Artificial intelligence could reduce the roughly $2 billion cost of developing a successful drug by improving success rates and shortening approval times, Andreessen Horowitz said in a newsletter published Nov. 13.
The venture capital firm, which invests in life sciences startups, said development takes an average of 13 years, with failures adding to the cost of successful medicines. Its newsletter, “Big Pharma Can Rival Big Software,” argued that AI could change those economics, allowing companies to pursue treatments that would otherwise cost too much to develop.
Andreessen Horowitz said its analysis compared pharmaceutical and software businesses and examined drug sales, development costs and potential uses for AI, drawing partly on 2025 revenue figures. It gave no single study period. The firm said it has invested at the intersection of technology and biology since 2015, backing insitro, Formation Bio, BigHat Bio and more than 75 other life sciences startups.
The firm said fewer than 10% of drugs entering human trials ultimately win approval, while fewer than 5% make it from the broader drug-candidate stage to U.S. approval. It put annual approvals of novel medicines by the U.S. Food and Drug Administration at about 40 to 50, despite prescription drug spending more than doubling over the past decade.
Large pharmaceutical companies obtain about two-thirds of new drugs through acquisitions or licensing, the firm said.
Andreessen Horowitz said AI could help companies select biological targets, choose promising molecules, design trials, recruit patients and draft regulatory submissions. Better success rates and shorter development times could reduce costs after accounting for the risk of failure.
The firm qualified that prospect by saying some timelines remain subject to regulatory requirements, including long-term toxicity studies. Drug development also requires extensive evidence from physical experiments.
The firm said successful drugs can produce years of protected sales, with a typical medicine retaining patent exclusivity for 10 to 14 years after launch. It estimated that the top quarter of drugs generate about 85% of drug revenue.
For the 10 largest pharmaceutical companies, about half of revenue comes from their three biggest products. Those concentrations make commercial results depend heavily on a small number of successes.
Andreessen Horowitz valued the 30 largest pharmaceutical companies at roughly $5 trillion combined and counted nine listed drugmakers worth more than $200 billion, compared with 16 software companies. The firm said AI could support both suppliers of software, models and data and companies developing medicines themselves. Lower development costs could make some previously uneconomic treatments viable and increase returns on research spending.
