Vijay Pande, long known in academic circles as a Stanford chemistry professor and the creator of the Folding@home distributed computing project, spent more than a decade building Andreessen Horowitz’s life‑sciences practice. Under his leadership that practice grew to manage nearly $4 billion.
In June of last year Pande left a16z and launched a much smaller firm. The new company, VZVC, which he co‑founded with longtime investor Zach Werner, deliberately pursues a small number of concentrated bets each year, operates without junior associates, and relies heavily on AI for day‑to‑day work.
Why a concentrated portfolio?
Pande argues that in the current market it makes sense to focus on a few high‑conviction investments rather than many smaller ones. VZVC’s plan is not to place dozens of bets annually — instead they expect to make roughly five investments per year, a highly selective approach. For Pande and Werner each new investment is a major commitment and they emphasize long‑term, hands‑on partnerships with founders who exhibit integrity and long‑term thinking.
The role of AI in drug development
Pande says biology is shifting from a ‘‘science of discovery’’ toward an engineering discipline: AI and machine learning can wrap computational understanding around extremely complex systems to pick drug targets, design molecules and even improve clinical trials — the most expensive stage of drug development. While AI and synthetic data can reduce some costs, clinical trials can still run into the hundreds of millions of dollars, and historically only about 20% of drug candidates make it successfully from phase I through phase III.
A major reason for failures is that early experiments are often performed in animal models like mice, which frequently do not predict human response well. AI won’t be perfect, Pande says, but it can surpass animal models in predictive power, and crossing that threshold would be transformative.
Precision and personalized medicine
Pande distinguishes precision medicine from crude population‑level comparisons: instead of comparing a patient’s lab values to population averages, the goal is to understand whether those values are abnormal for that specific individual. Genomics provides a baseline blueprint, but proteomics and other measurements better reflect the body’s current state. Robotic, automated measurement systems coupled with AI are particularly powerful for making these assessments.
Data limitations and siloed datasets
A critical constraint in applying AI to biology is that biological data cannot be ‘‘scraped’’ en masse from the public web; companies typically build proprietary, walled‑off datasets. That reduces the possibility of one universal model trained on shared data and makes model transfer harder. Pande notes, however, that there is momentum toward building biological atlases or foundation models; if open‑source foundation models take hold in biology as they have in language models, they could have broad impact.
Related companies and investment focus
Pande remains involved with companies that spun out of his Stanford lab, such as Genesis Therapeutics, and with Insitro, founded by former Stanford colleague Daphne Koller. VZVC’s main investment themes are AI for healthcare delivery and AI for improving clinical trials. When evaluating founders he prioritizes trustworthiness, integrity and founders who are willing to build long‑term partnerships rather than simply competing to beat others.
Lessons learned
Pande says he faced skepticism more than a decade ago when he advocated for machine learning in biology and medicine; much of that resistance has diminished. He also stresses that go‑to‑market execution is often as hard or harder than the underlying technology, and scientists and product founders need to apply their creativity to commercialization as well.
How VZVC operates
The firm’s name combines the founders’ first names (Vijay and Zach). VZVC is intentionally small and flat: investment decisions are made primarily by Pande and Werner, and built‑in automation (agents and AI tools) reduced the initial need to hire associates. Their focused model also means they rarely compete for the hottest rounds — founders often make room for them because they value the founders’ hands‑on support.
What’s overhyped?
Pande cautions against claims that AI will ‘‘solve everything.’’ AI can uncover insights beyond human reach, but it is only as powerful as the data it learns from. When high‑quality, comprehensive biological data are lacking, AI cannot miraculously overcome that deficit.
In short, Pande has moved from leading a large, broad bio fund at a16z to running a small, highly concentrated firm that leverages AI for drug discovery and clinical development. He sees real potential for AI to improve target selection, chemistry and trial design, but also highlights persistent challenges: siloed data, expensive clinical testing, and the crucial importance of strong go‑to‑market execution.
