Industry Research No. 11. Stepping away from compute, let's look at a completely different logic — innovative drugs. Its allure and danger both come from the same underlying structure: the power law.
1. First, Understand the Industry's Foundation: The Power Law
If you approach innovative drugs with the same playbook you use for other industries, its unique structure will humble you fast. So let's start with its foundation: Innovative drugs is an industry that runs on a power law distribution.
What does that mean? In most industries, outcomes for companies are relatively continuous — do well, earn more; do poorly, earn less; big swings are rare. But not in pharma. Outcomes here are extremely binary and extremely unequal:
- A drug's R&D is highly likely to fail (clinical trial success rates overall are very low — nine out of ten fail). When it fails, the hundreds of millions or billions spent are basically zeroed out.
- But the rare successes (blockbusters) produce astronomical returns — a successful blockbuster can generate tens or even hundreds of billions in annual sales, more than covering all the failures.
That's the power law: Most pipelines contribute nothing; a tiny number of winners capture nearly all the industry's returns. A pharma company's value often hinges on one or two blockbuster drugs; a biotech portfolio's returns depend on a very small number of big winners.
This structure determines everything about innovative drugs — its R&D logic, its valuation methods, and most importantly, how it should be bet on. Only by understanding 'power law' will you avoid using the wrong posture in this sector.
2. In a Power Law World, the Biggest Temptation Is Also the Biggest Trap
The power law distribution creates a direct, fatal temptation — picking that one winner.
Since the rare winners take everything, what if you could predict which clinical candidate will succeed and get rich overnight? Countless people turn pharma investing into gambling on a single clinical trial outcome: betting on the Phase 3 data for a particular drug. Guess right — the stock might double or triple overnight. Guess wrong (data miss, clinical failure) — the stock might drop 50% or more overnight.
This is the biggest trap in pharma investing. Because — betting on a single clinical readout is essentially placing a large bet on a binary event whose outcome you can barely predict. Even the most experienced scientists and management teams can't know the result in advance (otherwise they wouldn't run the trial). How can a retail investor predict it? This 'pick the winner' gamble looks tempting, but it's using a heavy position to bet on a near-coin-flip binary event.
This violates the iron rule I've repeatedly stressed (recall the asset allocation leverage article, Taleb's ergodicity): Never bet in a way that a single mistake can take you out of the game. Betting on a single clinical readout is exactly that: one wrong bet can permanently wipe out a big chunk of your capital. The biggest temptation of the power law world (picking that one winner) is precisely the posture most likely to get you eliminated.
So what is the right posture in a power law world?
3. The Right Posture: Barbell, Not Single Bet
My core framework for betting on innovative drugs is the barbell — the structure I discussed in the asset allocation framework article, and it finds its most fitting application here in pharma.
Barbell means: allocate to the two extremes, skip the middle. Applied to innovative drugs —
One end of the barbell (safe end): mature, commercialized pharma companies with real cash flows. These companies already have approved blockbusters that are selling and generating profits. Their value no longer depends on 'betting on a clinical outcome' but on proven sales. Their returns won't double overnight, but they won't go to zero on a single clinical failure. This is the 'stable' end of the portfolio.
The other end (aggressive end): small, diversified holdings of early-stage, high-beta names. Here, you embrace the power law — but using 'small positions + diversification,' not 'heavy position + single bet.' You're not betting on 'one specific winner' but on 'a basket where a few will win.' Each one could go to zero (downside limited because position sizes are small), but if one or two become blockbusters, their 10x-20x returns will cover all the zeros (massive upside). This is using the power law against the power law — it's essentially a portfolio of convex options, and the correct posture for convexity is always small, diversified, and affordable to lose (remember the Bitcoin and space articles).
The middle to skip: heavy position in a single unapproved drug. The barbell skips the most tempting middle ground — putting big money into one or two pre-approval candidates. This position has neither the certainty of the safe end nor the 'diversification' protection of the aggressive end. It's 'heavy bet on a binary outcome' — the worst posture in a power law world.
