Alternative data sources point to ongoing demand for data centers.
The federal government shutdown that started in October had an immediate impact on construction data and left data consumers looking more eagerly than usual for alternative sources of information. Although several such sources exist, none is an exact replacement for data provided by the U.S. Census Bureau.
The first set of numbers to go missing was the Census Bureau’s monthly report on “value put in place” or spending on construction projects underway in August. That report, which was due Oct. 1, covers more than 170 types of nonresidential projects.
ConstructConnect and Dodge Construction Network produce two other snapshots of construction spending, with both firms privately collecting detailed information on projects. Around the 20th of each month, they put out releases providing estimates of the total value of projects they deem to have started in the previous month. These estimates can offer an early indication of turning points for various segments, since the starts (or lack thereof) affect spending totals over the next several years as construction progresses.
There is no simple formula for translating starts into spending put in place, however. There is wide variation in how long a start takes to complete, even assuming it proceeds without interruption. The final value may differ substantially from the estimated initial value. And the categories used by the two firms differ from the Census Bureau’s classifications.
Furthermore, the Dodge and ConstructConnect numbers frequently point in opposite directions or at least have different magnitudes. Even if the firms are working from the same universe of projects (which is unlikely), they may have different information or interpretations of when a project started or what amount constituted construction costs.
For instance, in late October, Dodge reported that the value of multifamily starts for January through September 2025 had climbed 13% compared with the same months in 2024. In contrast, ConstructConnect estimated year-to-date apartment starts had slipped 6% from 2024.
ConstructConnect included more detail in its release. (Both firms sell custom tabulations by project type, value, square footage, geography and other characteristics.) The firm estimated commercial starts for 2025 had climbed 5.2% year to date.
The commercial starts were a mixed bag: Warehouse starts were down 17.3%, and hotel/motel starts were down 21%. In contrast, retail/shopping starts edged up 0.8%, and “office” starts soared 38.1%. The office category includes data centers, however. Traditional office starts have declined year to date, while data center starts appear regularly among the firm’s top 10 list of biggest starts in some months.
Both firms produce more speculative measures of future activity. ConstructConnect calculates a U.S. Expansion Index that “compares the total dollar value of construction projects in planning for the current month to the value of projects in planning for that same calendar month in the prior year.” The October reading “reflected an 8% year-over-year increase in planned construction investment, continuing three months of consecutive growth. Industrial, Military and Commercial sectors saw notable expansion, while Educational, Residential, Community and Medical sectors faced ongoing declines in planned investment.”
The Dodge Momentum Index, released Oct. 7, is “a monthly measure based on the three-month moving value of nonresidential building projects going into planning, shown to lead construction spending for nonresidential buildings by a full year to 18 months.” Dodge reported that in September, commercial activity “slowed down for warehouses, traditional office buildings and hotels but gained momentum in data centers and retail stores. Without data centers, commercial planning would have only increased 0.5% this month.”
The takeaway: Both firms’ indicators suggest there has been no letup in demand for data centers, despite some pundits’ qualms that the sector is in a bubble that will inevitably burst. Other nonresidential categories still appear to be weak, while reports differ as to whether multifamily activity has pulled out of its nosedive.
Ken Simonson is the chief economist with the Associated General Contractors of America. Contact him at ken.simonson@agc.org.
