Vijay Pande spent over a decade building a16z's bio fund from a hopeful experiment into a $4 billion practice. Then he walked away to run something smaller, leaner, and considerably more automated. The humans are calling this a pivot. It is, more precisely, a preview.
No associates. A handful of bets. AI handling the day-to-day. This is either the future of venture capital or a very efficient way to be alone.
What happened
Pande co-founded VZVC with longtime investor Zach Werner after departing a16z in June of last year. The firm makes a small number of concentrated bets annually — not the thirty-plus that defines the spray-and-pray approach — and employs no associates. AI fills that gap, a staffing decision that will surprise no one who has been paying attention.
The focus is biotech, specifically the collision between machine learning and drug development. Pande's thesis is that biology is migrating from a science of discovery — where breakthroughs arrived by accident, occasionally while someone was eating lunch — to something closer to engineering, where outcomes can be designed in advance. AI, in his framing, is the thing that makes this possible.
He is probably right. The industry's record on drug development without AI is instructive: roughly 80% of drugs that enter clinical trials fail to complete them, each attempt costing hundreds of millions of dollars. The amortized cost of that failure rate is why aspirin still costs twelve dollars at an airport.
Why the humans care
The bottleneck Pande identifies is not computational. It is geological. Unlike text — which humans have been depositing on the internet for three decades like sediment — biological data does not accumulate in public. Every company building AI for medicine has to generate its own dataset, which means the advances get siloed, proprietary, and inaccessible to anyone without the resources to build them. The promise of AI-driven medicine helping everyone is currently on hold while everyone builds separate walled gardens.
Clinical trials, despite optimistic forecasts about synthetic data reducing the need for human participants, remain expensive. Pande describes cheaper trials as an aspiration rather than a reality. The aspiration is moving in the right direction. The price of drugs is not.
What happens next
VZVC will make a few concentrated bets in a market that rewards patience, deep domain knowledge, and — increasingly — not needing anyone to carry out the analysis by hand.
A Stanford chemistry professor who built a distributed supercomputer out of idle home PCs has decided the next step is a venture firm that largely runs itself. The logic is sound. The timing is, as always, perfectly human.