Could the AI Boom Keep Interest Rates Higher for Longer?

How the largest technology investment cycle in decades could reshape economic growth, borrowing costs, and investment returns.

Aurevia Capital | Investment Perspectives | October 2026

Executive Summary

Artificial intelligence is increasingly becoming a story about physical investment as much as software. Data centers, advanced chips, power generation, transmission equipment, and other infrastructure require enormous amounts of capital. In principle, stronger demand for investment can put upward pressure on borrowing costs. But the evidence is more nuanced: despite extraordinary AI spending by a handful of companies, aggregate U.S. private investment has not yet risen dramatically relative to the size of the economy. At the same time, AI could eventually raise productivity and reduce some costs. The investment question, therefore, is not whether AI is simply “good” or “bad” for interest rates, but how a large capital-spending cycle interacts with productivity, inflation, financing needs, and asset valuations.

AI Is Becoming a Physical Investment Story

When people think about artificial intelligence, they usually imagine software. Yet the current AI boom depends on an enormous physical buildout: specialized processors, data centers, cooling systems, electrical equipment, and increasingly sophisticated power networks.

According to the Federal Reserve Bank of Minneapolis, capital spending by five of the largest investors in AI data centers—Alphabet, Amazon, Meta, Microsoft, and Oracle—is projected to rise from roughly $200 billion in 2024 to nearly $1 trillion annually by 2027.[1] 

That headline number needs an important caveat. The same Minneapolis Fed analysis notes that aggregate U.S. private investment as a share of GDP has remained roughly flat since 2018. In other words, enormous spending on AI infrastructure has not yet translated into an equally dramatic rise in economy-wide investment. Some of the increase is being offset by weaker spending elsewhere, including housing and other construction. AI is already a large investment story, but it is not yet dominating the macroeconomy.[1] 

Individual corporate commitments nevertheless illustrate the scale of the buildout. Meta expects $130 billion to $145 billion of capital expenditures in 2026, including principal payments on finance leases.[2] Microsoft reported $41 billion of capital expenditures in its June quarter, with roughly two-thirds going to shorter-lived assets such as CPUs and GPUs; the company also said it added 31 data centers during the quarter.[3] 

Key data point	What it tells us
Nearly $1 trillion projected annual capex by five major AI investors by 2027 [1]	The AI buildout is becoming economically significant, even though aggregate investment has not risen proportionately.
Meta $130–145 billion 2026 capex guidance [2]	AI infrastructure spending is already material at the individual-company level.
Microsoft $41 billion quarterly capex [3]	Large amounts of current spending are going directly into computing and data-center capacity.
About 945 TWh of global data-center electricity demand projected for 2030 [4]	The investment cycle extends well beyond technology into power and infrastructure.

From Investment Boom to the Cost of Capital

Why might all this spending matter for interest rates?

At its simplest, businesses are competing for capital. Building data centers and power infrastructure requires equipment, land, labor, electricity, and financing. Some large technology companies can fund substantial portions from their own cash flow. Others depend more heavily on bonds, bank loans, leases, private credit, or specialized financing structures.

The buildout is already producing examples of sizable external financing. In October, Reuters reported that SpaceX was seeking roughly $40 billion in bank loans and investment-grade debt to finance purchases of Nvidia chips for AI data centers.[5] The specific transaction may evolve, but it illustrates how increasingly capital-intensive AI projects can extend beyond internally funded spending.

If the number of attractive projects expands faster than the savings available to finance them, the price of capital can rise. That does not mean that every dollar spent on AI automatically produces higher interest rates. Borrowing costs are also shaped by inflation, Federal Reserve policy, government financing needs, global savings, and broader financial conditions. The more defensible conclusion is that AI has created a potentially important new source of capital demand—one that may add upward pressure to borrowing costs if it becomes large enough at the economy-wide level.[1] [6] 

This distinction matters. The Minneapolis Fed's evidence suggests that the pressure is present but, so far, moderate in the aggregate. That is a more balanced conclusion than claiming the AI boom is already a primary driver of interest rates.

