Technology Shocks, the AI Boom, and the U.S. Current Account

July 14, 2026

Technology Shocks, the AI Boom, and the U.S. Current Account

Giuseppe Fiori, Colleen Lipa, and Erik Nuenninghoff

The current artificial intelligence (AI) investment boom in the United States provides a powerful boost to imports of high-technology capital goods. The AI buildout bears the hallmarks of an investment-specific technology shock-a process in which rapid technological progress makes each new generation of capital equipment significantly cheaper and more powerful than the last, but where reaping those efficiency gains requires continuous and substantial investment to acquire and deploy the new vintage of capital goods.

Empirical analysis suggests that investment-specific technology shocks are associated with a persistent current account deterioration of roughly 10 percent relative to its historical average and rising import prices. With regard to the investment boom related to AI in the United States, these patterns could be more pronounced than in previous episodes given the geographic concentration of key inputs-approximately 90 percent of the relevant equipment is sourced from East Asia.

Two risks warrant attention. First, the historical pattern of rising import prices following investment-centered technology booms could pose greater risk to inflation since AI-related input prices have been on the rise in the last two years, after decades of decline. The price of computers and semiconductors has turned upward since the pandemic, suggesting that the cost of sustaining the AI buildout may prove higher than past experience would imply. Second, the persistence of the AI buildout is underpinned by a broad, ongoing race to accumulate computing infrastructure, with no clear ceiling on scale or duration. The resulting drag on the current account-through sustained import demand for semiconductors and related equipment-may prove considerably more durable than past episodes would suggest. The note proceeds as follows: Section 1 documents the import footprint of the AI boom. Section 2 identifies U.S. technology shocks using a structural VAR, and Section 3 presents the resulting impulse responses. Section 4 traces the international spillovers of U.S. investment-specific shocks, and Section 5 examines the risks arising from foreign price pressures and persistent current account imbalances.

1. Motivation: The AI Boom and Its Import Footprint

The AI investment boom in the United States depends critically on imports of advanced inputs with approximately 90 percent of equipment goods for high-technology sectors originating abroad, as shown in Figure 1. This dependence on foreign inputs is more pronounced for high-tech equipment than for equipment more broadly. In addition, the foreign suppliers of these inputs are concentrated in East Asian economies which have specialized in the production of semiconductors and related hardware.

Given this dependency on imported inputs, the domestic AI buildout should cause the U.S. external balance to deteriorate while stimulating demand in the supplying foreign economies. To assess whether these observations are consistent with the macroeconomic experience of past technology-driven investment cycles, we conduct an empirical analysis using historical data from the 1990s through 2019.

2. Identifying U.S. Technology Shocks: A Structural VAR Approach

To isolate the macroeconomic effects of U.S. technology shocks, we estimate a structural VAR model using U.S. quarterly data over the period 1993-2019. The approach decomposes the co-movement of key economic variables-the growth rates of output, the price of capital goods relative to consumption goods, labor productivity, import prices, and the current account-into distinct drivers. Following the seminal works of Galí (1999) and Fisher (2006), we distinguish between two types of technology shocks: investment-specific technology shocks lower the relative price of capital goods persistently whereas neutral technology shocks raise overall productivity persistently without changing the relative price of capital goods. We identify these two shocks using long-run restrictions: investment-specific shocks are identified as the only shock with a permanent effect on the relative price of capital goods, while neutral technology shocks are identified as the shock, orthogonal to the investment-specific shock, with a permanent effect on labor productivity. All other shocks are left unrestricted.

The distinction between investment-specific and neutral technology shocks maps naturally onto the two phases of a technology cycle: the former drives the costly buildup of infrastructure and capital, depressing the relative price of investment goods, while the latter governs the broader diffusion of productivity gains across the economy. The current AI boom - marked by surging expenditure on computers, chips, and data centers and sustained decline in the price of AI hardware, semiconductors, and computing equipment relative to consumption goods - bears the hallmarks of the buildup phase, suggesting that investment-specific forces are presently dominant. The second phase of technology cycle should eventually manifest as neutral technology improvements, without systematically altering the relative price of capital.

Investment-specific technology shocks reflect technological improvements that are embodied in new capital goods only. Thus, in order to harvest the technological advancements, firms must install new capital which raises investment spending. By contrast, neutral technology shocks capture broader advances in productive efficiency that raise output without systematically altering the relative price of capital.

