Andreessen Horowitz Makes Its Case That AI Will Turn Pharma Into an Even Bigger Business
Venture

Andreessen Horowitz Makes Its Case That AI Will Turn Pharma Into an Even Bigger Business

Andreessen Horowitz has published a detailed thesis arguing that artificial intelligence will make the pharmaceutical industry larger, not smaller. The essay, released on the firm's newsletter on November 13, contends that drugs already rival software in commercial scale and that AI could change how they are discovered and valued.

The core claim is simple: if AI improves the odds that a drug candidate succeeds and shortens the time it takes to reach approval, the economics of drug development shift substantially. Because the firm invests in drug-discovery startups, the argument doubles as an investment roadmap, a point Korean outlet Wowtale examined closely in its breakdown of the piece.

Why a16z Compares Drugs to Software

The essay opens with market size. The top 30 pharma companies are worth roughly $5 trillion together, and nine exceed $200 billion in value. Eli Lilly has passed the $1 trillion mark, largely on demand for its GLP-1 drugs Mounjaro and Zepbound.

According to the firm, Lilly's injectable GLP-1 portfolio generated nearly $40 billion in 2025, roughly comparable to Salesforce's annual revenue. About 29% of branded drugs, around 300 products, qualify as blockbusters with more than $1 billion in sales.

The business is highly concentrated, however. The top quarter of drugs produces about 85% of revenue, and GLP-1s accounted for 63% of Lilly's 2025 revenue while Keytruda represented 54% of Merck's.

Development Remains the Choke Point

Higher spending has not produced more medicines. U.S. prescription drug spending rose from $448 billion in 2016 to $915 billion in 2025, yet the FDA still approves only 40 to 50 novel drugs per year.

Fewer than 10% of drugs entering clinical trials win approval. Factoring in failures across the pipeline, a16z estimates the present-value cost of a single new drug at close to $2 billion and about 13 years of work.

How AI Could Shift Returns and Valuations

The firm points to two levers: a higher probability of success and a shorter path to approval. It sees AI contributing at nearly every stage, from selecting targets and screening molecules computationally to designing trials, recruiting patients and drafting regulatory filings.

In an illustrative model, cutting development cost from $1.6 billion to $800 million lifts the 10-year internal rate of return on a drug with $500 million in peak sales from 8% to 22%. That could make currently overlooked targets worth pursuing.

The firm also argues that public markets give little credit to early pipelines or discovery capabilities. If discovery becomes repeatable, investors might value a company's AI infrastructure as they value tech platforms, which a16z says is the rationale behind its Biotech Ecosystem Fund with Eli Lilly.

Startups, Partnerships and Fresh Capital

The essay describes bio-AI as nearing a "GPT-3 level" of market maturity, citing benchmark gains in antibody structure prediction and a growing list of pharma partnerships. Isomorphic Labs has deals with J&J, Novartis and Lilly, while Chai Discovery, Boltz and BigHat Bio each work with several large drugmakers.

a16z sees two paths for startups: selling data, models and tools to pharma and AI labs, with Veeva cited as precedent, or building full-stack biotechs that develop their own drugs. The second route demands more capital and risk but captures the full value of an asset.

Money is already flowing. Isomorphic Labs raised a $2.1 billion Series B in May, Chai Discovery raised $400 million in July, and Superluminal Medicines raised $60 million in September. Nvidia and Lilly have also launched a $1 billion joint AI drug-discovery lab.

Caveats Worth Weighing

The firm has a direct stake in the outcome. It has invested in life sciences since 2015, counts insitro, Formation Bio and BigHat Bio in its portfolio, and co-runs the Lilly fund.

Some comparisons are loose. The group of 16 software companies worth over $200 billion includes Apple, Amazon, Tesla and Samsung, and a single drug's peak sales are compared with entire companies' annual revenue. a16z itself labels the cost-halving scenario an illustration rather than a forecast.

Scientific hurdles remain. The Protein Data Bank holds only about 300,000 structures, far from internet scale, and better molecular design does not guarantee an effective drug if the wrong biological mechanism is targeted.

What the Thesis Means for Founders and Investors

For founders, the essay signals strong investor appetite for both AI tooling and AI-native drug developers, alongside rising expectations for proof. The firm frames the real test as whether model improvements translate into better drugs entering the clinic.

Large funding rounds and pharma partnerships suggest the bet is being taken seriously. Whether it pays off will ultimately be determined by clinical results rather than benchmarks.

Source: Wowtale

Topics:

JeffkomStory Team

Editorial, JeffKom Story

The JeffKom Story newsroom covers startups, founders, funding and the technology shaping what’s next.

Keep reading

Jensen Huang's Nvidia Exit Revives Debate Over Dual-Class Shares in Big Tech
Technology

Jensen Huang's Nvidia Exit Revives Debate Over Dual-Class Shares in Big Tech

More in Venture