Goldman Sachs Byte-ology: Quantifying AI's Value Creation in Drug Development
Summaries are public. PDF access requires an active membership; all members have the same access. Sign in
Goldman Sachs quantifies the value artificial intelligence creates in drug development by analysing 96 AI-designed drugs discovered by 234 specialised biotech firms. AI improves the overall probability of success from preclinical through Phase III by approximately 1.6 times, achieving 10.1% compared with a historical baseline of 6.4%. Phase I success rates reached 82% versus 70% historically, while Phase II came in at 44% versus 33%. Phase III sample sizes were insufficient, so historical rates of 27.5% were applied.
AI applications in drug discovery and clinical trial patient recruitment can reduce time to market by 20-25% and cut costs by 25-30%. The present value of improvements across time, cost, and success rates over ten years is estimated at $83-412 billion, using discount rates of 8-21%. This represents 7-20% of the present value of global pharmaceutical R&D spending. Based on value-of-statistical-life methodologies, higher success rates could generate $18 billion in present value within one year and $96 billion over seven years.
Large-cap biopharma companies are already demonstrating tangible benefits. Bristol-Myers Squibb quantified roughly $250 million in supply chain savings, while Pfizer identified over $700 million in annual operational impact. Novartis reported $10-15 million in savings per facility. Therapeutic areas are heavily concentrated in oncology at 54%, neurology at 16%, and immunology and inflammation at 14%. Small molecules account for 78% of the pipeline.