In what follows, we focus on investment-specific shocks as they account for approximately 60 percent of the variance in the current account in our structural VAR. Neutral technology shocks, by contrast, do not trigger significant external adjustment: because they raise productivity broadly rather than requiring a buildout of capital, they do not generate the investment-saving imbalance that drives current account movements. This further supports our focus on investment-specific shocks.

Before turning to the estimation results, we confirm that our identified technology shocks do in fact raise productivity. Examining the response of Fernald's (2014) TFP series-a standard benchmark for productivity measurement in the U.S., adjusted for the degree of input utilization-we find strong support for the identification scheme. Following an investment-specific shock, utilization-adjusted TFP rises by approximately half a percentage point in response to a one-standard-deviation shock. This finding is consistent with the interpretation that productivity gains materialize gradually as new capital is installed and deployed, rather than accruing immediately upon impact.

3. U.S. VAR Dynamics: Impulse Responses

The structural VAR (SVAR) provides a disciplined account of the historical patterns, persistence, and magnitude of the external adjustment that follows technology shocks. Figure 2 plots the impulse response of selected variables to a one-standard-deviation investment-specific shock (continuous line), together with 84 percent confidence bands (dashed lines). Following the shock, investment (not shown) rises by approximately 2 percent over the first year, while output expands gradually by around half a percentage point as the new capital is deployed. The investment surge generates a deterioration of net exports and a widening current account deficit. Measured as a share of GDP, this deficit widens by approximately 0.2 percentage points for several quarters-equivalent to roughly 10 percent of the average deficit of 3 percent of GDP recorded over the sample period-before gradually unwinding.

These dynamics align closely with theoretical predictions. Given the high import content of U.S. equipment investment, the demand for foreign goods rises sharply, with imports outpacing exports. In addition, the historical experience suggests that import prices rise by approximately 1 percentage points over the first year following the shock-a pattern that helps account for both the speed and the magnitude of the current account deterioration observed in the data.

4. International Spillovers of U.S. Investment-Specific Shocks

Through raising demand for foreign goods, U.S. technology shocks have significant effects on economic activity and inflation abroad. Spillover effects on technology suppliers exceed what aggregate trade linkages alone would predict. Countries such as Taiwan and Korea - whose economies are oriented toward manufacturing and concentrated in technology-intensive sectors - are affected more strongly than their overall trade exposure to the United States would suggest. By contrast, close trading partners such as Canada, despite deep aggregate trade ties, display positive but comparatively modest responses. This contrast points to a transmission channel that operates specifically through trade in technology goods, amplified by the sectoral structure of the recipient economy, rather than through broader aggregate demand spillovers.

We quantify these effects through country-by-country regressions of GDP and inflation on U.S. investment-specific shocks, controlling for the first four lags of domestic GDP and inflation as well as U.S. and core inflation. The results, summarized in Figure 3, show that increased U.S. demand for high-technology capital goods provides a meaningful boost to the manufacturing sectors in exporting nations, primarily East Asian economies and Mexico. Mexico's response is comparable in magnitude to Taiwan's and Korea's, despite occupying a different position in U.S. capital-goods supply chains: Mexico's manufacturing base is integrated into U.S. production as an assembly and component stage, while Taiwan and Korea specialize in semiconductor fabrication and other upstream technology-intensive inputs. One possible interpretation is that a U.S. investment-specific shock propagates through the supply chain at multiple stages simultaneously, generating similar real activity effects in countries that supply different parts of the same chain, even as the composition of their exposure differs. The inflationary effects are modest, with the exception of economies with highly concentrated technology export sectors, such as Taiwan.

These spillover dynamics have important implications for the persistence of the current account adjustment. As foreign economies expand alongside the U.S., their demand for American exports will rise over time, providing a partial offset to the initial trade balance deterioration. However, the asymmetric structure of the AI supply chain-in which the U.S. imports capital-intensive goods and exports knowledge-intensive services-suggests that this rebalancing may be slow to materialize.

5. Risks from Foreign Price Pressures and Persistent Current Account

The AI-investment boom with the distinct surge of U.S. imports of high-technology inputs from Mexico and East Asian suppliers appears consistent with the notion of technology shocks that require investment into new capital (investment-specific shocks). Following our historical estimates, the economic effects are likely to be persistent.

