Bernstein: About 70 AI drug discovery programs have entered clinical trials; definitive answers will still take 3 to 5 years

TechFlow reports, according to Trend Research, Bernstein's August 26, 2026 research report indicates that approximately 70 AI-derived drug development projects have entered the clinical pipeline, with the vast majority still in early stages. The true test will lie in the next three to five years. DeepMind and Anthropic CEOs recently claimed they would "cure most diseases within 5 to 10 years," but Bernstein considers such statements more akin to defending AI's social value rather than reflecting the practical realities of drug development. From target discovery to clinical trials and healthcare system capacity, every stage faces bottlenecks. Ultimately, biological validation must occur in the human body, a pace that cannot be compressed by algorithms.

Bernstein argues that AI's core value in pharmaceuticals lies in enhancing decision-making quality, not merely accelerating workflows. A 20% increase in success rates generates substantially greater R&D returns than equivalent reductions in time and costs. When AI is applied to tackle hard-to-treat diseases with incomplete biological understanding, the value it creates can be even greater despite a lower overall success rate. Major pharmaceutical companies exhibit significantly different AI deployment strategies: Eli Lilly maintains the broadest external partnerships (25 deals), Roche holds the largest computing power investments, and Amgen possesses the deepest data moat through deCODE Genetics. AI has yet to produce a visible impact on FDA approval data; the average of 48.5 new drugs approved annually over the past five years primarily reflects regulatory modernization and the rise of biologics. While AI has already contributed to molecular design, virtual screening, and target discovery, "curing all diseases" requires overcoming multiple systemic bottlenecks—a challenge that algorithmic iteration alone cannot solve.