The simple version

The Census Bureau releases its construction spending report on Tuesday morning, covering July. The series has a formal name that explains itself once you slow down on it: the value of construction put in place.

Put in place means the work actually performed at the site during the month. Not projects approved, not projects begun, not projects announced. A tower under construction contributes in every month that crews build on it, in proportion to how much of it got built.

That makes it a different kind of number from the housing statistics it sits alongside, and the difference is the whole reason it exists.

The numbers

• The July 2026 construction spending report is scheduled for release Tuesday, September 1, 2026, at 10:00 a.m. Eastern (U.S. Census Bureau economic indicator calendar)
• The most recent report came out Monday, August 3, 2026, at 10:00 a.m., covering June 2026, release number CB26-126 (Census)
• Census defines the value of construction put in place as a measure of the value of construction installed or erected at the site during a given period (Census, construction spending definitions)
• The total for a period is the sum of the value of work done on all projects underway during that period, regardless of when each project started or when contractors were paid (Census definitions)
• June 2026 total construction was at a seasonally adjusted annual rate of $2,166.5 billion, 0.1 percent (plus or minus 0.8 percent) below the revised May estimate of $2,168.5 billion (Census)
• Private construction was at an annual rate of $1,622.5 billion in June, split between residential at $877.1 billion and nonresidential at $745.3 billion (Census)
• Public construction was at an annual rate of $544.1 billion in June, including educational at $113.1 billion and highway at $150.9 billion (Census)
• Across the first six months of 2026, construction spending amounted to $1,046.9 billion, 3.5 percent (plus or minus 1.0 percent) below the $1,084.5 billion for the same period in 2025 (Census)
• Census publishes 90 percent confidence intervals and states that if a range does not contain zero the change is statistically significant, and if it does contain zero it is uncertain whether there was an increase or decrease (Census explanatory notes)
• Census states that statistics for the current month are preliminary estimates subject to revision in following months, and that it may take 2 months to establish an underlying trend for total construction and as long as 8 months for specific categories (Census)
• The figures are at an annual rate, adjusted for seasonality but not for price changes (Census)
• Census defines a data center as a building containing the hardware needed for storing, processing, and transmitting digital information, and classifies it within office construction rather than as a separate published line (Census definitions)
• Racks and servers inside a data center are excluded from the value put in place, as special purpose equipment rather than construction (Census definitions)

Four numbers, four different moments

Construction statistics confuse people because several of them describe the same building at different points in its life, and coverage rarely says which one it is quoting.

A permit is a legal step. It records that a local authority approved a project, which is a plan rather than a commitment. Permitted projects do not all break ground.

A start records that construction began. That is a stronger signal than permission, and it still says nothing about how long the work will take or how much of it happens in any particular month.

Put in place is the one that measures activity rather than intent. Census describes the total for a period as the sum of the value of work done on all projects underway during that period, regardless of when each project started or when the contractors were paid. That last clause is what separates it from a cash-flow measure: it is tracking work, not payments.

So a single large project does not arrive as one number in one month. It spreads across every month that crews are on site, which is why an announcement and its appearance in this series can be years apart.

Where a buildout would appear in the data

The report separates private construction from public, and within each it separates residential from nonresidential. Those halves are driven by different things, which is why the split is published rather than just the total.

In June, private construction ran at an annual rate of $1,622.5 billion against $544.1 billion for public. Inside the private figure, residential was $877.1 billion and nonresidential was $745.3 billion. Residential responds to mortgage rates and household formation. Nonresidential responds to business investment: warehouses, factories, offices, and power infrastructure.

The classification detail worth knowing is where a large computing buildout lands. Census does define a data center, as a building containing the hardware needed for storing, processing, and transmitting digital information. But it is a classification within office construction, not a separate line in the monthly table, so there is no published data center number to read.

There is a sharper limit than that. Census explicitly excludes racks and servers in data centers from the value put in place, treating them as special purpose equipment rather than construction. The building counts. The machines inside it do not. Any figure describing what a company spends on computing capacity is measuring something this series was never designed to capture.

The Real Cost lens on announced versus performed

The practical value here is knowing which numbers describe intention and which describe work that has already happened, because they are reported in the same paragraphs and mean different things.

• An announced investment figure is a plan, and plans are revised, delayed, and cancelled without any statistic recording that it happened
• Put in place records work performed at a site, which cannot be un-performed, though the estimate of it can be revised
• Census is direct about that second point: current-month figures are preliminary, and it says establishing an underlying trend can take 2 months for the total and up to 8 months for specific categories
• The published ranges are doing real work too, because a change whose confidence interval contains zero leaves it uncertain whether there was an increase or a decrease at all
• The figures are adjusted for seasonality but not for price changes, so a rise in the dollar value is not by itself a rise in the amount of building

That last point is the one most easily missed. This series is measured in dollars, and dollars buy less construction than they used to, so the quantity of building and the value of building are two different lines moving for two different reasons.

What this means

When a construction or housing figure appears in a story, the question that resolves most of the confusion is which moment it measures: approval, beginning, work performed, or completion. Census publishes all of them, separately, on different schedules, and the report arriving Tuesday is the one measuring work performed.

The broader habit is distinguishing announced from measured anywhere it comes up. A number describing what someone intends and a number describing what occurred are different kinds of evidence, and only one of them is a statement about the past.

What this is NOT

This is not a prediction of Tuesday's construction spending release, of construction activity, housing supply, home prices, or the economy, and no forecast or expectation figure appears here. Every figure cited is a published estimate for a stated reference month. This is not a claim about what caused any change in any category, and in particular no cause is attributed to the office category or to data center construction, which this article describes only as a matter of how Census classifies buildings. No company is named and no sector's prospects are assessed. This is not advice about buying, selling, building, or any financial decision, and it is not advice about any security or fund. Census figures are estimates subject to revision and carry published margins of error, and national totals do not describe any local market. This is not investment or financial advice of any kind.

Sources

• U.S. Census Bureau, Construction Spending: https://www.census.gov/construction/c30/c30index.html
• U.S. Census Bureau, Construction Spending definitions: https://www.census.gov/construction/c30/definitions.html
• U.S. Census Bureau, Construction Spending methodology: https://www.census.gov/construction/c30/meth.html
• U.S. Census Bureau, economic indicator release calendar: https://www.census.gov/economic-indicators/calendar-listview.html
• U.S. Census Bureau, New Residential Construction: https://www.census.gov/construction/nrc/index.html