With the barbell, you enjoy the extreme upside of pharma's power law (the convexity of the aggressive end) without being wiped out by any single failure (safety net from the safe end + diversification on the aggressive end). This is the only rational structure I see for individual investors facing this sector.
4. Two Relatively High-Conviction Themes
Within the barbell framework, two themes stand out with higher conviction, worth focusing your research on.
Theme One: The GLP-1 and Metabolic Revolution. This is the biggest, most certain wave in innovative drugs right now — the GLP-1-based weight loss/metabolic drugs (dominated by Eli Lilly and Novo Nordisk) are opening an unprecedented massive market. What makes it special: it addresses not some niche disease, but the huge population of hundreds of millions affected by obesity/metabolism, with efficacy repeatedly proven and commercialization already generating real revenue. This makes it no longer a binary 'clinical trial' bet — it has real, huge, and rapidly growing sales support, placing it closer to the safe end of the barbell. Around GLP-1, a whole value chain is emerging (including telehealth distributors like HIMS, which I've written about, that help distribute these drugs). GLP-1 is one of the rare themes in pharma with 'high certainty, huge market, and already commercialized.'
Theme Two: The Patent Cliff and M&A Wave. This is a structural, time-driven theme. A batch of blockbuster drugs at large pharma companies will face patent expirations in the next few years (the patent cliff) — once patents expire, generics rush in, and original drug revenue plummets. This forces big pharma to acquire innovation externally to fill the looming revenue gap. The result is a predictable M&A wave: biotech companies with high-quality innovative assets (especially clinical-stage candidates with blockbuster potential) become acquisition targets for big pharma. The investment logic of this theme: those biotechs with truly differentiated assets, even if they don't make it to market on their own, could be acquired at a premium — adding an 'acquisition exit' protection to the aggressive end.
(As for AI-driven drug discovery — using AI to accelerate drug discovery and reduce failure rates — this is a longer-term, more imaginative, but less validated theme. If it works, it could change the failure economics of pharma's power law itself. But right now it's too early; I classify it as 'worthy of tracking but not yet a bet.')
5. Final Thoughts
Innovative drugs is a charming and brutal industry that runs on a power law. Its charm lies in the astronomical returns of the rare blockbusters; its brutality lies in the zeroing out of most pipelines.
And the human instinct — 'pick that one winner' — is precisely the most dangerous reaction to a power law. It turns investing into a binary gamble on a single clinical result, using a heavy position to bet on a near-coin-flip event — the easiest way to get eliminated.
The right posture is to use structure against the power law — the barbell: one end with mature commercialized pharma companies with real cash flows (certainty, downside protection); the other end with small, diversified positions in early-stage, high-beta names (convexity, extreme upside); skip the tempting middle trap of 'heavy single bet.' Then layer on the two relatively high-conviction themes — the GLP-1 metabolic revolution and the patent-cliff-driven M&A wave — and you can enjoy the industry's enormous upside without being crushed by its brutality.
This approach is fundamentally the same as my asset allocation philosophy — not prediction (you can't know which clinical will succeed), but constructing a structure that won't be destroyed by any outcome and always captures the winners. Innovative drugs is the most fitting training ground for 'convexity thinking' and the 'barbell structure.'
If I had to leave just one sentence —
Innovative drugs runs on a power law, and the right way to play a power law is not 'pick that one winner' — it's a barbell: put one end in proven cash flows, and the other end in small, diversified convexity bets. Never bet heavy on a single clinical readout — that's the most tempting and fastest way to get eliminated in this industry.
Final article: we'll look at a sector bringing 'digital intelligence' into the physical world — robotics and physical AI, the biggest imagination playground for the next decade.
—
Risk disclaimer: This article is an industry chain study. The companies mentioned are for analysis illustration only and do not constitute any investment advice. Markets are risky; invest with caution.
On investment analysis, market insight, and asset allocation — not chasing short-term noise, but understanding what truly drives long-term returns.
