Treasury told its borrowing committee that artificial intelligence accounted for nearly 30 percent of GDP growth through 2025, and listed it among the risks in the same document
The quarterly document Treasury writes for the dealers who buy its debt is not usually where the federal government puts a number on artificial intelligence. Monday's does.
The economy statement released at 3 p.m., alongside the department's borrowing estimates for the quarter, says AI-driven investment accounted for nearly 30 percent of GDP growth throughout 2025, and that the trend continued through the first two quarters of 2026. It defines the measure in the same breath: business fixed investment in software, in computers and hardware, and in data centre structures, measured relative to pre-large-language-model trends.
That is a counterfactual, not a direct reading. The statement does not print the trend it is measuring against, and it does not show the calculation.
The same document files it under risk
Four items appear under the statement's medium-run risks to the outlook. Energy prices and geopolitical uncertainty, labour markets, business fixed investment and the manufacturing outlook, and artificial intelligence.
On the last of them the department is careful. It says the timeline and magnitude of productivity gains is uncertain, as is the ultimate effect on the composition of labour markets. It sets out two paths, one in which the technology behaves like a standard improvement and productivity growth returns to historical norms after a period above trend, and one in which it proves transformational and permanently lifts the path of productivity. It then adds a third possibility in a single sentence, that artificial intelligence could have disruptive impacts on the economy and labour markets, with firms and workers slow to adapt left at a competitive disadvantage.
The rest of the numbers
Real GDP grew 1.5 percent at an annual rate in the second quarter on the advance estimate, after 2.1 percent in the first. The statement argues the headline understates underlying demand, and gives the figure it prefers: private domestic final purchases, meaning consumption plus business fixed investment plus residential investment, rose 3.9 percent at an annual rate, which it calls the strongest pace in over three years.
Net exports subtracted 1.0 percentage points from growth. Inventories and total government spending together subtracted 0.8 points, and the statement attributes the government line largely to an accounting identity, because sales from the Strategic Petroleum Reserve count as demand elsewhere in the accounts.
On the labour market it reports 334,000 net new jobs in the second quarter, unemployment claims in mid-July at the lowest since 1969, and a private-sector layoffs and discharges rate of 1.1 percent in June. The median private-sector hires rate has been 3.6 percent for two years, and the statement draws the uncomfortable inference itself rather than leaving it: low turnover means less slack to absorb a shock, so a large one could push the market out of balance quickly.
What this document is
It is worth saying plainly what a reader is holding. This is Treasury's own account, written under the current administration for the administration's own advisory committee, and it reads that way. It opens by calling the economy strong under the Trump Administration, credits the tax law it names the Working Family Tax Cuts for a 6.1 percent rise in fixed business investment through the second quarter, and credits administration policy for the economy's resilience to oil supply shocks.
None of that makes the figures wrong. It does mean the framing is the department's and is reported here as the department's.
Second-quarter productivity data, which the statement flags as the next test of its capital expenditure argument, is due on Thursday 6 August.
