Artificial Intelligence and government debts... Are we entering an era of more expensive financing?

We are watching the artificial intelligence race to see which company will win, which stock will rise, and which jobs might disappear. But I believe another question deserves attention: What happens to the cost of capital when this boom requires massive amounts of money, while governments and other sectors also need funding?

In my view, we are facing a structural shift in the demand for capital. The expansion of AI requires data centers, power plants, networks, cooling systems, and equipment, along with financing that spans years. This means that competition among technology companies extends beyond software and services to encompass the resources underpinning the entire economy.

Numbers illustrate the scale of this transformation. The International Energy Agency expects capital expenditure by the largest data center operators to exceed $700 billion in 2026, with most of it directed toward developing new facilities. Companies are funding part of these investments through their profits and cash flows, but they are also turning to debt markets to meet their growing needs.

According to data published by Reuters on November 6, five major technology companies raised more than $370 billion in bonds since the beginning of the year, accounting for nearly 20% of the value of U.S. investment-grade corporate bond issuances. These issuances finance various activities, with AI expansion being among the drivers of this borrowing.

This is where the issue that concerns me begins. This financing is being sought at a time when governments are borrowing to fund their budgets and refinance their debts, and when industrial, housing, transportation, and service sectors are looking for funding for their investments. If the demand for funds grows faster than investors’ willingness to supply them, they may demand higher returns. Consequently, the cost of financing could rise for companies that have no direct connection to AI.

This does not necessarily require central banks to raise interest rates. The cost of borrowing over long periods is also influenced by the volume of issuances, expected inflation, investor appetite, and the risks they are willing to bear. Researchers at the Federal Reserve Bank of Dallas have warned that long-term issuances to finance AI may exert upward pressure on long-term bond yields and compete with other borrowers for the same funds. Thus, while official interest rates may remain unchanged, a manufacturer or a housing developer may find that available financing has become more expensive, leading them to delay expansion or reassess the viability of their projects.

This is what I mean by crowding out: when increased demand from one sector narrows financing options for other sectors or raises their costs.

In interconnected markets, the impact does not remain confined to the United States. All Gulf countries are affected to varying degrees by the cost of dollar-denominated financing, which governments and companies rely on to meet their needs. Rising yields on U.S. Treasury bonds may increase the cost of their new dollar-denominated issuances, even if local interest rates remain stable. Thus, the effects of financing the AI boom may extend to the cost of projects far removed from it.

Competition is not limited to capital. When network capacity is constrained, a data center and a factory may compete for the same energy supply, and equipment prices may rise or delivery times may lengthen. Conversely, opportunities expand for electricity and equipment suppliers, although growing demand does not guarantee the success of every project.

However, higher borrowing costs are not an inevitable consequence of every increase in investment. The boom may attract additional savings, generate income, and stimulate production expansion. Markets have so far absorbed the issuances, and a study by the Bank for International Settlements in October found limited and short-lived effects of AI financing news on bond markets. Yet, what might happen if the required financing continues to expand at this pace deserves close attention.

In my view, the challenge lies in timing. Funds, energy, and equipment are needed today to build data centers, while the diffusion of productivity gains takes time. A manufacturer may incur higher costs now, only to benefit later from technology that reduces waste and improves output. But postponing investment during this period could mean forfeiting an opportunity that is difficult to recover.

Therefore, I believe the AI boom warrants a reassessment of borrowing cost expectations and the capacity of other sectors to finance their own growth. The question that should accompany discussions of tech profits is whether the economy’s resources are expanding quickly enough to fund both this boom and the growth of other sectors simultaneously. The transformation may begin within data centers, but its impact on financing costs could extend to distant sectors.

With large government bond issuances, private sector financing needs, and the AI boom converging, demand for funds is likely to remain high in the near future, increasing pressure on financing costs for everyone.