A Bloomberg analysis published October 15 detailed how major pharmaceutical companies have committed billions of dollars to AI-driven drug discovery partnerships over the past two years, even as the industry still has no drug on the market that was substantially designed using artificial intelligence, underscoring a widening gap between investment enthusiasm and demonstrated clinical results.
The Scale of Pharma’s AI Bet
Deals announced in just the past several months include Eli Lilly’s agreements with Insilico Medicine worth up to $2.75 billion and with Profluent worth up to $2.25 billion, AstraZeneca’s partnership with Insitro, Zealand Pharma’s roughly $1.2 billion obesity-drug collaboration with Generate Biomedicines, and GSK’s deal with Relation Therapeutics, among many others. Collectively, these agreements represent tens of billions of dollars in potential milestone payments across the industry’s largest drugmakers.
Why the Results Haven’t Caught Up
Drug development timelines, even when accelerated by AI-driven candidate identification, still require years of preclinical testing and multi-phase clinical trials before any new therapy can reach patients, meaning the current wave of AI drug discovery deals is years away from producing approved medicines regardless of how promising early computational results appear. Industry researchers caution that AI has proven useful for narrowing candidate pools and predicting molecular properties, but has not yet demonstrably shortened the overall timeline from discovery to approval in a way that has produced a marketed drug.
The Business Logic Behind the Deals Anyway
Pharmaceutical executives argue that even incremental improvements in hit rates or reduced attrition during early-stage development justify the investment, given that traditional drug development costs can exceed $2 billion per approved therapy when failures across the pipeline are accounted for. Partnering with AI-native biotechs also allows large pharma companies to access specialized machine learning talent and infrastructure without building it entirely in-house.
Skeptics Within the Industry
Some biotech investors and scientists have grown more cautious, pointing to the gap between impressive-sounding computational predictions and the historically high failure rate of drug candidates in human trials, regardless of how they were initially identified. Critics argue that AI drug discovery deals carry elements of speculative enthusiasm similar to broader AI investment trends, with valuations sometimes outpacing demonstrated scientific validation.
Where Early Evidence Does Exist
Some AI-discovered candidates have advanced into human clinical trials faster than historical averages, and companies including Insilico have pointed to specific molecules that moved from computational design to first-in-human testing in under 18 months, a pace that would be unusual under traditional discovery methods. Whether these faster timelines translate into higher approval rates remains to be seen as the first cohort of AI-originated candidates works through later-stage trials.
What’s Next
The industry’s first real test of whether AI drug discovery investment pays off will come as candidates from the earliest deals, signed roughly two to three years ago, reach late-stage trial readouts over the next several years. Until then, continued deal-making suggests pharmaceutical executives remain willing to bet on AI’s eventual payoff even without proof it has yet delivered an approved medicine.
Photo: Thirdman / PEXELS via Pexels
