Artificial intelligence infrastructure is radically reshaping United States construction spending. It is driving unprecedented expansion in data center projects while several other vital building sectors contract significantly. Recent economic data compiled by Harvard Kennedy School economist Jason Furman and highlighted in an analysis by writer and podcaster Derek Thompson reveals a stark divergence in the modern American economy. In this shift, explosive tech demand stands in sharp contrast to declines across traditional building markets.
According to the figures, data center construction experienced an astounding 64.5% growth rate from 2025 to 2026. This far outpaced every other category tracked in the analysis. This massive surge underscores how heavily recent gains in construction spending have become concentrated within specialized technological real estate rather than distributed broadly across the national economy.
The Expanding Divide in U.S. Construction Spending
The hyper-focus on artificial intelligence infrastructure has left traditional construction sectors struggling to maintain momentum. Economic indicators show that new energy infrastructure was the only other major category to record positive growth, increasing by 4.1%. This modest yet crucial uptick is directly tied to the surging electricity demands required to power and operate intensive AI hardware and cooling systems.
Conversely, construction activity contracted across a wide array of foundational economic sectors:
- Multi-family home building fell by 5.6%
- Healthcare facility construction tumbled by 10.1%
- Commercial retail building dropped by 10.1%
- New educational construction declined by 12.6%
- Manufacturing construction collapsed by nearly 24% from 2025 to 2026
These contractions highlight a complex macroeconomic reality. While the overall economy continues to post growth figures, that expansion is increasingly predicated on the capital expenditures of a handful of hyperscale technology companies. These firms are investing heavily in generative AI and cloud infrastructure.
The Power Grid Bottleneck and Regional Shifts
As data center investments approach historic highs, with full-year spending for 2026 projected to climb dramatically, the industry faces critical bottlenecks. Chief among them is electricity availability. A single advanced AI training facility can demand hundreds of megawatts of continuous power, straining local and regional electrical grids. Consequently, power infrastructure construction is expanding rapidly to keep pace with these immense utility loads.
Geographic hubs are also shifting to accommodate these mega-campuses. While Northern Virginia remains a primary anchor, states like Texas, Ohio, and Louisiana are seeing massive multi-gigawatt developments break ground. Projects such as Meta’s Prometheus campus in Ohio and expansive developments in Central Texas demonstrate the immense physical scale of modern data center engineering.
Market Implications and Future Outlook
The heavy concentration of capital in data center development presents both a structural economic opportunity and a notable market distorter. Essential construction materials, specialized electrical components, structural steel, and skilled labor are being absorbed by hyperscale projects. This absorption occurs at rates that constrain supply for residential and civic infrastructure.
As regulatory scrutiny and local community resistance mount, with dozens of projects facing restrictions or rejections from municipal governments, industry stakeholders must navigate increasingly complex approval timelines. Whether this hyper-focused boom can sustain its current trajectory or eventually cool down will depend heavily on enterprise AI adoption rates, power grid modernization efforts, and broader macroeconomic stability.
