AI data center investment is its own asset class now. The public conversation about artificial intelligence runs on chips and software, but every model has to live somewhere, and “somewhere” is a fast-growing stock of data centers that need land, buildings, power, cooling, and money. That physical layer is where a large share of patient capital is quietly flowing, and it is the layer Charles Mizrahi’s American Prosperity Report pitch is built around. Here is how investing in it works and where the pitch fits.
The Three Basic Routes In
There are three practical ways to put money into AI data centers, and they carry very different risk profiles. The first is owning the real estate itself through a data center REIT, the landlord layer that collects rent on the buildings hyperscalers lease. The second is owning the operators and developers that design, build, and staff the facilities. The third is owning the asset managers and private funds that raise the capital and take a fee for deploying it.
Each route exposes you to a different slice of the same underlying trend. REITs deliver income and relative simplicity but limited upside when demand spikes. Operators carry the most upside when the market accelerates, along with the most execution risk. Asset managers sit in the middle, earning fees on every dollar that flows into the buildout whether any single facility is a home run or not. We cover the landlord layer in our best data center REITs explainer and the full operator-versus-financier split in our data center stocks piece.
Why Demand Keeps Climbing
The demand story is the easy part to verify. Hyperscalers such as Google, Microsoft, and Meta have committed enormous sums to data center construction because their AI ambitions depend on owning vast compute capacity. The American Prosperity Report cites a “$7 trillion pipeline” of AI-infrastructure deals, listing Google at $3 billion, Microsoft at $10 billion, Meta at $29 billion, NVIDIA at more than $1 billion, and the US government at $80 billion. Those commitments are the kind of numbers that keep a financing-driven business busy for a decade.
The distinction worth holding onto is that these are infrastructure commitments, not software bets. AI needs electricity and physical space before it needs a cleverer algorithm, and that ordering is what gives the data center layer its defensive character. The chip designer and the model builder are one kind of investment; the company that builds and powers the building is a different kind entirely.
How the Money Is Actually Made
The economics of a data center project explain why asset managers gravitate toward them. A facility is financed with a mix of equity and debt, built over two to four years, and then leased to tenants on long contracts that pay for power, space, and connectivity. The developer earns a construction margin, the operator earns a management fee, and the capital provider earns a return tied to the cash the building throws off. The full mechanics of that fee model are covered in our alternative asset managers explainer.
For an asset manager, the recurring fee income is the attraction: it compounds as assets under management grow, largely independent of any single project’s outcome. That is the model Brookfield operates. It raises third-party capital into funds, deploys it into infrastructure including data centers and the power behind them, and charges management and performance fees on the way. The AI boom is not a direct Brookfield product line; it is the demand wave that fills its pipeline.
Where the Brookfield Thesis Sits
Brookfield Corporation is the reveal behind the American Prosperity Report pitch, and it is instructive because it is not a data center company at all. It is a global alternative asset manager with roughly $1.3 trillion in assets under management that finances and operates the power and data center infrastructure AI depends on. That is an indirect AI exposure, which the pitch leans on with its “$7 trillion pipeline” and “40% undervalued” claims.
The honest framing is that Brookfield is a financing-and-operating bet on the AI buildout rather than a chip or software bet. Its chief executive, Bruce Flatt, has run the firm for more than twenty years and is often compared to Warren Buffett for his long-term, value-oriented style. That pedigree is real, but it does not make the stock an AI pure play, and investors should price it accordingly.
What to Watch Before You Invest
Two things deserve attention before committing capital. First, this is a capital-heavy business, and Brookfield carries roughly $270 billion in consolidated debt, most of it non-recourse to the parent. Non-recourse means a lender can seize a specific asset but cannot chase the parent, which is a meaningful distinction when reading a scary headline. Second, a portion of the portfolio sits in commercial real estate and office assets that have been under pressure.
Neither item cancels the AI thesis, but both are reasons to size the position like an infrastructure investment rather than a high-growth technology position. The data center theme is durable; the valuation and the balance sheet are where an investor’s attention should stay.
Ready to see the research? Click here to access Charles Mizrahi’s report.
NewsletterVetter is an independent publication. We receive compensation from some of the services we review through affiliate links. Nothing on this site is investment advice. Always do your own research.