Part 2: Case for Monopolies & Prize Funds
Don't Fund Research. Buy Results.
When did America's most important research move from universities to corporations? And more interestingly—why didn't anyone notice?
Something strange happened over the last 30 years. MIT still exists. Stanford still has labs. The NIH budget keeps growing. But when you look at where breakthrough research actually happens—from AI to antibiotics, from quantum computing to cancer drugs—it's coming from companies, not universities.
We didn't plan this. Nobody voted for it. It just happened.
Pfizer runs more advanced biotech research than any university. Google solved protein folding. Microsoft is building AGI. Moderna created a COVID vaccine in 48 hours. Even Walmart runs more sophisticated logistics research than academic operations research departments.
The privatization of American research isn't coming—it's complete.
Most people see this as a problem to solve. They're having the wrong debate. They're arguing about whether corporations should do research. But that ship has sailed. The question now is: how do we point this massive private research apparatus at problems that don't have natural markets?
The answer is surprisingly simple. But to understand it, we need to first understand how we got here.
The Great Migration
In 1970, if you were a brilliant physicist, you probably worked at a university or government lab. By 2024, you almost certainly work for a corporation. What happened?
The simple answer is money. But that's not the whole story. The real answer is that companies became better at research than universities. [1]
This sounds backwards. Universities are supposed to be about pure knowledge. Companies are supposed to be about profit. But somewhere along the way, those roles flipped.
Universities optimized for publication counts. Companies optimized for breakthroughs. Universities chased grant money. Companies chased solutions. Universities built bureaucracies. Companies built labs.
The migration was gradual, then sudden. First it was just the applied researchers. Then the theorists. Then even the pure mathematicians. Google now employs more top machine learning researchers than any university. Microsoft has more quantum physicists than most national labs.
But here's the interesting part: this might actually be good.
Companies have something universities lost: urgency. When Microsoft funds fusion research, they want fusion to work. When Pfizer develops a drug, they need it to actually cure people. The feedback loop is tighter. The results matter.
There's just one problem.
Markets Point in the Wrong Direction
Markets are optimization functions. They're incredibly good at finding solutions—but only to problems that make money.
This works great for smartphones and social networks. It works terribly for antibiotics and pandemic preparedness.
Consider antibiotics. By 2050, antibiotic resistance will kill 10 million people annually. We know how to develop new antibiotics. Pharma companies have the labs, the talent, the capability. They just don't have a reason.
The math is brutal. Developing a new antibiotic costs $2 billion. But good antibiotics are used sparingly to prevent resistance. Your blockbuster drug that saves humanity generates maybe $100 million in revenue.
Spend $2 billion to make $100 million? That's not a business model. That's charity.
So pharma companies work on what pays: chronic conditions, lifestyle drugs, expensive cancer treatments. Not because they're evil. Because they're rational. [2]
The same logic applies everywhere. Energy companies won't develop grid-scale storage that makes their existing assets worthless. Tech companies won't build AGI safety solutions that slow down AGI development. Nobody will build pandemic preparedness systems that sit idle between pandemics.
Markets fail when the benefit is public but the cost is private. And for civilization's biggest problems, the benefit is almost always public.
The $6.5 Million Miracle
In 2004, DARPA did something strange. Instead of funding research into autonomous vehicles, they offered $1 million to anyone who could build one.
The research community thought this was insane. Autonomous vehicles required breakthroughs in computer vision, sensor fusion, path planning, and machine learning. You can't just offer a prize and expect magic to happen.
106 teams disagreed.
What happened next was beautiful. Every team failed. Carnegie Mellon's heavily funded effort made it 7 miles before driving off a cliff. Caltech's entry caught fire. One motorcycle team's bike fell over at the starting line.
But here's what made it beautiful: every failure was public. No NDAs. No trade secrets. No patents. Just 106 different approaches, failing in 106 different ways, where everyone could see.
The learning was exponential. Teams could see exactly what didn't work and why. The computer vision team learned from the sensor fusion team's mistakes. The path planning team learned from the machine learning team's failures.
By 2005, five teams finished the course. By 2007, six teams were navigating urban traffic.
Then something predictable happened: the winners became founders. Sebastian Thrun, who led Stanford's winning team, started Google's self-driving project—now Waymo, worth >$50 billion. Chris Urmson, who won with Carnegie Mellon, later founded Aurora ($13 billion). Other Challenge veterans launched Cruise (acquired by GM for $1 billion), Argo AI (Ford/VW, $7.4 billion), and Nuro ($8.6 billion).
Venture Capitalists had watched the Urban Challenge from the audience. Within months, they were writing checks. The entire autonomous vehicle industry—every major player—traces back to those desert failures.
Total DARPA investment: $6.5 million. Value of self-driving car industry: $500+ billion. ROI: 76,923x.
But the real lesson wasn't about self-driving cars.
Buying Results, Not Research
DARPA discovered something profound: you get what you pay for. If you pay for research, you get research. If you pay for results, you get results.
This sounds obvious. But somehow we've built a $55 billion grant system that does the opposite. The NIH doesn't pay you to cure cancer. They pay you to study it. [3]
Think about the incentives. If you're a grant-funded researcher, success means getting more grants. The way to get more grants is to publish papers. The way to publish papers is to work on problems that are interesting but not too hard. Solving the problem completely would actually be bad—then you couldn't get grants to study it anymore.
