Building the AI Economy: Why Infrastructure at Scale Is a National Asset

In the first half of 2025, investment in data centers and information technology did most of the work of growing the American economy. By the estimate of Harvard economist Jason Furman, absent that buildout, annualized GDP growth would have been roughly 0.1% — a near standstill. The AI economy is being built right now, in steel and silicon and transmission lines, and the scale of the undertaking is precisely what makes it possible.

This is the clearest contemporary example of why large American companies are the country's infrastructure. The compute layer of the AI economy requires capital, land, power, and engineering capacity at a magnitude only a handful of firms can marshal. Understanding what that investment delivers — and how to manage its real costs responsibly — is one of the defining economic questions of the decade.

The Investment Carrying the Economy

The numbers are without modern precedent. The four largest hyperscalers, Microsoft, Amazon, Alphabet, and Meta, were on track to spend roughly $344 billion on capital expenditure in 2025, up from $228 billion the year before, according to Bank of America. Counting other technology firms pushes the total toward an estimated half a trillion dollars. Globally, data center capital expenditure reached approximately $770 billion in 2025, surpassing investment in both oil and gas and solar, according to analysis from Rystad Energy.

That spending is now a macroeconomic force in its own right. Barclays estimated that capital flowing into software, data centers, and computing hardware added about one percentage point to annualized U.S. GDP growth in the first half of 2025, roughly half of total growth for the period. J.P. Morgan reached a similar conclusion, finding that AI-related capital expenditure outpaced the American consumer as an engine of expansion. For the first time on record, by Renaissance Macro Research's calculation, the contribution of data center investment to GDP growth surpassed that of all consumer spending.

This is what large-scale enterprise produces: investment substantial enough to move the entire economy, concentrated in the productive capacity that determines national competitiveness in the technology that will define the century.

The AI Economy Runs on Power

Computation requires electricity, and the demands of AI are reshaping the American power system. Data centers consumed about 4.4% of total U.S. electricity in 2023 — roughly 176 terawatt-hours, up from 58 terawatt-hours in 2014, according to the Department of Energy's Lawrence Berkeley National Laboratory. The same body of research projects data centers could reach between 6.7% and 12% of national electricity use by 2028. The Electric Power Research Institute projects a range of 9% to 17% of U.S. electricity generation by 2030.

Meeting that demand is itself an infrastructure achievement, and companies at the center of the AI buildout are increasingly funding it. Data center operators are shifting from passive utility customers into active grid stakeholders — co-investing in transmission upgrades, securing long-term power contracts, and deploying on-site generation and storage to add capacity rather than simply drawing it down.

How much electricity do data centers use in the United States?

Data centers used approximately 4.4% of total U.S. electricity in 2023, about 176 terawatt-hours, according to the Department of Energy's Lawrence Berkeley National Laboratory. That share is projected to grow as AI workloads scale, with credible forecasts placing data center consumption between roughly 7% and 17% of national electricity by 2028 to 2030, depending on the pace of construction and efficiency gains.

Managing the Cost Responsibly

Rising electricity demand raises a fair and serious question: who pays for the grid that serves it? The honest answer begins with the evidence on what has actually moved electricity prices. A Congressional Research Service analysis drawing on Lawrence Berkeley National Laboratory research found that the main driver of electricity price increases from 2019 through 2025 was utility investment in grid infrastructure — much of it responding to aging equipment and resilience needs — alongside natural gas prices and recovery from natural disasters. That analysis did not identify data centers as a major influence on electricity prices in most areas of the country during that period.

The framework for keeping it that way is already taking shape, and it rests on a sound principle: large new electricity users should pay the full cost of serving them. Utilities and regulators have moved quickly, establishing at least 38 large-load tariffs since 2018, with 30 of those adopted in 2025 and 2026 alone, according to the energy consulting firm E3. Pennsylvania's Public Utility Commission advanced a model tariff in 2026 placing the cost of interconnection upgrades directly on large-load customers, shielding residential and small-business ratepayers. Colorado ordered a similar structure.

Well-structured, large customers can benefit everyone on the system. In a facility-level analysis of Amazon data centers across four utility territories, E3 found that each site generated on average roughly $3.4 million in net surplus revenue, payments to the utility exceeding the cost to serve the facility, with the difference available to ease pressure on other customers' rates. Scale, paired with sound rate design, becomes a contributor to grid affordability rather than a threat to it.

Scale Is the Precondition for the AI Economy

A data center campus capable of training frontier AI models represents billions of dollars in concentrated investment, multi-year construction timelines, and direct partnership with utilities to bring new power online. This is infrastructure in the original sense of the word — the foundational capacity on which an entire economy operates — and it can be built only by firms large enough to absorb the upfront risk and deploy capital at national scale.

The United States holds a lead in artificial intelligence today because American companies are building the physical foundation for it faster and at greater scale than anyone else. That foundation is generating economic growth now, anchoring high-wage construction and engineering work, and positioning the country to lead in the defining technology of the era. Sustaining it calls for a policy environment that allows investment at scale to proceed while applying the proven tools, transparent rate design, full cost responsibility for large loads, and continued grid investment, that protect every customer on the system.