The Trade Deficit Shrunk $189.3 Billion as Four Tech Import Lines Rose $183.5 Billion

By Cumulant Research

June produced a simple headline: the monthly U.S. goods-and-services deficit narrowed from $77.6 billion to $73.3 billion because imports fell more than exports. The larger result sits behind that monthly move. Across January-June 2026, the deficit was $189.3 billion smaller than in the same months of 2025 even though imports were $9.0 billion higher. Exports rose $198.3 billion, more than absorbing the importimportAn import is a good or service purchased by a U.S. resident from a resident of another country. [BEA](https://www.bea.gov/data/intl-trade-investment/international-trade-goods-and-services) increase. [Census and BEA](https://www.bea.gov/news/2026/us-international-trade-goods-and-services-june-2026)

That is an accounting result, not a verdict that the economy became $189.3 billion stronger. The headline figures are seasonally adjustedseasonally adjustedSeasonal adjustment removes recurring calendar patterns so different months or quarters are easier to compare. [FT-900 explanatory notes](https://www.census.gov/foreign-trade/Press-Release/ft900/ft900_2606.pdf) current-dollar values, so they combine changes in prices and physical quantities. [FT-900](https://www.census.gov/foreign-trade/Press-Release/ft900/ft900_2606.pdf)

The broad totals conceal unusually large movements in a few import rows. Pharmaceutical-preparation imports were $89.82 billion lower and finished-metal-shape imports were $76.27 billion lower than in the first half of 2025. Holding everything else constant, those two declines improved the goods balance by a combined $166.09 billion. [FT-900 Exhibit 8](https://www.census.gov/foreign-trade/Press-Release/ft900/ft900_2606.pdf)

Technology hardware moved in the opposite direction. Computer imports rose $95.36 billion, computer-accessory imports rose $37.17 billion, telecommunications-equipment imports rose $27.02 billion and semiconductorsemiconductorA semiconductor is a material with controllable electrical conductivity that forms the basis of electronic chips. [NIST](https://www.nist.gov/semiconductors) imports rose $23.97 billion. Together, the four rows increased $183.52 billion. [FT-900 Exhibit 8](https://www.census.gov/foreign-trade/Press-Release/ft900/ft900_2606.pdf)

Think of the balance as a bathtub with many separate faucets and drains. Lower pharmaceutical and metal imports slowed two large incoming streams. The four technology streams opened wider. The level still fell because exports and the other favorable flows more than offset the technology increase.

At the six-category level, food, industrial supplies, automobiles, consumer goods and the residual other-goods category together contributed $317.7 billion toward a narrower Census-basis balance. Capital goodsCapital goodsCapital goods are the FT-900 category containing equipment used in production, such as computers, industrial machinery and telecommunications equipment, while excluding automobiles. [FT-900 Exhibit 6](https://www.census.gov/foreign-trade/Press-Release/ft900/ft900_2606.pdf) moved the other way by $136.0 billion, offsetting 42.8 percent of that favorable subtotal and leaving about $181.7 billion of Census-basis goods narrowing. [FT-900 Exhibit 6](https://www.census.gov/foreign-trade/Press-Release/ft900/ft900_2606.pdf)

Capital-goods imports increased $186.50 billion. Computers, computer accessories, telecommunications equipment and semiconductors supplied $183.52 billion, or 98.4 percent, of that increase. The remaining capital-goods rows netted to an increase of roughly $2.98 billion after their gains and declines were combined. [FT-900 Exhibit 8](https://www.census.gov/foreign-trade/Press-Release/ft900/ft900_2606.pdf)

Technology hardware is our shorthand for those four separately published rows. It is not an official Census aggregate, and it is not automatically an artificial-intelligence measure. Census end-use categories group merchandise by principal use, while each broad row can contain products bought for many different purposes. [Census Bureau](https://www.census.gov/newsroom/blogs/global-reach/2012/03/end-use-trade-term-of-the-month-2.html)

Federal Reserve researchers make the same measurement warning. Computers and peripheral equipment include servers used in data centers, but also ordinary office computers. Communications equipment can include data-center networking gear as well as mobile-phone and broadcasting equipment. The researchers conclude that aggregate statistics do not contain an agreed AI-investment line item. [Federal Reserve](https://www.federalreserve.gov/econres/notes/feds-notes/the-ai-buildout-and-the-economy-publicly-available-data-to-assess-ais-impact-20260717.html)

Michael Waugh's more detailed research classifies individual HS10 customs products associated with computing, electrical, cooling and networking infrastructure. Its published monthly analysis extends through January 2026, not June. Extending that product-level method through June would be necessary before treating the latest increase as an AI-import total. [NBER](https://www.nber.org/papers/w35053), [Federal Reserve Bank of Minneapolis](https://www.minneapolisfed.org/article/2026/much-more-than-microchips-trade-soars-in-ai-related-goods-driving-us-trade-deficit)