Energy and Productivity Pull in Opposite Directions

The investment cycle also extends well beyond technology companies because AI requires large amounts of electricity.

The International Energy Agency projects that worldwide data-center electricity demand could more than double by 2030 to around 945 terawatt-hours, with AI the largest driver of the increase. In the United States, data centers could account for almost half of electricity-demand growth through 2030.[4] Meeting that demand will require additional generation, transmission, grid connections, and other infrastructure.

That creates a potential near-term tension. Heavy investment can support economic activity, but when demand for equipment, power, construction, or specialized inputs exceeds available supply, costs can rise. Federal Reserve policymakers discussed this dynamic in 2026, noting that strong demand for AI infrastructure was putting pressure on some technology and electricity prices and that the productivity benefits of AI might take longer to emerge.[7] 

The longer-term story could look very different.

AI may eventually allow companies to produce more with the same amount of labor and capital. It could automate administrative work, improve logistics, accelerate research, and lower the cost of delivering services. If those improvements become widespread, the economy could grow faster without generating the same degree of inflationary pressure. Both Federal Reserve policymakers and New York Fed research have highlighted this potential, while emphasizing considerable uncertainty around the timing and size of the gains.[6] [7] 

For now, however, the investment is easier to observe than the productivity payoff. Minneapolis Fed economists note that broad measures of aggregate productivity through early 2026 did not yet show a clear AI-driven boom. Their comparison with earlier technologies such as electricity and computers also suggests that major innovations can require substantial complementary investment and organizational change before their full benefits appear.[1] 

That creates one of the central tensions in the AI story: the cost of building the infrastructure is visible today, while the full economic return remains uncertain.

What It Means for Bonds and Equities

Long-term government bond yields are influenced by where investors believe short-term interest rates and inflation are headed, as well as the additional return they may demand for tying up money for many years. There is no need to assume that AI directly determines Treasury yields; it is one influence among many. [8] 

AI could matter if the investment boom keeps economic growth stronger, creates persistent demand for financing, or contributes to cost pressures. Conversely, large productivity gains could reduce inflationary pressure. New York Fed research therefore treats AI as something that could change both potential economic growth and the economy's underlying level of interest rates, rather than as a simple one-directional shock.[6] 

For stock investors, this creates an equally important tension. AI may increase revenues, margins, and long-term earnings for some companies. But higher interest rates make future profits less valuable in today's dollars and can make investors less willing to pay very high prices for growth that may take years to arrive.

That means AI can be economically transformative without every AI-related stock necessarily producing attractive returns.

Profitability is therefore becoming just as important as spending. Meta, for example, reported $31.08 billion of capital expenditures in the second quarter of 2026 while guiding to $130–145 billion for the full year; Microsoft reported $41 billion in quarterly capex while continuing to emphasize investment in AI infrastructure.[2] [3] The investment question is increasingly whether those expenditures generate sufficient future cash flow and returns—not simply how large the spending numbers become.

Portfolio Implications

For long-term investors, the implications extend well beyond technology stocks.

If AI supports stronger economic growth while also keeping borrowing costs elevated, bonds may offer more attractive income, although longer-term bonds could lose value if interest rates rise further. Stocks could benefit as AI improves productivity and profitability, while companies valued primarily on distant future earnings may face greater pressure if financing costs remain high.

Energy, utilities, data-center infrastructure, electrical equipment, and related businesses may also participate in the buildout. The IEA's projections make clear that AI's physical requirements are likely to create investment needs far beyond semiconductors and cloud computing. But greater demand does not eliminate execution risk, regulatory constraints, financing costs, or the possibility that too much capacity is built relative to eventual demand.

