Jul 15, 2017 · 13m · a16z
On Machine Learning in Medicine and More
0:00 / 0:00
⌖ your search result is the highlighted band (5:05–5:38). Playback starts there
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this interview hosted by David Clark, Andreessen Horowitz General Partner Vijay Pande outlines the convergence of computer science, machine learning, and biology. He explains how software-driven business models, cross-disciplinary founders, and digital health platforms are disrupting traditional biotechnology and transforming modern healthcare delivery.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The host as informed peer 3.3 Guest teaching 3.0 Guest disagreement 1.5 The host pushing back 2.3
speaking balance: gold is the host, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 0:27 Rejection of traditional biotech model
Vijay explicitly pushes back against standard biotech investing, criticizing its high risk, heavy capital requirements, and stochastic outcomes compared to software.
Hardest push from the host ▶ 10:17 Multiples and fund return math challengeDavid Clark challenges Vijay's thesis by citing empirical database figures showing traditional life science returns cap at 10X, questioning how fund-level economics work.
Biggest teaching moment ▶ 5:07 Hybrid founders eliminating unknown unknownsVijay educates the host on how single founders proficient in both biology and computer science eliminate communication silos and unknown unknowns.
The host holds their own ▶ 10:17 Citing database evidence on return capsHost David Clark demonstrates deep industry expertise by citing hard database stats on the return limits of traditional life science venture investments.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The host as informed peer | Guest teaching | Guest disagreement | The host pushing back | Why |
|---|---|---|---|---|---|---|
| Differentiating Bio-Software from Traditional Biotech | 2 | 3 | 2 | 1 | David Clark asks conversational intro questions while Vijay Pande distinguishes his software-driven bio fund from traditional high-risk biotech. Vijay details the investment categories including digital therapeutics and computational medicine. | |
| Timing and Technological Drivers in Healthcare Innovation | 3 | 3 | 1 | 2 | David Clark pushes gently on historical VC underperformance in healthcare tech to ask what has changed. Vijay explains the convergence of mobile, machine learning, and hybrid founders who master both biology and code. | |
| Navigating Inflexible Incumbents and Working Within the System | 3 | 4 | 2 | 3 | David Clark asks how new entrants can disrupt an industry filled with lucrative vested interests. Vijay responds with a self-driving car analogy, explaining why companies must work within existing systems and reframe regulatory obstacles as competitive moats. | |
| Expansion of Exit Horizons and M&A Opportunities | 5 | 2 | 1 | 3 | David Clark demonstrates VC domain knowledge by citing fund database metrics showing life science returns topping out around 10X versus tech's 100X+. Vijay asserts that bio-software companies deserve software-like return multiples because of their high margins and software core. |
Statements from this episode (16)
Pande: Traditional biotech requires massive capital and yields highly stochastic outcomes
“So biotech has a lot of things I don't like. It has a lot of risk. It has a requirement of a lot of money before you get to some answer, and the answer is very stochastic.”
Pande: Software enables investors to incrementally de-risk investments at every stage
“What I love about software is that software, we can de-risk at every stage of investment, and we can see greater and greater revenue and help the company get bigger and bigger.”
Pande: Computational biology tasks requiring global computing power in 2000 run easily on AWS today
“What we could do 15 years ago required almost all the computers on the planet. Now that's something that we can do easily on something like Amazon.”
Pande: Type 2 diabetes and PTSD are behavioral problems needing non-pill solutions
“So the area of digital therapeutics lies in the fact that for many areas, such as type two diabetes, anxiety, depression, PTSD, I think our go-to for cures for this is a pill, but actually these are behavioral problems, and it's much more natural to have a beh…”
Pande: Freenome's genomic ML will drive cancer cure rates to 80-100%
“So a company like Freenome will use machine learning and deep learning with genomics to be able to tell whether you have cancer early enough, such that the cure rate, success rate could be 80%, a hundred percent.”
Pande: Compute will enable university-developed behavioral therapies to scale
“There are programs for all these behavioral issues, whether it's type two diabetes, or anxiety, or PTSD, and the intriguing thing is these programs work, often using things like cognitive behavioral therapy. The challenge is that they're often sitting in unive…”
Pande: All a16z bio portfolio founders possess dual CS and biology fluency
“And I would say all of the investments I've made have founders of this character that I could have a deep discussion with them in the computer science as well as the biology.”
Pande: Health tech startups must operate within existing healthcare systems
“Similarly, we could toss out the healthcare system and maybe start fresh and it, and that'd be an intriguing sort of, ah, thought discussion. But in reality, we can't do that either. And so what we're gonna see is people working within the system.”
Pande: Software as a drug will revolutionize medicine like biologics
“You think about, like, Genentech in the early days of biologics. In the early days there, people may have said, you know, a protein can't be a drug. But now that's something that's well recognized. I think some people might say, software can't be a drug. But s…”
Pande: Early cooperation with regulators creates competitive barriers to entry
“One of the things that I feel is that people often think regulation is a challenge and in some ways it is, but it's also a huge opportunity. In that if, especially in this new space, you can work with regulators and get there first. You can actually have an op…”
Pande: Digital health startups could be acquired by pharma or tech giants
“I could see a Pfizer buying Omada, you know, and maybe coupling with a drug or maybe not. I could see Google buying a company like Omada. I could see Freenome being bought either by a therapeutics company that wants a diagnostic, or by, let's say, a sequencing…”
Pande: Alphabet restructuring places medical subsidiaries alongside core search
“If you look at the alphabet restructuring, it puts Verily and Calico and these other elements that are very much medical centric, much more in the forefront. And search and the core business is just another one of those verticals.”
Clark: Traditional life science investments consistently cap out at 10x
“One of the things we've seen with traditional life science investing, when we look at our database, is that there's very much a limit on the sort of multiples that you can achieve on a single investment. The best we see consistently is around the 10 X, and it'…”
Pande: Bio-IT return multiples will look like software multiples
“The return multiples in this space I think will look like software multiples because these are software companies. These are companies that have software in their heart, they have high margins, they have all the things we like about software, we de-risk them t…”
Pande: Freenome's test will cost about as much as a dental visit
“Freenome's test that's something that you could do once a year, you could do twice a year, it's comparable to the cost of the dentist, and something where one could catch cancer immediately, like catching a cavity at a very early stage, and it gets cured, and …”