AI Infrastructure Spending Surpasses Housing Investment in US Economy
Data center investments now exceed residential spending, signaling a major economic shift.
1Historic economic shift
For the first time, US spending on information processing equipment ($752B) has surpassed residential investment ($748B), marking a pivotal transition from housing-led growth to AI-driven expansion. This reversal comes after residential investment dropped 18% from its 2021 peak while tech spending surged 51% during the same period. The San Francisco Fed highlights this as a structural change in investment priorities, with hyperscalers leading the charge through massive data center construction.
2Trillion-dollar AI race
Major tech firms are projected to spend $1.3 trillion annually by 2027 on AI infrastructure, up from $470B in 2025. Alphabet, Amazon, Microsoft, Meta, Oracle and SpaceX are driving this unprecedented capex growth that currently outpaces revenue. S&P Global warns this aggressive build-out risks overcapacity if demand doesn't materialize as expected, with 2028 seen as a potential inflection point when capex may stabilize against accelerating revenues.
3Interest rate divergence
The housing market remains highly sensitive to borrowing costs, with 30-year mortgage rates near 7% as Treasury yields hit 2007 highs. This has created a 'lock-in effect' where homeowners refuse to sell, constraining supply. Conversely, AI investment shows remarkable rate resilience - hyperscalers continue borrowing aggressively despite higher costs, with Alphabet even reporting negative cash flow. Treasury Secretary notes companies are becoming 'yield-agnostic' given anticipated AI returns.
4Construction sector split
While multifamily housing starts dropped 2.6% to 1.275 million annualized in August, single-family homes saw limited gains. Builder sentiment hit yearly lows as high rates depress demand and construction costs rise. Meanwhile, data center construction booms with no signs of slowing - the six largest hyperscalers will collectively have negative operating cash flow in 2026-27 due to relentless infrastructure spending.
5Political and social backlash
64% of voters oppose local data center construction per NBC polling, creating midterm election headwinds. AI expansion also exacerbates cost-of-living pressures through higher electricity bills and tech product prices. This compounds housing affordability crises, as limited supply keeps prices elevated despite weakened demand - existing home inventory remains constrained while new construction lags.
6Fed policy implications
The Fed's inflation-fighting rate hikes have disproportionately impacted housing while barely slowing AI investment. This divergence challenges traditional monetary policy transmission mechanisms, as the new growth engine (tech) proves less rate-sensitive than the old (housing). Capital Economics predicts further housing start declines as borrowing costs keep rising, with no similar brake on tech capex.
7Corporate debt dynamics
Hyperscalers are increasingly funding AI builds through debt issuance rather than cash reserves, despite higher yields. This 'build at any cost' mentality reflects conviction in AI's long-term returns, with companies willing to accept near-term negative cash flows. The trend mirrors early 2000s telecom infrastructure overbuilds, though current players have stronger balance sheets to sustain losses longer.
8Structural economic transformation
The AI infrastructure boom represents more than sector rotation - it's reshaping US growth drivers as profoundly as the housing boom did pre-2008. Unlike housing's cyclical nature, AI investment appears more durable given its global strategic importance. However, risks loom if demand fails to justify capacity, potentially creating tech-specific overinvestment bubbles while housing shortages persist.
Original article: Fortune — AI Infrastructure Spending Surpasses Housing Investment in US Economy