NextFin News - US manufacturing is expanding at its fastest pace in more than four years, with the ISM Manufacturing PMI climbing to 55.6 in July 2026 - the strongest reading since May 2022 - yet the revival is being powered by an AI-driven capital spending supercycle and a post-contraction rebound, not by the tariffs the White House credits for the turnaround. The data show a sector gaining momentum on output and orders while still losing ground on jobs, import dependence, and productive capacity. The boom is real. The explanation is not.
The Boom in the Data - and the Contradictions Beneath It
The headline numbers are genuinely strong. The Institute for Supply Management said its Manufacturing PMI rose to 55.6 percent in July, up 2.3 percentage points from June's 53.3 and the highest since May 2022, when the index hit 55.9. The sector has now expanded for seven consecutive months following a 10-month contraction. Production drove the gain: the Production Index jumped 6.3 points to 58.5, its highest level since November 2021. New orders grew for a seventh straight month at 56.7, backlogs rebuilt to 55.0, and customer inventories stayed thin at 40.7 - a combination that typically forces factories to keep running.
But the same report carries the counter-evidence. In July, 38 percent of purchasing managers' comments were positive and 62 percent negative, a 1-to-1.6 ratio of positive to negative sentiment. Pricing volatility was mentioned in 57 percent of negative comments, the Iran war in 43 percent, and tariffs in 18 percent. Input costs remain elevated: the Prices Index stood at 71.1, a level where more than half of respondents report paying more. A factory sector can print a strong PMI while its managers are, on balance, unhappy - and that is exactly what the July report shows.
The employment picture is the sharpest contradiction. Manufacturing payrolls stood at 12,611,000 in July 2026, according to Bureau of Labor Statistics data - only marginally above June and still below the roughly 12.7 million workers employed at the start of 2025. Revised figures show the sector shed about 103,000 jobs between January 2025 and January 2026. The July PMI employment sub-index did return to expansion at 52.8, the first time in 33 months, but survey sentiment and actual payrolls are describing two different recoveries: one in output, one in people.
Even the Federal Reserve's hard output data, which the ISM survey correlates with, shows the limits of the rebound. Industrial production rose just 0.2 percent in July after 0.3 percent in June, and manufacturing capacity utilization stood at 76.3 percent - still 3.1 percentage points below the 1972-2025 average. Factories are running harder than they were, but they are not running hot. There is slack left in the system, which is another way of saying the boom has room to run without adding a single unit of new capacity.
What Is Actually Driving the Rebound
Three forces explain the upturn, and none of them is the tariff wall.
First, the AI and data-center capital spending supercycle. S&P Global's Regulatory Research Associates forecasts roughly $1.3 trillion in aggregate US energy utility capital expenditure between 2026 and 2030, a record driven primarily by data-center demand. That spending flows into electrical equipment, machinery, and fabricated metals - the industries showing the most life in the ISM report. Second, inventory restocking: customer inventories at 40.7 signal firms ran stocks down and now must rebuild, which mechanically lifts production. Third, simple cyclical math - the sector is rebounding from a 10-month contraction, and the first readings after a trough are almost always the steepest.
History is the cleanest way to separate the tariff effect from the cycle. The last time tariffs were the central trade-policy story, in 2018-2019, the ISM Manufacturing PMI fell from 60.8 in September 2018 to 47.8 by August 2019 - a contraction directly tied to the trade war, not a boom. The last time the index rebounded from a contraction this fast, in 2020-2021, the driver was post-lockdown restocking and fiscal stimulus, with the Production Index hitting 60.5 in November 2021. And the most recent contraction, which ran 10 months into early 2026, was itself a textbook inventory correction after the 2021-2022 over-ordering cycle. Three cycles, three different drivers - and the tariff episode is the one that produced contraction, not expansion.
The tariff channel runs the other way. The Yale Budget Lab estimates the average effective tariff rate at 17.5 percent, the highest since 1932, with a median cost of about $1,400 per household per year. Since roughly half of US imports are inputs into domestic production, those duties land on factory cost lines before they ever reach consumers. The ISM Prices Index at 71.1 - with steel and aluminum tariffs cited alongside Middle East oil premiums - is the transmission mechanism in real time. Tariffs do not stimulate output; they raise the cost of producing it.
