Eisman put the odds of OpenAI “dying suddenly” at 30%, and that’s no small number. He’s not simply saying that something could go wrong at one company. He’s saying that if this extreme scenario plays out, the U.S. could be dragged straight into a recession. That’s because this round of AI hype isn’t built on a single product; it rests on an entire chain powered by computing capacity: data centers need electricity, campuses need land, construction needs workers, and public-company valuations depend on index weightings to sustain demand. If any link in the chain breaks, the fallout can spread through balance sheets and employment.

Take electricity: Data centers currently account for about 3% of U.S. power consumption, and some estimates put that figure at 6%–7% by 2028. In some regions, grid connection queues are already measured in years. On the real estate side, data center campuses and their substations have become the new “infrastructure real estate.” A single large project can cost billions of dollars and require hundreds of megawatts. Construction can create thousands of temporary jobs, but if capital spending slows, vacancies, financing costs, and local government budgets can all turn into liabilities. On the employment front, expected job growth across semiconductors, servers, optical modules, construction, and operations and maintenance is tied to AI capital spending. And on the index side, the top 10 stocks in the S&P 500 now account for more than 30% of its weight. A handful of AI leaders are spending a combined total of more than $200 billion a year on capital expenditures. If those leaders become volatile, the index and pension accounts will sway along with them. The reason Eisman’s question—“Who actually profits from all this?”—hits a nerve is that the revenue goes to a handful of platforms and suppliers, while the costs and risks are spread across power, real estate, employment, and ordinary investors’ portfolios.

I’ve learned this lesson myself. When I was sitting on big unrealized gains, I was actually at my most cautious. I didn’t add when I should have, and now I’m watching others move higher. This isn’t about whether my call was right or wrong; it’s about letting the narrative drive my position sizing. So the more useful question now is: Which asset you hold is most vulnerable if the AI narrative loses momentum? High-valuation AI computing stocks, data center REITs, electric utilities, or broad-market index funds that rely on index weightings? Let’s see how things look over the weekend and compare notes then.