Brookfield Enters the Cloud

Your Landlord Wants Your Margin

The hierarchy of the cloud has always been well-defined. At the top sit the "hyperscalers"—companies like Amazon, Microsoft, and Google—who write the software, manage the customers, and capture the massive margins. At the bottom sit the infrastructure providers: the companies that own the real estate, the fiber optic lines, and the power plants.

In this world, the tech companies are the master tenants and the infrastructure firms are the landlords. It is a comfortable arrangement. The tech companies get to pretend they are asset-light software businesses, and the landlords get steady, boring checks from some of the most creditworthy companies on earth.

But the AI boom has a way of turning comfortable arrangements upside down.

Brookfield Asset Management—a firm that manages over $1 trillion in assets and generally spends its time buying things like hydroelectric dams, office towers railroads, and pipelines—is reportedly launching its own cloud computing business. Here's SiliconAngle:

A report in The Information yesterday suggests it’s planning to launch a cloud computing business that would lease artificial intelligence chips directly to customers. It would operate a business model that can lower the costs of building and running AI data centers, the report added.

The new business would be operated by a subsidiary of Brookfield called Radiant, and is linked to a new $100 billion AI infrastructure program that was announced in November. That program is anchored by Brookfield’s Artificial Intelligence Infrastructure Fund, which has already committed $10 billion to the initiative, with half coming from a group of institutional and industry partners, including the chip giant Nvidia Corp. and the Kuwait Investment Authority.

This is where the story gets rather strange, because Brookfield is, culturally speaking, the anti-Silicon Valley. The tech world's mantra is "move fast and break things." Brookfield's is "move slowly and whatever you do, do not break the dam." Tech companies obsess over software and ephemeral assets like GPUs that are obsolete in five years. Brookfield obsesses over physical infrastructure with a half-century lifespan.

The traditional logic of the cloud was that the "brains" (the software) was the hard part, and the "body" (the power and the dirt) was the commodity. Brookfield is betting that the polarities have reversed.

In a world where you can't build a data center without first finding a few spare gigawatts of power, owning the power is no longer a commodity service. It is the ultimate competitive moat. Brookfield looks at its portfolio of renewable energy and real estate and sees a vertical integration opportunity. If you already own the power plant and the land, why let AWS sit in the middle and collect the premium for turning those electrons into AI tokens?

This shift is part of a broader, stranger trend in how AI is being financed. We talked recently about how Meta is using "off-balance-sheet" structures to fund its build-out, essentially getting firms like Blue Owl Capital to buy the assets so Meta can just rent them back. It is a strategy designed to keep the massive debt of the AI race from weighing down a tech company's stock price.

But Brookfield is coming at the problem from the opposite direction. While tech companies are trying to move their physical assets off their books to please their investors, Brookfield is using its mastery of physical assets to stage a hostile takeover of the tech companies' business model.

There is a certain irony here. For a decade, Silicon Valley has been trying to "disrupt" every physical industry it could find, from taxis to hotels. Now, the people who own the most physical assets on the planet—the literal landlords of the global economy—have decided that the "cloud" is just another piece of infrastructure they are better suited to own.

If Brookfield succeeds, it suggests that the most valuable part of the AI stack isn't the code or even the model. It is the ability to navigate the messy, physical reality of the power grid. It turns out that in a race to build a digital god, the person with the keys to the transformer station might be the one who actually dictates the terms.

More on AI Infrastructure

• Why AI Companies May Invest More than $500 Billion in 2026 (Goldman Sachs)
• SoftBank to buy data center firm DigitalBridge for $4 billion in AI push (CNBC)

On Our Radar

Our Intelligence Desk connects the dots across functions—from GTM to Operations—and delivers intelligence tailored for specific roles. Learn more about our bespoke streams.

The Compute Gigafactory

• The Headline: Elon Musk's xAI is rapidly expanding its "Colossus" data center complex, aiming to create a nearly 2-gigawatt AI training facility in a brute-force bid for compute supremacy. (Bloomberg)
• ARPU's Take: This is the AI arms race expressed as a civil engineering manifest. Musk is applying the Tesla Gigafactory playbook directly to AI: win by building an insurmountable physical advantage in raw power and GPUs. He's betting that the ultimate victor won't be the one with the cleverest algorithm, but the one with the biggest factory.
• The Operations Implication: xAI's strategy is designed to create a massive demand shock for the two most constrained resources in the AI supply chain: top-tier Nvidia GPUs and grid-level electrical power. By attempting to corner a significant portion of both, Musk forces rivals into a defensive CapEx cycle and shifts the primary operational bottleneck from securing financing to securing the land and utility contracts needed to deploy compute at this scale. The competitive arena is no longer just the lab; it's the construction site and the local power substation.

P.S. Tracking these kinds of complex, cross-functional signals is what we do. If you have a specific intelligence challenge that goes beyond the headlines, get in touch to design your custom intelligence.

You received this message because you are subscribed to ARPU newsletter. If a friend forwarded you this message, sign up here to get it in your inbox.