AI infrastructure spending projected to outpace America’s past building booms

Elon Musk has never been shy about thinking on a planetary — or interstellar — scale, and his latest comment does exactly that. Reacting to a chart comparing AI infrastructure spending to America’s biggest historical building booms, Musk wrote that “an Earth economy is less than a trillionth the size of a K2 economy.” The remark, tossed off in a repost on X, lands at a moment when the sheer scale of AI-related capital spending is already drawing comparisons to the railroad age, the highway system, and the dot-com fiber boom.

Key takeaways

• AI infrastructure spending is projected to average 3.63% of US GDP between 2025 and 2032, according to the chart Musk reposted.
• That share outpaces past US infrastructure booms: railroads averaged 2.24% of GDP from 1870 to 1890, and highways peaked at 1.13%.
• Columbia Business School researcher Stijn Van Nieuwerburgh estimates the total AI buildout near $8.2 trillion across data centers, power, and IT equipment.
• Musk compared Earth’s entire economy to a theoretical “K2 economy,” saying it is less than a trillionth the size.
• Investor Ray Dalio has separately warned about AI bubble liquidity risks, including possible spillover into pension funds.

Record-Setting AI Infrastructure Spending

The numbers behind Musk’s comment tell a striking story on their own: AI infrastructure spending is on track to eclipse every major US building boom of the past 150 years, at least as a share of the overall economy. That alone is worth pausing on before getting to the interstellar comparisons.

How AI Spending Stacks Up Against Past Booms

The chart that prompted Musk’s reaction tracks average annual US infrastructure spending as a percentage of gross domestic product across different historical eras. Between 2025 and 2032, AI-related outlays are projected to average 3.63% of GDP. That figure sits well above the railroad expansion of 1870 to 1890, which averaged 2.24% of GDP, and the interstate highway push, which topped out at 1.13%. Telecom and fiber-optic buildouts during the dot-com era reached roughly 1.1%, while canals and electrification trailed further behind at 0.66% and 0.5%, respectively. In short, if the projection holds, the current wave of AI infrastructure spending would represent the largest infrastructure commitment, relative to the size of the economy, that the country has ever undertaken.

Estimated Total Buildout Costs

Behind that percentage sits a dollar figure large enough to reshape entire industries. The chart’s data is credited to Stijn Van Nieuwerburgh, a real estate professor at Columbia Business School, whose DataCenterJEP research estimates the total AI buildout at close to $8.2 trillion. That figure spans data centers, the power generation needed to run them, and the IT equipment inside. It is one of the more concrete attempts to put a single number on what is otherwise a sprawling, multi-year investment cycle spread across chipmakers, utilities, cloud providers, and construction firms.

Elon Musk’s K2 Economy Analogy

Musk’s framing pushes the conversation well past dollars and GDP percentages and into the territory of astrophysics. His point, in essence, is that comparing AI spending to past infrastructure booms still undersells how small the entire Earth economy is next to what he calls a “K2 economy.”

What the Kardashev Scale Says About Earth’s Economy

The “K2” reference draws on the Kardashev scale, a framework used to rank civilizations by how much energy they harness. A Type II civilization, in this scheme, would capture nearly all the power output of its own star — an amount of energy so large it makes current human energy use look negligible by comparison. That is the backdrop for Musk’s claim that “an Earth economy is less than a trillionth the size of a K2 economy.” It is not a financial statement so much as a thought experiment about what economic scale could eventually look like if energy capture expanded from a planet to an entire star system.

Space-Based Industry and Economic Expansion Plans

This is not the first time Musk has framed the conversation this way. He has previously argued that SpaceX could eventually outvalue Earth’s economy by tapping vastly more solar energy than humanity currently uses, with space-based industry growing large enough to dwarf today’s global output. His nearer-term ambitions are considerably more grounded. In June, Musk outlined a path toward what he described as quadrillionaire status built around factories on the Moon and Mars. Even that vision, ambitious as it sounds, would still fall far short of capturing anything close to a full star’s energy output — the defining feature of a true K2 civilization.

Investment Risks and Market Concerns

Not everyone watching the AI buildout is focused on interstellar comparisons. Several voices in finance are raising more immediate concerns about what happens if the current spending cycle doesn’t deliver the returns investors are pricing in.

Warnings From Ray Dalio and Others

Stijn Van Nieuwerburgh, whose research underpins the $8.2 trillion buildout estimate, has warned that if enthusiasm for AI infrastructure fades, losses could spread beyond tech balance sheets into pension funds and other lenders exposed to the financing behind these projects. Investor Ray Dalio has issued a separate warning about AI bubble liquidity risks. Other skeptics have pointed to circular funding arrangements within Big Tech, where companies invest in each other’s AI ventures in ways that can inflate perceived valuations without necessarily reflecting underlying demand.

Implications for Investors and Future Prospects

Put together, the two threads in this story — record-breaking AI infrastructure spending on one side, and a Kardashev-scale vision of civilization-wide energy capture on the other — frame a real tension for anyone trying to size up where AI investment goes next. The historical comparison suggests the current buildout is already unprecedented relative to the US economy; Musk’s framing suggests, at least rhetorically, that it is still tiny compared to what a truly star-powered economy could look like.

For investors weighing near-term decisions, the practical question is simpler and far more immediate: can the returns generated by data centers, power infrastructure, and IT equipment justify spending that already exceeds the railroad and highway eras as a share of GDP? Musk’s own nearer-term roadmap — factories on the Moon and Mars — is itself a modest step relative to his long-term vision, underscoring just how wide the gap is between what’s being built today and the scale he’s describing when he talks about a K2 economy.

FAQ

How much is AI infrastructure spending projected to represent of US GDP?

AI infrastructure spending is projected to average 3.63% of US GDP between 2025 and 2032.

How does projected AI infrastructure spending compare to past US infrastructure investments?

The projected AI spending share surpasses historical booms like railroads, which averaged 2.24% of GDP, and highways, which reached 1.13%.

What does Elon Musk mean by comparing Earth’s economy to a K2 economy?

Musk states Earth’s economy is less than a trillionth the size of a K2 economy, a theoretical civilization on the Kardashev scale that harnesses almost all the energy produced by its star.

What risks are associated with the large AI infrastructure investments?

There are warnings that losses could spread to pension funds and other lenders if enthusiasm fades, alongside broader concerns about a potential AI bubble and circular Big Tech funding deals.

Article produced with the assistance of artificial intelligence and reviewed by the editorial team.