Vijay Pande is moving from scale to focus. After more than a decade building a major healthcare and life sciences investing practice at a16z, he is now running VZVC with co-founder Zach Werner around a much smaller operating model.
The shift is not only about fund size or staffing. It reflects Pande’s view that AI in medicine is advancing, but that the hardest questions still come down to data, clinical trials, and whether founders can build companies that work in the market as well as in the lab.
From a16z scale to a smaller firm
Pande was once better known in academic settings than venture capital. He was a Stanford chemistry professor and was known for building Folding@home, a distributed-computing project that connected millions of home PCs into a supercomputer for disease research.
That changed a dozen years ago, when Marc Andreessen and Ben Horowitz moved a16z into healthcare and life sciences after previously avoiding the category. Pande led that effort, and over the next decade-plus it grew into a practice managing close to $4 billion.
In June of last year, Pande left that platform and started VZVC with Zach Werner. The new firm is intentionally small. On the investment side, Pande says it is essentially just the two co-founders.
The firm had considered hiring associates, but Pande says the agents they built made that unnecessary. That gives VZVC a structure that is meaningfully different from the larger platform he previously helped run.
Why VZVC is choosing concentration
VZVC is not built to place dozens of bets across a wide market. Pande said, "We’re [not] driving 30 bets per year," and described the target as "probably five" investments.
That concentration changes the way the firm thinks about each company. Pande compared adding a company at a typical fund to adding a Facebook friend. For him and Werner, the decision is much more serious: "wanting to have another child."
The model also changes competition. Pande says VZVC is usually not chasing a hot round in the same way a firm might compete for a hot Series A or Series B. Instead, he says people make room for them because they want the two partners involved and because of how hands-on they can be.
He pointed to Antonio Gracias at Valor, Thrive, and a16z as influences on how he thinks about building the firm. The common thread, as presented by Pande, is a more deliberate approach to involvement rather than simply maximizing the number of companies in a portfolio.
AI’s promise in biology depends on data
Pande argues that biology is moving away from a pure "science of discovery" and toward something closer to engineering. In his view, AI and machine learning can help computers reason through biological complexity, including drug targets, drug design, and clinical trials.
Clinical trials remain a major pressure point. Pande described synthetic data reducing the need for people in trials as an aspiration, not a solved reality. He said getting to clinical trials has become faster and cheaper, especially with AI, but running a trial can still cost hundreds of millions of dollars.
The risk is also severe. According to Pande, the probability of a drug moving successfully from the first trial through the third trial is just 20%. As he framed it, if 8 out of 10 fail and trials cost hundreds of millions of dollars, the broader cost burden becomes very high.
One reason for that failure, in Pande’s explanation, is that drugs are often designed from experiments involving animal models like mice. He said those models are not very predictive of humans. AI does not need to be perfect to matter, in his view; it needs to be better than the animal model benchmark.
Precision medicine and the limits of silos
Pande also connects AI to precision medicine. His point is simple: when a patient faces a serious condition, doctors often have limited information and may try one drug, then another, then another. The better outcome would be choosing the right drug first.
He also questions how medical data is usually interpreted. Blood test values are typically compared with population averages, but Pande says the more useful question is whether a result is unusual for that specific person.
That requires more than genomics. Pande said precision medicine was long based on genomics, but compared the genome to the blueprint for a house on day one. The body, like the house, changes over time. He pointed to proteomics and other measurements as more relevant for understanding disease and the body’s current state.
AI may also help with the fragmentation of medical expertise. Pande acknowledged the problem of different specialists not syncing well, using oncology and endocrinology as an example. In principle, he said, AI could become a specialist in everything and notice patterns that no single human could see.
The open question for AI biotech
The central constraint is data. Unlike text, biological data cannot simply be scraped from the internet. Pande said this makes biology unusual from a pure AI perspective because companies cannot all train on the same freely available data, and the data cannot simply be distilled from one model into another.
That creates walled-off datasets, but Pande also sees a broader shift toward atlases of biological information, typically foundation models from a technology standpoint. He expects open-source foundation models in biology to have a broad impact, comparing the possibility to open-source LLMs competing well against corporate models.
Still, he is careful about hype. Pande says AI can find insights humans cannot reach alone. But he is skeptical of sweeping claims that AI will cure everything, not because he doubts AI itself, but because he doubts whether the necessary data always exists.
That caution also shapes how he evaluates founders. Pande said he spends most of his time in two areas: AI for healthcare delivery and AI for clinical trials. He looks for founders with high integrity, long-term thinking, and a relationship that can last "5, 10 years plus" and ideally into their next company.
For Pande, technical brilliance is not enough. He said one lesson from investing is that even the most compelling technology comes back to go-to-market. For founders coming from science or product, he wants the same creativity applied to selling, adoption, and company-building, because the go-to-market side can be at least as hard as the technology itself.