Bridging Science, Medicine, and Returns
What this covers
Tal Zaks speaks from inside Moderna's COVID-19 vaccine effort, and the conversation moves between drug development, venture finance and public policy. He starts with the economics. Biotech venture investors aim for 3 to 4x, with 5x a great outcome, so they cannot be wrong as often as tech investors chasing 1000-to-1 upside. Part of the difficulty is the Biology and Pharmacology Risk Split: whether a protein really drives disease, and whether a drug can safely change it. He uses the Hoosac Tunnel story, and Albert Hirschman's argument about useful naivete, to explain why projects always prove harder than expected. He also treats the missing antibiotics as a failure of commercial incentive, not of science.
The middle of the conversation is the Moderna story. It covers a September 2019 meeting with Anthony Fauci, trial sites placed where COVID hotspots were predicted, and a phase three event count reached in three months. Zaks also defends the safety record, including the myocarditis signal in young adults. Other threads include a personalized mRNA cancer vaccine that cut skin cancer recurrence by about 50 percent in a phase two study, and a plan to use a lipid nanoparticle to deliver a corrective gene once. He also notes that mRNA can in principle be designed to alter DNA, which raises ethical questions.
The remaining ground is more diffuse. Zaks calls AI a commoditizer of discovery but not of clinical development, though machine-learned control arm predictions may shrink trials. He cites pathology-reading software that goes largely unused for lack of return on investment. He describes jargon in Multidisciplinary Silo Barrier terms, and mentions Amos Tanay's work suggesting that half the variance in normal blood values is false.
Tal Zaks argues that biotech returns depend on hard-to-predict biology and pharmacology, so progress requires multidisciplinary execution, durable return incentives, and public trust, and AI will help most in discovery and least in clinical development.
- Biotech returns hinge on capital required, time to value inflection, and probability of success, and biology and pharmacology risks are very hard to predict.
- AI is commoditizing drug discovery but clinical development still requires putting drugs into people, because there is no holistic model of a human.
- Durable progress depends on return on investment incentives, as shown by the antibiotic gap, and on open public dialogue rather than scientific authority alone.
Therapeutics returns require domain expertise and concentrated conviction, not venture-style power-law bets.
- Investors without formal domain training can earn strong returns in therapeutics by recognizing the multidisciplinary nature of the field and finding the right experts, since experience can be partly replaced by wisdom but only up to a point.
“you can replace experience with wisdom up to a point. Beyond that, you better go and ask somebody.”
- Zaks argued that executives' time is more valuable to them than capital, so his personal sweat equity at Moderna was his strongest signal of conviction, and he wants his capital to follow the sweat equity of people he believes in.
“It is my time on earth. ... that's how I believe in this company. And so it taught me the primacy of sweat equity”
- Therapeutics investment decisions turn on the thesis the team is trying to prove, the capital needed to reach the next value inflection point, the likelihood of reaching it, and the talent around the table, and early phase one data may not de-risk a vaccine enough to justify capital.
“the phase one data are probably not going to be de-risking. You're going to have to take it all the way to phase three to know”
- Scott Kupor's account of Andreessen Horowitz's model explains how tech VC can tolerate winning only one in a hundred because returns can reach 1000x, a model that does not transfer to therapeutics.
“Scott Kupor's The Secret of Sand Hill Road where Andreessen Horowitz talks about what it takes to be a VC investor in tech.”
Medicine must shift from sick care toward prevention, despite unpredictable biology and permanent gene-modification risks.
- Biology is not a system humans designed, so researchers are constantly learning about it, unlike engineers who build the systems they work on, which makes outcomes harder to predict.
“It's not a system we designed, so we're constantly learning about it versus engineers.”
- Permanent gene modification raises concerns about effects on future generations and unintended consequences for offspring, which will require new moral questions and debate to be resolved before it is widely accepted.
“There's a concern for what this means for future generations and making sure that we're doing it for the individual and not with unintended consequences for their offspring”
- The current system is a sick care system rather than a health care system, because interventions that detect and address abnormalities depend on drug development that has not yet caught up with earlier detection.
“that's where I agree with the folks that have been saying that we've got a 'sick care system' as opposed to a 'health care system.'”
- Vaccines take healthy people and make them healthier in the future, which challenges the traditional physician's focus on treating the sick, and Zaks argued this preventive shift is the direction medicine must take.
“But actually that's what vaccines do and that's what this new age of technology is going to enable us to do.”
Government backstops and reimbursement, not private capital alone, sustain medical innovation.
- The COVID vaccine manufacturing scale-up was financed by Moderna raising private capital in public markets, with government later backstopping purchases of doses rather than funding the investment directly.
“No, it was private capital. ... what the government did at a certain time point was backstop the investment and said, okay, we'll pay for these doses.”
- Moderna's pre-COVID relationship with the NIH and BARDA, built during the 2017 Zika vaccine effort, made government aware of the platform's speed and helped set up the 2019 NIH demonstration project.
“the government got to know us, BARDA and the NIH during Zika”
- The failure to develop new antibiotics is not due to a lack of scientific tools but to a lack of commercial incentive, so infection mortality is rising, and policy must reimburse these efforts or the innovation engine stalls.
“it's not a lack of scientific tools to understand bacteria and how to make better drugs. There's just a complete lack of commercial incentive to do so.”
- Steve Rosenberg's NCI lab produced an mRNA cancer vaccine in mice in the late 1990s that was published in 1998, and twenty years later Moderna's personalized cancer vaccine built on that foundation after Rosenberg was invited to collaborate.
“we tried in his lab to make an mRNA vaccine for cancer. We did it in mice. We published it back in 1998”
Ambitious naivete and well-funded flexibility drive early-stage company success.
- Early-stage investments fail when they are linear and underfunded, so successful companies need enough capital, team functions, and strategic flexibility to adjust, which Zaks learned from a failed small-company investment and from Moderna's parallel exploration of mRNA applications.
“this was probably not well enough funded and not given enough opportunity... to have that degrees and freedom to adjust.”
- Stephane Bancel's management style probes each function with repeated why questions and requires answers in plain English, so that a mentor's test that you cannot explain something to a kindergartner means you do not understand it.
“just ask the five whys, why is this, why is this, why is this?”
- Economist Albert Hirschman argued that economic progress is driven by the naivete of underestimating how hard projects will be, which is useful because excitement lets people start, and once committed they push through, as in the Hoosac Tunnel.
“His whole economic thesis was that what drives economic progress is the naivete of thinking how easy it's going to be.”