This created a system that optimizes for the wrong thing. When Amgen tried to reproduce 53 "landmark" cancer studies, only 11% worked. Bayer found similar results—just 25% of published findings could be replicated. We're spending $55 billion annually on research that's 75% noise. [4]
Prizes flip this entirely. You don't get paid to try. You get paid to succeed. No publications, no citations, no committees. Just results.
This changes everything about who can participate. With grants, you need the right credentials, the right institution, the right connections. You need to know how to write a grant proposal—a skill completely unrelated to solving problems.
With prizes, none of that matters. A garage inventor can compete with MIT. A startup can beat a national lab. Pharma companies can dust off shelved research. Anyone with an idea can try.
The payment structure maintains rigor: 10% for proof of concept, 30% for demonstration, 60% for full deployment. You get funding to attempt the solution, but the big payout requires actual success. This filters out cranks while enabling serious attempts from unexpected places.
When Stanford's team entered the DARPA challenge, they weren't the best funded or the most credentialed. They won because they learned fastest. That's what prizes select for: learning speed, not pedigree.
The Two Problems Worth Solving
Once you understand that we've privatized research and that prizes can direct it, the question becomes: what should we buy?
There are two categories of problems worth creating markets for:
Type 1: Acceleration Prizes These are technologies that markets will eventually develop, but too slowly. Self-driving cars would have arrived by 2040. DARPA pulled them forward to 2025. That 15-year acceleration will save hundreds of thousands of lives.
Type 2: Existence Prizes These are breakthroughs markets will never develop because the economics are permanently broken. Antibiotics. Pandemic vaccines for unknown diseases. Carbon capture. The science is possible. The business model isn't.
The key insight is that you don't need to fund the research. You just need to create the market. Companies will handle the rest.
Why America Can Do This and China Can't
Prizes aren't just another funding mechanism. They're free markets for research.
Think about what we've been doing with grants: centralized allocation. Government bureaucrats decide who gets money, what approaches to try, which institutions deserve funding. It's Soviet planning with American characteristics.
China should be great at this. Central planning is their entire model. But they're losing to our accidental free market—where private companies compete to solve moonshot problems.
Prizes make this explicit. Instead of government picking winners, it just defines winning. Then it steps away and lets the market work.
This is our systemic advantage. Free markets are the greatest information processing system ever invented. Millions of independent actors, trying different approaches, failing in different ways, learning from each other. No central planner, however brilliant, can match this.
When you offer a prize, you're admitting profound ignorance: "We don't know how to solve this problem. We don't know who can solve it. We don't know what approach will work." You're crowdsourcing intelligence to the entire market.
Imagine Xi Jinping announcing: "We don't know how to build quantum internet, so here's $5 billion for whoever figures it out. A garage startup in Shenzhen might beat Huawei. We're fine with that."
It's systemically impossible. Their entire model is built on the state knowing best, picking winners, controlling outcomes. They can copy our technology. They can't copy our market structure.
When Stanford's scrappy team beat GM in the DARPA challenge, we celebrated. That's not a bug—it's the feature. The ability for nobody to beat somebody, for chaos to beat planning, for markets to beat ministers.
This is 21st century government done right: Define problems. Create incentives. Get out of the way. Let free markets do what they do best—solve problems through competition.
We've been trying to out-plan China's planners. That's fighting on their terms. Prizes let us fight on ours.
The Implementation Is Obvious
We're already spending $55 billion annually on research grants. Cancel all grants and convert them to prizes. Pay for cures, not conferences.
Start with the obvious ones:
New antibiotic classes: $10 billion each
Pandemic vaccine platform (any virus, 100 days): $20 billion
Carbon capture at scale: $25 billion
Structure them intelligently. 10% for proof of concept, 30% for demonstration, 60% for deployment. Make them open to anyone—garage startup or Google.
But here's what's elegant: we don't need to figure out all the prizes. Once you establish the principle, problems will reveal themselves. Markets are good at that.
The Real Insight
Last week I wrote about how America's tech monopolies became our research labs. Readers panicked, thinking I was defending monopolies.
They missed the point. I wasn't defending anything. I was describing reality.
We've built the most powerful research engine in human history. It just happens to be distributed across corporate balance sheets instead of university endowments. Google has $170 billion in cash. Apple has $165 billion. Pfizer has $44 billion.
Fighting this is like fighting gravity. The smart move is to use it.
Prizes don't replace corporate R&D. They multiply it. When there's a $10 billion antibiotic prize, watch Pfizer dust off shelved research. When there's a $50 billion AGI safety prize, watch Google's deployment timeline suddenly include safety stops.
We keep trying to rebuild the 20th century research system. But we've already built something better. We just need to point it at the right problems.
The future belongs to whoever figures out how to unleash innovation, not control it. We've already done the hard part—building organizations that can tackle impossible problems.
Now we just need to make the impossible profitable.
[1] I'm using "better" in a specific sense here: more likely to produce usable breakthroughs. Universities are still better at producing papers.
[2] This is why the "ESG" movement mostly fails. You can't make companies act against their interests. You can only change what's in their interest.
[3] Yes, some grants have milestone payments. But the core metric is still publications, not solutions.
[4] This is also why prizes are more transparent than grants. When someone wins a prize, you can see exactly what they built. When someone gets a grant, you can only see what they promised to try.
Very well written.