozen people at Indian drug-discovery startup Peptris is using artificial intelligence to work on more than a dozen potential drug programmes simultaneously.

Peptris co-founder Venkatasubramanian Narayanan said the company advanced one of its drug development programmes to a late preclinical stage in about 18 months, compared with more than four years using conventional methods.

Such fast development could signal a fundamental shift in India’s pharmaceutical sector, helping it to expand beyond its traditional focus and areas of success.

Industry scientists told CNA that India has historically been strongest in generics, manufacturing and contract research for international firms, while original drug discovery remains less developed.

The scale of India’s pharmaceutical industry was highlighted at the Bloomberg New Economy Forum in New Delhi on Tuesday, where Prime Minister Narendra Modi said the country supplies one in every five generic medicines used globally. The government is now investing in drug discovery efforts.

In September, India’s Department of Pharmaceuticals approved about 16 billion rupees (US$168 million) in funding for 41 pharmaceutical and medical technology research projects, including novel drug discovery and clinical development.

Drug developers and scientists say that AI could make it cheaper and faster for more Indian companies to develop their own medicines by identifying promising compounds earlier. But that addresses only part of the challenge.

Noubar Afeyan, co-founder and chairman of US-based pharma giant Moderna, told CNA that India’s ability to turn promising science into medicines will depend on building much more of the infrastructure needed before and during human trials.

“It’s probably more important to get preclinical and clinical capabilities established at scale than just AI,” he said. “I don’t think AI will fix that.”LOWERING THE BARRIERS TO DRUG DISCOVERY The potential significance goes beyond India’s pharmaceutical exports - which rose by 2.1 per cent to about US$31 billion in the financial year ended March 2026, according to government data.

Non-communicable diseases accounted for 49 per cent of deaths in India in 2021, or about 6.4 million, according to the World Health Organization. The financial burden is also significant, with households paying 43.4 per cent of the country’s total health expenditure out of pocket in financial year 2022-23, according to India’s National Health Accounts.

Scientists said that lower research costs could help more Indian companies develop new treatments without the resources of a large pharmaceutical firm.

Megha Goyal, a drug-discovery scientist and co-founder of Abhinavayan Biotech, said AI is already helping smaller Indian biotechnology companies identify promising drug targets, screen potential compounds and design molecules before committing to laboratory experiments.

“AI has not made drug discovery easy; it has made it more accessible and more efficient,” she said.

Peptris’ Narayanan said Indian scientists and contract research organisations already perform sophisticated drug-discovery work, even if Indian companies have historically taken on less of the financial risk involved in owning the medicines being developed.

Goyal said the bigger challenge for Indian drug developers comes after the initial discovery stage.

“Drug discovery is inherently a long-term and high-risk endeavour that requires patient capital, sustained commitment [and] long-term vision,” she said.

Vishesh Rajaram, founding partner at venture capital firm Speciale Invest, said he is now seeing more scientists build companies around drugs of their own rather than provide research services or manufacture existing medicines.

“Today we see (Indian) scientist-founders building novel-chemistry, biologics and AI-first discovery companies with global ambitions from day one,” said Rajaram, whose firm has invested in Peptris.

He said cheaper computing and AI tools can reduce the number of compounds researchers need to make and physically test, allowing smaller teams to identify promising drug candidates with less money.

But Rajaram said those savings are concentrated largely at the beginning of drug development.

“AI lowers the capital needed to create an asset,” he said. “It doesn’t yet lower the capital needed to prove that the asset works in patients.”THE HARD PART Santanu Datta, co-founder and chief scientific officer at Bengaluru-based biopharma firm Bugworks, said AI works best with large amounts of reliable data. This includes the shape of a molecule, its intended target and how similar compounds have behaved.

A drug may appear to interact with its intended target as hoped, but behave differently in patients, affect other proteins or cause unforeseen effects, said Dutta.

He compared it to predicting how a person behaves alone versus trying to predict what happens when that person is placed in a crowded city.

“Once you put the person in a city with a million people, things change,” he said.

Datta has spent more than three decades in drug discovery, including more than 20 years working on infectious diseases at AstraZeneca India.

Bugworks, which he later co-founded, is developing treatments for drug-resistant infections and cancer, and its lead antibiotic has progressed to Phase 1 human trials - the first stage of testing an experimental drug in people, primarily to assess its safety.

But there is still a long road to an approved drug. Only 7.9 per cent of drug programmes that entered Phase 1 trials between 2011 and 2020 ultimately received approval, according to an analysis by the Biotechnology Innovation Organization, Informa Pharma Intelligence and QLS Advisors.

According to the United States Food and Drug Administration, a medicine typically passes through three phases of clinical human trials before regulators decide whether to approve it.

“From idea to drug candidate, there are massive improvements (using AI),” Afeyan of Moderna said. “From drug candidate to human proof of concept, there are smaller improvements.”Another challenge is that large pharmaceutical companies have years of data from failed drug programmes that could help AI identify what is unlikely to work, but much of it remains private, Datta said. ‘VERY NASCENT STAGE’There are early signs that Indian companies are beginning to create novel drug candidates that attract interest beyond the country.

Peptris, for example, has licensed one of its drug candidates to US-based pharma firm Revio Therapeutics for further development, giving it an overseas partner to take the programme forward.

Industry executives said how far Indian companies can develop their own drugs before seeking partners will determine how much value India retains.

Speciale Invest’s Rajaram said licensing Indian-developed drugs overseas was “a genuine step up, but not yet the destination”. A bigger shift, he said, would come when more Indian companies take their own treatments into early human trials before seeking larger partners.

Kiran Mazumdar-Shaw, founder of Indian biopharmaceutical major Biocon, said it was still too early to conclude that AI was transforming drug development in the country.

“We are still at a very nascent stage, and nothing of this is proven,” she said. “Proof of concept at pilot scale is no reflection on what is really the true opportunity at scale.”