How Does Life Become Worth Investing In?
On May 1st 2025 Meta and Microsoft reported high earnings and added ~$360 billion in market cap, as much as the value of the pharma companies Pfizer, Amgen, and GlaxoSmithKline combined. The biotech index XBI was flat that day. The next earnings report on August 1st resulted in ~$440 billion added market cap, as much as Bristol-Myers Squibb, AstraZeneca, and Sanofi combined. The biotech index XBI index was flat that day. AI software advances are clearly a massive deal. Nevertheless, it’s surprising that they completely outclass innovation that determines life and death.
This piece explores why biomedicine, and longevity in particular, is not a booming industry and how that might change. We set the stage with a primer on how medicines are made: generally, ideas for how to treat disease come from a university research group or sometimes existing companies. Turning the idea into a way to affect human bodies, e.g. a molecule in a pill, takes a few years and more than ten million dollars. Unlike most businesses, the company is not allowed to sell these pills to patients. First, they have to work with the healthcare system to run controlled tests in patients (‘clinical trials’) for a number of years, which can cost more than a hundred million dollars. If the pill works well in these trials (most don’t), it can be sold (in a complicated way via doctors and pharmacies) to a precisely defined group of patients. Targeting aging to improve health takes all of these dynamics to the extreme: clinical studies take longer, the probability of success is unknown, and markets are theoretically huge but untested in practice.
A feature of every booming industry is productive specialization, where different organizations get good at parts of a larger, valuable production. Improvements from any part benefit the whole. New ideas can also open up new markets, using the tools developed in the industry in new ways. This dynamic happens often when the barrier to entry is low, as for software written by a tiny team and offered for sale to the industry. But it can work for capital-intensive industries too, as long as future profits can be predicted reliably. However, the dynamics of biotech disincentivize specialized collaboration. Let’s contrast space and biotech, both capital-intensive:
Space: Company A makes communications satellites. They sign a $50M deal to provide wifi in rural Alaska. They spend $10M building small satellites and pay rocket company B another $10M to launch them (and get their costs back if the rocket explodes). Assuming their satellite works, Company A gets to fulfill their contract without learning to build rockets, while Company B gets paid right away using revenue from a customer base they weren’t serving. Customers get something they didn’t have before, and both companies become more valuable. The more companies come up with useful reasons to put things in space, the more B’s market expands.
Biotech: Company C has data that a certain protein improves heart failure in mice. They want to raise $50M to run clinical trials to get proof in humans, after which they can raise more money to get the drug to approval. These trials will take >5 years, after which they may have lots of profits. Company D has a way to target proteins to the heart better, which they claim would improve the odds of success for Company C’s trials. They suggest that Company C pays them $20M to license this, but C does not have this money because they don’t have revenue before the drug is approved. They ask investors for $70M instead, but their competitor E is only asking for $50M. C tries offering D royalties on future sales instead, but D needs cash now to pay their employees. Ultimately, C, D, and E are distributing the same pie of revenue if drug gets approved.
This type of problem should be solvable by borrowing on future gains. If the protein targeting technology really improves success rates and thus expected value, it should be worth investment today. Even if the company doesn’t have the money, professional underwriters exist to assume long-term risks with positive expected value. But this model relies on being able to estimate the future expected value. High uncertainty scenarios turn low-margin-but-net-positive underwriting into speculative investment. The core event triggering profits in biotech, clinical trials, is necessarily attempting something new and happen on average once or twice per company, so relevant feedback is scarce. To illustrate how this further disincentivizes productive specialization, let’s revisit our space and biotech companies:
Space: Company A from the example above doesn’t actually get the $50M contract right away, so they don’t have enough money to build satellites and pay for the launch. But because the future revenue is clear, underwriters are happy to bridge the gap. The remaining risk is that A fails to build a usable satellite or B fails to launch. Usually there is a track record for these, and success rates are very high. The risk of failure can be accounted for in financing terms. This arrangement allows progress that otherwise would not have gotten off the ground.
Bio: For most biotech companies, a successful trial means large profits while failure means the end of the company. The average trial has a ~10% chance of success, and improving that number would be amazing. The problem is that the trial either succeeds or fails based on many factors (C had the right disease understanding, found the right patients, picked the right drug dose, etc). Unlike rockets and satellites, we can’t independently test which part succeeded or failed, so determining the effect of D’s technology would take many such trials, each long and expensive. C and others need to commit to D’s technology before they lock in trial design with the FDA, without clear proof of benefit. Because every company faces this dilemma, it’s difficult for D to generate evidence about their technology’s impact. In most cases, companies like D end up using their technology for making therapies and running their own trials.
Feedback feeds financing. When there are no good signals to gauge ROI, financing comes from discretionary spending rather than core investment funds. This is evident in the current state of the longevity field: A few highly capitalized companies are funded by individuals who believe in a long-term mission. An increasing number of pharmaceutical companies have research groups focused on aging biology, but they are small and exploratory. A large fraction of funding and progress comes from non-profit organizations, compared to biotechnology overall (and to other fields).
What would enable longevity to become a true industry, with an accelerating rate of progress fueled by productive specialization? From the above, we see two options:
The first is political: Make feedback faster by reducing the burden of running clinical trials, which would mean cheaper and more abundant feedback for companies trying to improve the work of others. This could be done responsibly, e.g. by allowing drugs to go on the market earlier but monitoring their effects more closely when they do. However, there are limits to this option - taken to the extreme, e.g. eliminating clinical trials, would eliminate incentives to make drugs actually work and create an industry of false marketing of ineffective products.
The second is more technological: Make feedback better by improving our ability to predict eventual outcomes, which would let transactions around future events happen confidently. The threshold that would unlock new financial models isn’t obvious, but even before that happens, individual trials would have higher success rates. And there is no limit to the benefits from improving this metric.
These are not alien ideas - biomedicine in general would like faster time to market and better predictions, and attempts these in various ways. But for longevity there’s a stronger imperative: assumed trial success rates are 0%, and proving otherwise takes more time and money. And there are actionable opportunities: everyone ages, both humans and animals. We can build predictors in short-lived model organisms while incorporating data collected from human blood samples. We can create public benchmarking of predictors, to incentivize what works robustly over what can be published. And anything that really works can be scaled through existing consumer health offerings. It’s possible to work towards the first predictive measures of longevity, and if we want this industry to boom, it’s the most important thing to focus on.
