The AI Buildout Meets the 5% World

BlackRock’s central claim is that competition for capital will rise. AI investment and government borrowing could lift annual U.S. financing demand above $7.5 trillion by 2030.

Takeaways

• BlackRock’s central claim is that competition for capital will rise. AI investment and government borrowing could lift annual U.S. financing demand above $7.5 trillion by 2030.

• The source of higher yields decides the equity implication. Growth and credible policy can coexist with risk-taking; inflation and rising term premium pose a harder test.

• AI demand extends beyond the next frontier model. BlackRock expects inference and adoption to sustain demand for physical computing capacity, while financing costs still test individual projects.

• Stay selective across stocks and bonds. BlackRock favours U.S. and emerging-market equities, shorter-duration income and credit backed by durable cash flows.

The AI Buildout Meets the 5% World

Stocks are being asked to survive a peculiar test: the investment boom supporting their earnings is making capital more expensive. AI companies need power, data centres, chips and financing on a vast scale. Washington needs financing too. With the 10-year Treasury yield above 5 %, BlackRock remains overweight U.S. equities and broadly pro-risk. Its argument is that the source of higher yields matters more than the level alone.

Jean Boivin, Wei Li, Vivek Paul and Ehsan Khoman put the AI buildout and heavy government borrowing on the same side of the capital ledger. BlackRock estimates annual U.S. financing demand could exceed $7.5 trillion by 2030, with AI’s capital needs a major driver. The broader AI and data-centre bond universe accounts for about 14% of U.S. investment-grade issuance this year, its commentary calculates, up from 5% in 2025 and 1% over the preceding decade.

There is the bull market’s quiet irony. Investors cheer another round of AI spending because somebody earns the revenue. The Treasury market sees another determined bidder for capital. Both can be right, and strong earnings can keep stocks moving while the cost of funding the next stage rises. The harder question is how much of that cost the builders, lenders and eventually shareholders will have to carry.

BlackRock makes a distinction worth carrying from the research note onto the trading desk. Slower progress at the frontier of AI models does not necessarily mean fewer data centres are needed. Much computing demand comes from inference, the work of running models already in use, and adoption remains young. A market frightened by a pause in the next model release may therefore be looking past the electricity and hardware demand created by deploying the models we already have.

I find that persuasive as a reason to separate the physical buildout from the daily drama of AI headlines. It is less persuasive as a reason to ignore price. Sustained demand for compute supports the infrastructure story; it does not guarantee every project earns its financing cost. Once yields are around 5%, the gap between “the world will need this” and “this particular borrower will make money from it” becomes large enough to trade.

BlackRock’s cross-asset picture shows how unevenly this year’s shocks have travelled. In its mid-month snapshot, Brent crude was up 94% year to date and emerging-market equities 73%, while Bitcoin was down 33%. U.S. equities remained positive despite the rate reset. “Risk-on” has not meant every risk asset marching to the same beat. Energy scarcity, AI growth and the rising cost of capital have rewarded different owners at different times.

Oil is the other claimant on this story. With the Strait of Hormuz effectively closed in BlackRock’s assessment and Brent above $100/bbl, higher fuel costs can lift headline inflation quickly. If they feed into wages and other business costs, the problem becomes harder for a central bank to look through. Heavy AI financing demand and an energy shortage can both push yields higher, but they ask very different things of the Fed.

BlackRock argues that the Fed’s hike helped restore credibility after the 10-year Treasury yield crossed 5%. In its reading, the rise in yields has mostly reflected higher real rates and expectations for tighter policy, rather than a sharp jump in term premium, the additional return investors demand for holding a long bond. It believes markets risk reading too much into Kevin Warsh’s hawkish tone.

That is the fulcrum for the equity trade. A bond selloff driven by resilient growth, investment demand and a credible Fed can coexist with rising stocks, particularly where earnings are strong. A bond selloff increasingly driven by inflation or doubts about fiscal and monetary credibility is a different animal. It asks investors to pay a higher price for uncertainty just as companies are paying more to borrow. The 5% handle tells you where yields are; it does not tell you which story the market is buying.

BlackRock places that contest inside a broader set of forces: AI, geopolitical competition, energy-system change, demographics and a changing financial architecture. The diagram looks tidy, but the market experience is anything but. A data centre sits at the intersection of several circles. It is an AI investment, a demand on power, a financing decision and, depending on where it is built, a question of supply security. A trader who treats it only as a technology earnings story leaves much of its risk with somebody else.

That framework leads BlackRock to a selective version of risk-taking. It favours U.S. and emerging-market equities, with particular attention to the scarce inputs of the AI buildout: power, chips and data centres. It is wary of long bonds, which have become more sensitive to rates and, in its view, less dependable as portfolio protection. In credit it wants clear cash flows, lender protections and recovery value rather than simply the highest coupon on the screen.

The common thread is the ability to bear a higher cost of capital. A strong company can finance a project and wait for its revenue to arrive. A heavily leveraged borrower may have to refinance before the promise becomes cash. In a world competing this fiercely for funding, I would rather know who has time on their side than be told a theme has a long runway.

The detailed positioning makes the distinction clearer. BlackRock is overweight U.S. and emerging-market equities, while Europe, the UK, Japan and China are neutral. It sees opportunities around physical AI in Japan and China, but does not assume that cheap AI models automatically mean attractive profits for the companies providing them. Exposure to a growing technology and ownership of a profitable business remain separate decisions.

On the bond side, BlackRock prefers shorter and medium maturities to long U.S. Treasuries and long investment-grade credit. It is overweight U.S. agency mortgage-backed securities and emerging-market local-currency debt, and underweight Japanese government bonds. Its credit view is selective even where headline yields look generous: tight spreads and heavy issuance leave less room for a borrower whose cash flows disappoint.

I would read those tables as the consequence of BlackRock’s macro argument, rather than as a shopping list. It is staying with the equity earnings created by the buildout while trying to limit exposure to the parts of the bond market most vulnerable to a further rise in long yields. Credit income is attractive only when the borrower can survive the financing environment that made the coupon attractive in the first place.

The question now is whether that separation holds. If the Fed retains credibility, AI adoption keeps physical demand firm and strong businesses turn spending into earnings, a 5% Treasury world need not end the equity advance. If oil-driven inflation persists or the sheer quantity of debt begins lifting term premium, the pressure can reach even good companies through valuation before it reaches their income statements.

BlackRock’s call is to stay pro-risk and choose carefully. My trader’s translation is to watch the reason for the next rise in yields before treating every move above 5% as the same signal. Growth can push a bull up a steep hill. Inflation and lost credibility can turn that hill into a wall. The yield on the screen may look identical; the trade on the other side of it is not.