The Superintelligence Gamble Transcript and Summary

Mark Zuckerberg is betting the GDP of Kenya on superintelligence, wooing top researchers with homemade soup and billion-dollar contracts.

[Speaker 1]: If you want to understand the single biggest gamble in the history of the modern internet, you don’t start in a boardroom. You actually start in a living room in Palo Alto, about six months ago. [Speaker 2]: Right. The soup era. [Speaker 1]: Exactly. The soup era. It’s mid-2025. There’s a frantic war for talent happening in Silicon Valley. And the image that keeps coming up is Mark Zuckerberg, personally hosting dinners for top AI researchers. And he’s not just hiring caterers. He is literally serving them homemade soup. [Speaker 2]: Which sounds like a quirky PR detail, until you look at who was at the table. These weren't just coders. These were the architects of the next generation of intelligence. And to get them to sign, Meta wasn't just offering soup. Reports suggest some of the compensation packages for people like Andrew Tulloch reached one point five billion dollars over six years. [Speaker 1]: That is a staggering amount of money for human capital. But it turns out, the soup and the salaries were just the down payment. Because today, we got the bill for the rest of the plan. [Speaker 2]: And the number is shocking. Meta has released its guidance for 2026, and they are planning to spend one hundred and thirty-five billion dollars on infrastructure in a single year. [Speaker 1]: Just to contextualize that-one hundred and thirty-five billion dollars is roughly the entire GDP of Kenya. Spent in twelve months. On chips, data centers, and power. [Speaker 2]: Most companies cut costs to survive. Today, we’re looking at why Mark Zuckerberg thinks the only way to survive is to spend the equivalent of a mid-sized African economy on plumbing. [Speaker 1]: It’s Tuesday, February 3, 2026, and you’re listening to The Angle. [Speaker 2]: So, usually when a mature company announces they’re going to light a hundred billion dollars on fire to build "infrastructure," Wall Street panics. Investors want profits, they don't want science projects. [Speaker 1]: Right. And we actually saw that happen yesterday with Microsoft. They announced huge AI spending, admitted they haven't figured out how to monetize it yet, and their stock dropped. The market punished them. [Speaker 2]: But then you look at Meta this morning. They announced an even bigger number-that hundred and thirty-five billion cap-and the stock is up almost ten percent. [Speaker 1]: Which feels like a contradiction. Why is the market punishing Microsoft for spending billions, but giving Meta a standing ovation for spending even more? [Speaker 2]: It comes down to one really critical distinction: The difference between a feature and a utility. Microsoft is building AI features that they hope you’ll pay a subscription for. Meta is claiming they are building the electric grid of the internet. [Speaker 1]: And unlike everyone else, they’ve found a way to make the construction costs pay for themselves immediately. [Speaker 2]: Exactly. This isn't charity. If you look at the Q4 earnings from last week, Meta posted net income of nearly twenty-three billion dollars. They have the cash. But the reason investors are letting them spend it is something called GEM. [Speaker 1]: Generative Experience Models. [Speaker 2]: Right. And this is the mechanism that explains why the stock is up. For the last decade, digital advertising was static. I show you a picture of a sneaker. If you like sneakers, maybe you click. [Speaker 1]: Simple enough. [Speaker 2]: GEM changes the physics of that interaction. Think of GEM as a dynamic director that lives inside the ad server. It’s not just picking an…

Related coverage: Business & Economy · Popular stories