Analyst: AI Capital Demand Is Reducing Appetite for US Treasuries

If private capital is being redirected toward AI infrastructure at scale, the federal government may face structurally higher borrowing costs at a moment when deficit spending remains elevated.

Rebecca Patterson, former Chief Investment Strategist at Bridgewater Associates and Senior Fellow at the Council on Foreign Relations, said in a September 25, 2026 interview with Bloomberg's "The Close" that artificial intelligence investment is competing directly with demand for US Treasury securities, potentially pressuring yields upward as capital flows toward private AI infrastructure rather than government debt.

Patterson's remarks came during a broader discussion with anchor Romaine Bostick about whether signals in bond markets, equity markets, commodities, and energy are pointing toward a single coherent economic story or whether any one asset class is more predictive than the others, according to Bloomberg.

The mechanism Patterson describes is capital competition. When large institutional investors and sovereign funds allocate a greater share of available capital to AI-related private assets, including data center construction, semiconductor supply chains, and model training infrastructure, they hold proportionally less in low-yield government paper. Reduced demand for Treasuries, all else equal, pushes yields higher and raises the cost the federal government pays to finance its debt.

The US Treasury Department's Office of Debt Management reported in its August 2026 quarterly refunding documents that the government planned to issue approximately $1 trillion in net new marketable debt across fiscal year 2026. Borrowing costs at higher yields directly increase the annual interest expense carried in the federal budget, which the Congressional Budget Office projected in its June 2026 baseline at $892 billion for fiscal year 2026, making interest the third-largest federal expenditure category behind Social Security and Medicare.

Patterson's framing adds a structural variable to that budget pressure. Traditional crowding-out theory holds that large government deficits raise interest rates by competing with private borrowers for available loanable funds. The argument Patterson raises inverts a portion of that dynamic: private sector demand for capital, specifically AI-driven demand, may itself be pulling institutional investors away from the Treasury market, achieving a similar yield-widening effect from the demand side rather than the supply side.

The political dimension of rising Treasury yields has drawn attention across both parties in Congress. Members of the House Budget Committee held hearings in July 2026 on debt sustainability, with Republican members citing interest cost projections as evidence of unsustainable fiscal trajectory and Democratic members disputing the near-term severity while acknowledging long-run risk. The hearing record is publicly available through the House Budget Committee's official archive.

AI infrastructure spending has accelerated sharply in the period Patterson is referencing. Federal procurement and private capital deployment in AI have both grown. The White House convened AI executives in a policy session earlier this year, as covered by Trump and Johnson to Meet AI Executives at White House Tuesday, indicating that AI investment is now a recognized factor in federal economic planning discussions.

Federal Reserve data on Treasury holdings, published in the Z.1 Financial Accounts of the United States, would show whether institutional investor categories, including mutual funds, insurance companies, and pension funds, have reduced their aggregate Treasury positions in recent quarters. The Federal Reserve's most recent Z.1 release, for the second quarter of 2026, was published in September 2026 and provides the most granular publicly available breakdown of who holds US government debt and in what amounts.

It is not yet established from public records how large the reallocation away from Treasuries and toward AI assets has been in dollar terms, which investor categories are driving any shift, or whether the effect Patterson describes is temporary or represents a durable change in institutional portfolio strategy. Treasury International Capital data, published monthly by the Treasury Department, would show whether foreign institutional holders are also reducing positions, and the next TIC release is scheduled for October 2026.

Patterson did not specify a dollar figure for the crowding-out effect or cite a particular data series to quantify it in the Bloomberg interview summary. What public records would further clarify the claim include the Federal Reserve's Z.1 report, TIC data, and any forthcoming Treasury refunding statements that address changes in the domestic investor base for government securities.