Everyone's obsessed with which company will build the best AI model, but BlackRock (BLK) thinks investors are looking in the wrong place. According to Ben Powell, the firm's chief investment strategist for Asia-Pacific, the real money in AI isn't going to the model builders—it's flowing to the companies supplying the infrastructure that makes it all possible.

Following the Money Trail

Speaking at Abu Dhabi Finance Week, Powell laid out a straightforward thesis: the AI boom is driving unprecedented capital spending, and we're nowhere near the peak. The beneficiaries? Companies providing what he calls the "picks and shovels"—chipmakers, energy producers, and even copper-wire manufacturers.

"The money is very, very clear," Powell said, noting that major tech companies are just beginning to tap credit markets for the next wave of AI investment. Translation: there's a lot more capital coming.

Why the spending frenzy? Powell explains that hyperscalers are treating AI like a winner-takes-all battle. They're acting as if anything short of first place means getting pushed out entirely, which is driving aggressive spending even if it risks overshooting demand.

This isn't just about building better models anymore. It's about who can secure the infrastructure to support them—the hardware, the energy capacity, the physical components that keep data centers running. And that's where Powell sees the most durable investment opportunities.

From Spending Spree to Execution Challenge

BlackRock estimates that global AI capital expenditure could reach $5 trillion to $8 trillion by 2030, with the U.S. leading the buildout. The firm argues that this massive investment is creating a new capital-intensive economic regime that keeps growth resilient even as traditional business-cycle indicators cool.

But here's where it gets interesting: the game is changing. Jordi Visser of 22V Research argues that the next phase won't be about who spends the most—it'll be about who can actually execute. As the industry shifts from capital constraints to delivery constraints, success will depend on infrastructure readiness, project management, and the ability to turn contracts into real capacity while protecting margins.

By 2026, Visser predicts, the winners will be companies that can deliver under pressure, managing rising costs and tighter deadlines without sacrificing profitability.

The Trillion-Dollar Question

Amid all this infrastructure talk, I/O Fund CEO Beth Kindig made waves with a bold prediction: Nvidia (NVDA) could reach a $20 trillion valuation by 2030. Her logic? Wall Street keeps revising AI infrastructure spending estimates higher—from an initial $280 billion peak to McKinsey's recent estimate of $5.2 trillion just for AI data centers. If Nvidia maintains or grows its roughly 50% market share, that $20 trillion figure could actually materialize.

The bubble concerns that have shadowed AI investing haven't disappeared—critics still question whether demand will justify the explosive valuations. But Powell's point is that even if some AI companies stumble, the infrastructure providers are positioned to win regardless. Someone has to build the data centers, manufacture the chips, and supply the power. That's where the smart money is going.