Every decade or so, a truly massive capital reallocation happens in the global economy. The railroads. Electrification. The internet. And now, the energy transition. By some estimates, the world needs to invest $4 trillion per year through 2030 to meet decarbonization targets. Four trillion. Per year. That is not a typo, and that is not a projection from an optimistic environmentalist. That is the International Energy Agency.
Every few decades, a technology comes along that reshuffles the entire deck. The printing press. Electricity. The internet. And now, artificial intelligence. When disruption hits, the same pattern repeats: a few companies ride the wave to extraordinary profits, most get crushed under it, and investors – watching from the sidelines or worse, from the wrong side of the trade – wonder how they missed it.
There is a man in my old neighborhood who ran a dead horse rendering business. No competition. Steady demand. Customers came back whether the economy was good or bad, because dead horses do not wait for favorable interest rates. It was not glamorous, but it was reliable – and reliability made him very wealthy while flashier operators went bust every few years.
Here is a number that should make you uncomfortable: roughly two-thirds of all corporate acquisitions are duds. Not “slightly disappointing” or “took longer than expected.” Duds. The acquiring company pays a premium, announces synergies, integration teams get deployed, and five years later the aggregate profits are maybe one-quarter of what was projected. Meanwhile, the CEO who approved the deal has moved on, the investment bankers collected their fees, and shareholders are left holding a lighter wallet. And yet, companies keep doing deals. Hundreds of billions worth every year. So the interesting question is not “why do acquisitions fail” – that part is well documented. The interesting question is: what separates the rare deals that create enormous value from the expensive failures?
Every engineer who has built a system knows there is a difference between something that works and something that becomes impossible to replace. A database you can swap out in an afternoon is just software. A database that half your company’s workflows depend on, that thousands of employees have built tooling around, that new hires learn on day one – that is infrastructure. Network effects work the same way. They are the mechanism by which a product transforms from “useful” into “the only rational choice.” And for investors, businesses protected by network effects are the closest thing to a legal monopoly you will ever find. Visa processes over 200 billion transactions a year. Not because their technology is impossible to replicate – it is not – but because every merchant, every bank, every cardholder is already connected. Starting a competing payment network is theoretically simple and practically impossible. That gap between “theoretically simple” and “practically impossible” is where fortunes are made.
Every civilization runs on infrastructure. Roads, power lines, rail tracks, fiber optic cables. Nobody thinks about them until they stop working. Then suddenly everyone has very strong opinions. Infrastructure is the plumbing of the economy. Invisible when it works, catastrophic when it does not.
Here is a fun exercise. Open your brokerage app right now and look at any AI stock – pick one, does not matter which. Check the 52-week range. There is a very good chance the high is double the low, maybe triple. Palantir swung from $17 to $80 in 2024. Super Micro Computer went from $230 to $1,200 and then back to $300 in approximately the same time it takes to binge a Netflix series. NVIDIA moved 15% in a single week multiple times this year. These are not penny stocks on some obscure exchange. These are large-cap companies with real revenue, real engineers, and real products. And yet their stock prices behave like the heart rate monitor of someone who just discovered espresso.