The NY Fed checked whether the AI boom widened the US trade gap. Mostly, it didn't.

The AI buildout is easy to see in import data. A new Liberty Street Economics post from the Federal Reserve Bank of New York, published October 14, 2026 by Julian di Giovanni, starts there: imports of semiconductors, servers and computing equipment have climbed sharply, yet the US current account deficit has been largely unchanged. The post asks what the national accounts say about that gap between a loud import story and a quiet balance.

The accounting identity doing the work

The post leans on a simple identity. The current account balance equals a country's gross saving minus its gross domestic investment. A country can import a great deal of capital equipment without running a bigger external deficit, as long as it saves enough to pay for the investment itself. So for the AI boom to widen the deficit, it would have to raise investment without a matching rise in saving.

The author points to two ways that could happen. Equity gains could lift household spending (a wealth effect), or business saving could fall. Both moved together during the dot-com era. The post asks whether either is happening now.

Investment: reallocated, not expanded

The headline finding is that the aggregate investment rate barely moved. The table below gathers the post's main ratios, all expressed as shares of GDP.

The post says AI investment has not been small and has been rising, but it has coincided with weakness elsewhere. The surge in factory construction that followed the 2022 industrial policy legislation has faded, investment in structures and equipment outside the technology complex has been soft, and residential investment has been declining recently. Data centers and information-processing equipment helped take up the slack, not enough to lift aggregate investment relative to national saving.

Saving: not the dot-com pattern

During the dot-com expansion, the post notes, private saving fell by nearly 3 percentage points of GDP between 1995 and 2000 as households spent out of rising equity wealth.² Nothing comparable shows up today. The half-point drop in private saving since the pandemic is attributed as much to the unwinding of savings accumulated during the pandemic as to anything tied to AI.

Business saving is the other contrast. The dot-com buildout was concentrated in young telecom and equipment firms with thin or negative operating cash flow. The AI buildout is concentrated in what the post calls the most cash-generative firms in the economy, and business saving has held up through the capex surge. The post puts the dot-com trough at roughly 8.8% of GDP in late 2000 and early 2001.³

Where the wider deficit comes from

If AI is not the driver, what is? The post decomposes the roughly 2.4 percentage point widening in the saving-investment gap relative to the fourth quarter of 2019. Government dissaving accounts for about 1.75 points, lower private saving for about half a point, and higher investment for essentially nothing.⁴ Government saving fell from about -2.2% of GDP in 2015 to 2019 to about -4.0% since 2023.⁵ The author's reading is that fiscal policy is the major driver of current account dynamics, with no discernible investment channel.

The caveat that matters

The post is explicit that this is a snapshot, not a forecast. Announced capital spending plans of the major technology firms for 2026 and 2027 are considerably larger than what has been built so far. The offsetting weakness in residential and non-technology investment will not last forever. If those drags fade while AI capex accelerates, the post says, higher investment and lower business saving would both push the deficit wider.

It is also an opinion piece by one staff economist, and it carries the standard disclaimer that the views are the author's and not necessarily those of the New York Fed or the Federal Reserve System. It builds on NBER research, linked from the post, showing AI-related goods reached a high of 23% of US imports in 2025.⁶ We did not open that paper, so we treat the import share as the New York Fed's citation of it.

What to take from it

Trade data and national accounts answer different questions. A surge in gross imports tells you about where a boom is spending. The balance tells you who is funding it. For AI, the funding so far looks domestic and corporate: cash-rich firms are paying for the buildout out of their own saving, while the fiscal position is what widened the external gap. Watching whether business saving starts to slip is the cleaner signal to follow than watching import totals.