We conclude by highlighting two risks associated with the AI boom and its consequences for the external balance. The first concerns import prices. Historically, investment-specific shocks have been associated with higher aggregate import prices. While prices of computers and semiconductors had been on a declining trend until 2019-even after controlling for quality improvements-this trend slowed in the post-pandemic period and has since reversed, with these prices now posting positive growth rates. As shown in Figure 4, computers and semiconductors, which currently account for about 16 percent of U.S. goods imports, have seen prices rise relative to their pre-pandemic trend and in excess of overall import price inflation. As the AI buildout continues, the risk of further price pressures in this segment warrants close attention.

The second risk concerns the persistence of the current account deterioration. While the historical estimates suggest that the external deficit associated with a technology boom gradually unwinds over several quarters, the current episode may prove more protracted. The surge in equipment imports driven by AI infrastructure investment shows little sign of abating-major technology firms have signaled sustained and growing capital expenditure plans, suggesting that the import-intensive phase of the AI buildout is far from complete. This implies that the contribution of AI-related investment to the current account deficit may accumulate over a longer horizon than previous technology episodes would suggest, raising the prospect of a more significant and prolonged external imbalance than our historical estimates alone would indicate.

References

Fernald, J. (2014). A quarterly, utilization-adjusted series on total factor productivity. Federal Reserve Bank of San Francisco Working Paper 2012-19.

Fisher, J. (2006). The dynamic effects of neutral and investment-specific technology shocks. Journal of Political Economy, 114(3), 413-451.

Galí, J. (1999). Technology, employment, and the business cycle: Do technology shocks explain aggregate fluctuations? American Economic Review, 89(1), 249-271.

Runkle, D.E. (1987). Vector autoregressions and reality. Journal of Business & Economic Statistics, 5(4), 437-442.

Appendix

A. Figure 1: Import share of equipment and high-tech equipment calculation.
Gross import share - the gross import share is calculated with the following equation:

$$$${S}^{G}=\frac{{i}^{{ht}}}{{I}^{{ht}}}$$$$

Where $$S^G$$ is the gross import share, $$i^{ht}$$ is the imports of high-tech equipment, and $$I^{ht}$$ is non-residential private fixed investment in high-tech. Note that in the equation for net import share of equipment, $$i$$ and $$I$$ are the imports of equipment and non-residential private fixed investment in equipment, respectively.

Net import share of high-tech equipment - the net import share is calculated with the following equation:

$$$${S}^{N}=\frac{{i}^{{ht}}-\ {e}^{{ht}}}{{I}^{{ht}}}$$$$

Where $$S^N$$ is the net import share, $$i^{ht}$$ is the imports of high-tech equipment, $$e^{ht}$$ is the exports of high-tech equipment, and $$I^{ht}$$ is non-residential private fixed investment in high-tech. Note that in the equation for net import share of equipment, $$i$$, $$e$$, and $$I$$ are the imports of equipment, exports of equipment, and non-residential private fixed investment in equipment, respectively.

B. Data Collection for VAR
a. The United States data used in the SVAR is collected from the BLS, BEA, and FRB staff calculations. The data collected from the BLS is the producer price index of investment, the consumer price index of non-durables, and the consumer price index of services. The current account as a percentage of GDP, the value of GDP, and import prices comes from the BEA.

C. Variable creation for VAR
Relative price of investment - the relative price of investment is calculated with the following equation:

$$$$P^I\ =\ \frac{{PPI}_{investment}\ }{{CPI}_{non-durables\ +\ services}}$$$$

Where $$P^I$$ is the relative price of investment. The change in the relative price of investment as calculated as $$P_t^I-P_{t-1}^I$$

Output per capita - output per capita is calculated with the following equation:

$$$$Y_t\ =\frac{GDP}{p_t}$$$$

Where $$Y_t$$ is output per capita and $$p_t$$ is civilian noninstitutional population 16 years and over.

Growth rate of total import prices - the growth rate of total import prices is calculated with the following equation:

$$$${G}\ =\ \frac{P_t^i-P^i_{t-1}}{P^i_{t-1}}$$$$

Where G is the growth rate and $$P^i$$ is the import price.