Adding the monthly deficits in Exhibit 1 shows why the first-half result looks so large. The deficit totaled $374.76 billion in the first quarter of 2025 but $165.75 billion in the first quarter of 2026, producing $209.01 billion of year-over-year narrowing. [FT-900 Exhibit 1](https://www.census.gov/foreign-trade/Press-Release/ft900/ft900_2606.pdf)

The direction then reversed. The second-quarter deficit increased from $185.79 billion in 2025 to $205.48 billion in 2026, a $19.69 billion deterioration. Subtracting that offset from the first-quarter improvement leaves the reported first-half narrowing of $189.31 billion. [FT-900 Exhibit 1](https://www.census.gov/foreign-trade/Press-Release/ft900/ft900_2606.pdf)

This is a base effectbase effectA base effect occurs when a change appears unusually large because the earlier period used for comparison was itself unusually high or low. [Federal Reserve](https://www.federalreserve.gov/econres/notes/feds-notes/racing-against-tariffs-global-impacts-of-frontloading-20250801.html) in plain view: the first half is being compared with an exceptionally large first-quarter deficit in 2025. The second-quarter comparison shows that the favorable gap had already begun to reverse.

Federal Reserve researchers documented a surge in exports to the United States from several economies during the first quarter of 2025, concentrated in March ahead of expected tarifftariffA tariff is a tax or customs duty imposed when goods cross an international border. [U.S. Customs and Border Protection](https://www.cbp.gov/travel/international-visitors/know-before-you-visit/customs-duty-information) increases. Their sector examples included euro-area chemicals and Taiwanese information-technology products. The underlying U.S. import series in that analysis extended through May 2025. [Federal Reserve](https://www.federalreserve.gov/econres/notes/feds-notes/racing-against-tariffs-global-impacts-of-frontloading-20250801.html)

That evidence supports a timing story: some goods that otherwise might have arrived later were brought into the United States earlier. It also helps explain why comparisons with early 2025 deserve caution. But it does not provide a counterfactual showing what first-half 2026 imports would have been without front-loadingfront-loadingFront-loading means importing something earlier than planned, similar to buying next month's supplies today because a new tax is expected. [Federal Reserve](https://www.federalreserve.gov/econres/notes/feds-notes/racing-against-tariffs-global-impacts-of-frontloading-20250801.html). [Federal Reserve](https://www.federalreserve.gov/econres/notes/feds-notes/racing-against-tariffs-global-impacts-of-frontloading-20250801.html)

The $89.82 billion fall in pharmaceutical-preparation imports is therefore consistent with comparison against an elevated 2025 period, but the FT-900 does not identify how much was caused by tariff timing, prices, inventory decisions, country composition or other factors. [FT-900 Exhibit 8](https://www.census.gov/foreign-trade/Press-Release/ft900/ft900_2606.pdf)

The commodity analysis uses Census-basis goods data because that is where the detailed import and exportexportAn export is a good or service sold by a U.S. resident to a resident of another country. [BEA](https://www.bea.gov/data/intl-trade-investment/international-trade-goods-and-services) rows appear. The headline adds two layers. First, BEA adjusts the customs data to a balance-of-payments basisbalance-of-payments basisBalance-of-payments data adjust customs figures so they better represent transactions between U.S. residents and foreign residents. [BEA](https://www.bea.gov/news/2026/us-international-trade-goods-and-services-annual-revision), which reduced the calculated narrowing by about $2.1 billion. Second, the services surplusservices surplusA services surplus means the value of services exported exceeded the value of services imported. [FT-900 Exhibit 1](https://www.census.gov/foreign-trade/Press-Release/ft900/ft900_2606.pdf) was about $9.7 billion larger than a year earlier. Those steps take the result from roughly $181.7 billion of Census-basis goods narrowing to $189.3 billion for goods and services. [FT-900 Exhibits 1, 5 and 6](https://www.census.gov/foreign-trade/Press-Release/ft900/ft900_2606.pdf)

A narrower trade deficittrade deficitA trade deficit means the country imported more goods and services than it exported during the measured period. [BEA](https://www.bea.gov/data/intl-trade-investment/international-trade-goods-and-services) is not, by itself, a complete measure of economic health. In GDP accounting, imports are subtracted so spending on foreign production is not mistakenly counted as U.S. production. That accounting subtraction does not mean every import is economically harmful. [BEA](https://www.bea.gov/index.php/news/blog/2025-06-03/bea-blog-expenditures-approach-measuring-gdp)

The first-half figures also are not price-adjusted. They cannot tell us whether the four technology rows grew because the United States bought more physical equipment, paid higher prices or experienced both. The FT-900 publishes real goods series, but not the same detailed first-half commodity decomposition used here. [FT-900](https://www.census.gov/foreign-trade/Press-Release/ft900/ft900_2606.pdf)

The defensible conclusion is narrower: exports produced the first-half improvement, an unusual first-quarter comparison amplified it, and four technology-hardware import lines continued to pull strongly in the opposite direction.