Data center construction ran at $85.0 billion in August 2026, 63% of the $134.1 billion spent on US private office construction, according to the Census Bureau’s latest construction spending data, at seasonally adjusted annual rates. The August figures are preliminary.
In August 2022 data centers were $13.1 billion of a $83.7 billion office total, or 16%, by our arithmetic from the same series. The Energy Policy Research Foundation (EPRINC) put it this way in a 28 October chart note: “Data centers have gone from a minor part of U.S. office construction to its main driver.”
Business Pill · Grid Cost Allocation Explained
An evergreen Business Engineer explainer on who pays for grid upgrades. It teaches the general idea only and does not discuss the Census construction data in this piece.
Animated: data center share of US private office construction spending every three months, Census C30 data; shares are our arithmetic, August 2026 preliminary. Chart: FourWeekMBA.
The key insight: Data centers were $85.0 billion of the $134.1 billion of US private office construction in August 2026, 63% by our arithmetic from Census data, up from 16% in August 2022, while office building excluding data centers fell by about 30%.
The Census Bureau counts data centers as a line inside private office construction. In December 2019 that line was $7.9 billion of an office total of $80.0 billion, or about 10%.
The share first passed half in November 2025, at 50.3% by our arithmetic. By December 2025 data centers were $55.5 billion of $107.1 billion.
The pace has picked up in 2026. Data center spending rose from $65.7 billion in May to $85.0 billion in August, and from $79.0 billion in July, a revised figure. By our arithmetic, August was 73% above August 2025’s $49.1 billion and about 6.5 times August 2022.
US private office construction spending split into data centers and other office, December 2019, August 2022 and August 2026, seasonally adjusted annual rates. Source: US Census Bureau (C30); August 2026 preliminary. Other office and shares are our arithmetic.
The Rest of Office Shrank
Office construction excluding data centers was $70.5 billion in August 2022 and $49.2 billion in August 2026, about 30% lower, by our arithmetic. EPRINC says the decline came amid “elevated post-pandemic vacancies and higher borrowing costs.”
Over the same four years, total office spending rose by $50.4 billion while data centers rose by $71.8 billion, so data centers account for more than all of the growth, by our arithmetic.
Where the Power Demand Shows Up
EPRINC wrote that “Virginia and Texas stand out”: commercial electricity demand there has grown by 32.6 million and 32.7 million megawatt-hours respectively since 2019. In Virginia’s case, EPRINC says, that is about 3.7 GW of round-the-clock load.
Citing the Server Country project tracker, EPRINC lists $234.5 billion of announced projects in Virginia that would need another 19.2 GW, and $177 billion in Texas that would need another 42.9 GW.
The note names eight more states in the tracker’s top 10, including Missouri at $207.5 billion and 9.0 GW, Utah at $181.3 billion and 26.9 GW, and Pennsylvania at $128.6 billion and 24.1 GW. It adds that Louisiana’s total is led by Meta’s Hyperion campus, at 5 GW on its own.
As we read the Census table, the office category is now mostly a data center category: the data center line has been more than half of office spending every month since November 2025.
By our arithmetic, data centers rose by $71.8 billion between August 2022 and August 2026 while the office total rose by $50.4 billion, so office building outside data centers contracted over the period.
As we read EPRINC’s note, Virginia and Texas are where the power demand has already appeared, while the tracker’s announced projects point to Missouri, Utah, Pennsylvania and others for the next wave.
EPRINC, 28 October 2026
“Data centers have gone from a minor part of U.S. office construction to its main driver.”
Three Implications
FOR CONSTRUCTION SUPPLIERS Data center construction ran at $85.0 billion in August 2026, 73% above August 2025, by our arithmetic from Census data.
FOR OFFICE LANDLORDS AND LENDERS Office construction excluding data centers fell from $70.5 billion in August 2022 to $49.2 billion, by our arithmetic; EPRINC cites elevated post-pandemic vacancies and higher borrowing costs.
FOR GRID PLANNERS EPRINC lists $177 billion of announced projects in Texas needing another 42.9 GW and $234.5 billion in Virginia needing another 19.2 GW, citing the Server Country tracker.
The framework puts it this way: “But infrastructure reality is catching up with AI ambitions: a projected 19GW gap between new data center demand and available grid capacity threatens to deflate the AI bubble.”
As we read it, the Census construction data show the dollars of that build-out already flowing: data centers are now 63% of US private office construction spending.
August 2026 Census figures are preliminary and the July figure is revised; both can change. EPRINC notes that the tracker’s announced figures “are not directly comparable to the Census series”, since they cover multiyear buildouts and can include IT equipment. The shares, growth rates and multiples here are our arithmetic from the Census table.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Data centers were 63% of US private office construction spending in August 2026, up from 16% in August 2022, while the rest of office building fell by about 30%, according to Census data and our arithmetic.
A note on sourcing. Spending figures come from the US Census Bureau’s Value of Construction Put in Place private, seasonally adjusted table (August 2026 preliminary, July revised); state power and project figures come from EPRINC’s 28 October chart note, which cites the Server Country tracker. Shares, growth rates and multiples are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
