Brookfield bets on the atom: how a legendary investor is building the foundation for AI

While venture capitalists and AI enthusiasts argue over which language model writes code faster, serious players in the world of big money look at the situation in far more down-to-earth terms. They look at the power outlet. Bruce Flatt, longtime head of Brookfield Asset Management, has decided it is time to stop spending his time on day-to-day operations. He is handing the CEO post to Connor Teskey, but not for a well-earned rest on some island. Flatt is moving up a level, to the parent company Brookfield Corp, to carry out a plan that could make his company the main beneficiary of the AI boom, even if none of today's neural networks survive into the next decade.

There is a clear calculation behind this personnel move. Brookfield is a machine with nearly a trillion dollars in assets, and it is now turning toward the insurance business and heavy infrastructure. Flatt plans to turn the company into an investment-and-insurance behemoth, using capital from insurance premiums to finance the most expensive and longest-term projects on the planet. Why does this matter to the tech industry? Because AI is not only chips and algorithms; above all, it is colossal energy consumption. Brookfield is betting that in the coming years a shortage of capacity will become the main brake on progress, and whoever holds the "circuit breaker" will dictate the terms.

Nuclear power holds a special place in Flatt's strategy. We have already seen Microsoft strike deals to restart the Three Mile Island reactors, but Brookfield has been playing this game for a long time and professionally. Flatt understands that solar panels and wind turbines cannot provide the stable baseload that giant GPU clusters running 24/7 require. Nuclear power is the only way to satisfy the appetites of Google, Amazon and OpenAI without breaking climate commitments. Brookfield is effectively building the physical shell for a digital mind, investing in something that cannot simply be copied or retrained on new data.

Handing authority to Connor Teskey at Brookfield Asset Management frees Flatt's hands for more aggressive capital maneuvering. Teskey is a proven fighter who successfully ran the company's renewable energy business, so continuity is guaranteed. Flatt, for his part, will focus on building a structure that will feed the AI industry for decades. It is a classic "picks and shovels" strategy from a gold rush. While developers burn billions of dollars hoping to create AGI, Brookfield plans to earn money on every watt that AGI consumes.

The shift to an investment-insurance model is also interesting. In a world where interest rates remain unstable, owning insurance capital gives Brookfield access to cheap, "long" money. It is ideal fuel for building data centers and nuclear plants whose payback is measured in decades. Flatt is effectively creating a closed loop: insurance money builds infrastructure, infrastructure powers AI, and contracts with tech giants provide the steady returns needed to pay out insurance claims. It is an elegant and frighteningly large-scale game.

What does this mean for the market? We are entering a phase in which the success of AI companies will depend not on the brilliance of their engineers but on their ability to strike deals with giants like Brookfield. If you have no access to cheap energy and cooling, your smartest model will remain just a set of weights on a disk. Flatt understands this better than many in Silicon Valley. While the industry discusses software, he is building the world in which that software will live. And that world will be very expensive.

Bottom line: Brookfield is moving from managing other people's assets to building its own infrastructure empire for AI's needs. Will the tech giants have enough money to pay Bruce Flatt's electricity bills?