Vijay Pande Is Betting Small: A Few Concentrated Bets a Year and an AI-Native Fund
The architect of a16z's $4 billion bio portfolio left to build VZVC — a lean, AI-run fund built around a handful of concentrated bets and a provocative thesis about why biology can't be scraped.
Published: 2026-08-31 Category: Quick Take Sources: TechCrunch
The Hard Pivot
Vijay Pande was once best known as the Stanford chemistry professor behind Folding@home, the distributed-computing project that turned millions of home PCs into a supercomputer for disease research. Then, a dozen years ago, Andreessen Horowitz handed him the keys to a healthcare and life-sciences practice he grew to manage close to $4 billion.
So it was a surprise when, in June of last year, Pande walked away to start something much smaller. His new firm, VZVC, co-founded with longtime investor Zach Werner, is built around a handful of concentrated bets a year rather than dozens. It has no associates, and it relies heavily on AI for its day-to-day operations.
Biology Is Moving From Discovery to Engineering
Pande's core thesis is that biology is shifting from a "science of discovery" to something you can engineer. With AI, computers can wrap their understanding around something very complicated — identifying drug targets for specific diseases, making those drugs, and even helping in clinical trials, the most expensive stage of the pipeline.
He's candid that the economics are still brutal: the probability a drug moves successfully from first trial to third is just 20%. Eight in ten fail, and each can cost hundreds of millions of dollars, which is why drugs are so expensive. The typical failure causes, he argues, isn't bad biology — it's that drugs were designed on animal models like mice, which are poor predictors of humans. An AI model won't be perfect, but it will beat any animal model, and once it crosses that bar, "that's where it gets really exciting."
The Data Problem No One Can Scrape
The most provocative thread is about data. Unlike text, biological data can't be scraped off the internet. That means nearly every company ends up building its own walled-off dataset — and it's a place where one model's data can't be distilled into another's. This echoes medicine's familiar silo problem: an oncologist and an endocrinologist treating the same patient rarely sync up well.
Pande's answer is a bet on shared infrastructure: a shift toward building "atlases of biological information," typically foundation models. As these become more common, he expects the same dynamic that played out with open-source LLMs — open-source foundation models in biology having broad impact against corporate ones.
Analysis
The most interesting thing about Pande's pivot is what it signals about how elite investors are reorienting — not just toward AI, but to being operationally reshaped by it. A fund with no associates that leans heavily on AI to run daily operations is a pointed rejection of the bloated, spray-and-pray model that dominated the last decade. Concentrating on a handful of long-horizon, high-integrity founders (he calls for five-to-ten-year relationships) is a bet that judgment and trust, not volume, will compound in a market that's otherwise saturated with capital chasing the same AI names.
His biological "walled garden" thesis is worth sitting with. If data can't be scraped and can't be cheaply shared, the competitive moat in AI-driven biotech is whoever can accumulate proprietary, hard-to-replicate datasets — favoring incumbents with real-world clinical and manufacturing access. That's a structural advantage no amount of frontier-model compute can erode, and it reframes the "AI will crush drug discovery" narrative into something more sober: the winners will be those who own the data, not just the models.
Source: TechCrunch — "'We're not doing 30 bets a year': Vijay Pande on betting small after running $4 billion at a16z" (Aug 29, 2026).