The broader lesson is that the economic success of AI will not necessarily translate into equally strong returns for every investment. Some companies may use AI to improve productivity with relatively little new capital. Others may spend heavily simply to defend their competitive position. For investors, revenue growth, free cash flow, balance-sheet strength, returns on invested capital, and valuation may ultimately matter more than the size of an AI spending announcement.

The AI revolution is therefore more than a technology story. It is increasingly a story about how businesses allocate capital, how the economy finances new investment, and where the returns from that investment ultimately accrue.

The question is not simply how much AI will transform the economy, but how that transformation will affect the cost of capital—and the returns investors ultimately earn.

Evidence Limitations

The central thesis should be interpreted as a plausible economic channel, not a demonstrated causal claim. The largest AI companies are spending at extraordinary levels, but the Minneapolis Fed finds that aggregate private investment relative to GDP has not yet risen materially, which weakens any claim that AI has already transformed economy-wide capital demand. Measuring AI's contribution to growth is also difficult because much of the hardware used in the buildout is imported; the Federal Reserve has cautioned that simply counting gross AI investment can overstate its contribution to U.S. GDP.[9] Productivity effects remain especially uncertain in both timing and magnitude. Accordingly, this article argues that AI could contribute to higher capital costs—not that it is the primary cause of current interest rates.

This publication-ready version preserves the thesis of Aurevia's supplied working draft while narrowing claims where the evidence does not establish causality and simplifying the bond-market terminology for a client audience.

Footnotes

[1] Federal Reserve Bank of Minneapolis, How Is AI Influencing Interest Rates? Investment, Productivity, Prices, and More, 2026. The article estimates that capital spending by five leading AI-data-center investors could approach $1 trillion by 2027, while emphasizing that aggregate private investment as a share of GDP remains roughly flat relative to 2018.

[2] Meta Platforms, Meta Reports Second Quarter 2026 Results, July 2026. Meta reported $31.08 billion of second-quarter capital expenditures and guided to $130–145 billion for full-year 2026.

[3] Microsoft, Fiscal Year 2026 Fourth Quarter Earnings Conference Call, July 2026. Microsoft reported $41 billion in quarterly capex, with roughly two-thirds spent on shorter-lived assets including CPUs and GPUs, and disclosed continued expansion of its data-center footprint.

[4] International Energy Agency, Energy and AI, April 2025. The IEA projects global data-center electricity demand of about 945 TWh by 2030 and estimates that data centers could account for almost half of U.S. electricity-demand growth through 2030.

[5] Reuters, SpaceX Seeks $40 Billion Financing to Buy Nvidia Chips, Sources Say, October 2026. The reported financing provides a current example of AI infrastructure relying on significant external capital.

[6] Federal Reserve Bank of New York, Artificial Intelligence and Monetary Policy: A Framework and Perspective on Cyclical Transmission, Structural Transition, and Financial Stability, Staff Report No. 1192, April 2026. The paper examines how AI could affect inflation, potential output, the natural rate of interest, and financial stability.

[7] Board of Governors of the Federal Reserve System, Minutes of the Federal Open Market Committee, June 16–17, 2026, July 2026. Participants discussed both near-term price pressure associated with the AI buildout and the possibility that future productivity gains could reduce production costs and expand aggregate supply.

[8] Board of Governors of the Federal Reserve System, Monetary Policy, Price Stability, and Equilibrium Bond Yields: Success and Consequences, November 2019. Provides the economic framework for understanding long-term bond yields as reflecting expected future short-term rates plus compensation investors require for holding longer-term bonds.

[9] Board of Governors of the Federal Reserve System, The AI Buildout and the Economy: Publicly Available Data to Assess AI's Impact, July 2026. The note highlights measurement challenges, including the risk of overstating AI's contribution to domestic GDP when imported computing equipment is counted as gross investment without properly accounting for net exports.

Disclaimer: This article is provided for general informational and educational purposes only. It does not constitute personalized investment advice, a recommendation to buy or sell securities, or a forecast of future market performance. All investments involve risk, including the potential loss of principal.