US economic growth in the second quarter was stronger than it looks at first glance, thanks to a resilient consumer, as well as business investment, where growth broadened outside of artificial intelligence spending.
Gross domestic product growth was just 1.5% (annualized) quarter-over-quarter. But final sales to domestic purchasers, which strips out the volatile net exports and inventories categories, expanded by 3.2%. The latter is a better guide to the underlying trend in growth than the headline number.
Real GDP by Expenditure, % Quarter-Over-Quarter Growth (Annualized)
The solid growth in final sales to domestic purchasers was driven first by personal consumption, which expanded 3.2%, even as high oil prices bite household budgets. Partly this was just a rebound from a first quarter where bad weather inhibited services spending. But durables spending also jumped by 6.8%, which evinces some consumer optimism, perhaps driven by higher stock prices compared with the first quarter.
The strongest major component of GDP was nonresidential fixed investment, which grew 8.4% quarter-over-quarter. The lion’s share of growth in this category in recent years has come from AI data centers. But this quarter’s growth in nonresidential fixed investment was much broader. Information processing equipment (which captures much of the data center capital spending) did grow 8.3%, but that’s lower than the rocketing 25% year-over-year growth as of the first quarter. Hence, all capital spending by businesses excluding AI-related spending, which has decelerated since 2023, rebounded in the second quarter.
Consumers Are Spending, but Their Savings Are Dwindling
But there’s one red flag in the report. The other side of the coin of households’ resilient spending is a dwindling savings rate. The personal saving rate fell to 2.9% in the second quarter, as the jump in prices (especially energy) eroded households’ real incomes. That’s a huge shortfall compared with the 6.9% saving rate averaged over 2018-19 (immediately before the pandemic, which we take as a proxy for normal). The drop isn’t entirely attributable to the Iran war: The saving rate was just 3.9% in the first quarter, partly because real wage growth has been in a steady downtrend since 2024.
At some point, we expect consumers to pull back in order to boost lagging household savings. That’s one reason we expect GDP growth to slow through 2028, as discussed in our economic outlook.
What Investors Are Asking Us About the Economy
I recently appeared on The Morning Filter podcast with co-hosts Dave Sekera and Susan Dziubinski, where we discussed our market and economic outlooks for the remainder of 2026.
Here’s a question from someone who tuned in:
How can we know when AI is starting to affect the supply side and hence boosting GDP growth more significantly?
The backdrop here is that we’ve argued that AI won’t reliably boost GDP growth until it affects the supply side of the economy, via enhanced productivity growth.
How can we know when AI is actually boosting productivity?
Based on today’s data, it looks like labor productivity (real GDP per hour worked) is up 1.8% year-over-year on average in the first half of 2026. That’s a strong figure, in line with the 2020-25 average of 1.9%. We’ve entered a regime of higher productivity growth, after averaging a more meager 1.0% over 2010-19. But this productivity growth acceleration began well in advance of the mass uptake of LLMs in 2024 and 2025.
If the new generation of AI technologies were having a major impact on GDP growth, we’d expect to see productivity growth accelerate further beyond that 2020-25 average. So far, have not seen that.