The White House frames the same period differently. An April 2026 release titled "Trump Effect: American Manufacturing Is Roaring Back as Factory Activity Hits Four-Year High" declared "the largest reshoring wave in American history as companies invest trillions to build and expand here at home." The Made in America Week proclamation in July repeated the claim, saying the administration is "ushering in the greatest reshoring wave in our history" and adding "over 900,000 American jobs." Those claims rest on announcement-level investment figures, not on the physical measures of capacity or employment.
"In July, U.S. manufacturing activity remained in expansion territory, growing at its fastest rate in more than four years," said Susan Spence, chair of the ISM Manufacturing Business Survey Committee, in the August 3 release. "In July, 38 percent of the comments were positive and 62 percent negative, with a 1-to-1.6 ratio of positive to negative sentiment. Pricing volatility was mentioned in 57 percent of negative comments, the Iran war 43 percent, increasing lead times 22 percent and tariffs 18 percent."
The official headline and the panel's mood point in opposite directions: the fastest expansion in four years, delivered by managers who are, on balance, complaining.
The Reshoring Gap: Investment Up, Imports Up More
Kearney's 2026 Reshoring Index, released in April, measures the year-over-year change in the US manufacturing import ratio - the share of domestic output represented by manufactured imports from 14 Asian low-cost countries. The index improved from -115 to -91 but remained firmly in negative territory, meaning America is still shifting toward greater import reliance, not less. US imports of manufactured goods rose 4.6 percent.
The China story is real but often misread. Direct imports from Mainland China fell by $135 billion, and China's share of total US manufacturing imports dropped below 10 percent, down from 20 percent four years earlier. That looks like reshoring until you follow the goods: the other 13 Asian low-cost countries picked up $193 billion in absolute dollar value - more than China lost. Computer and electronics imports rose 29 percent while domestic output in that category grew only 2.8 percent. For laptops and smartphones, even after tariffs, production costs at typical source countries stayed below what the same goods could be made for in the United States.
Capacity is the quietest rebuttal to the boom narrative. Kearney found that despite US manufacturing investment tripling over the past four years, physical capacity grew only 1.5 percent - uncertainty and policy churn causing delays and abandonments. Construction spending tells the same story from a different angle: manufacturing construction spending was $174.8 billion at a seasonally adjusted annual rate in May 2026, down 21.9 percent from roughly $224 billion a year earlier, the steepest decline among all nonresidential construction categories. Announced investment is not the same as built capacity.
There is a second-order consequence that the reshoring narrative misses entirely. When capacity fails to materialize while demand accelerates, the result is not more domestic production - it is more imports. That is precisely the pattern in the Kearney data: investment announcements tripled, capacity grew 1.5 percent, and imports rose 4.6 percent. The gap between the ribbon-cutting and the operating factory is being filled from overseas. Tariffs, in this telling, are not pulling jobs home; they are pushing supply chains sideways into countries that can still serve the American consumer at a lower landed cost.
The Market Has Already Priced the Boom
The equity market's verdict on the manufacturing story is worth reading alongside the data. The Industrial Select Sector SPDR ETF (XLI) gained roughly 16 percent year-to-date through June 2026, with inflows of about $3.6 billion, as investors priced in data-center capex, defense spending, and reshoring themes. The sector's forward price-to-earnings ratio has climbed above 26 times - near technology-like multiples - which leaves little room for disappointment. In other words, the market is not discovering the manufacturing recovery; it is underwriting it at a premium.
That positioning matters because it means the conventional wisdom - factories are coming back, buy the industrials - is already embedded in prices. The second-order question is what happens when the cycle turns. If the inventory restocking leg rolls over, or if tariff-driven input costs force the Federal Reserve to hold rates higher for longer, the same multiple that amplified the rally becomes the transmission channel for the drawdown. The industrial sector's rich valuation is not a reflection of tariffs working; it is a bet that AI capex can outlast a cyclical upturn, and that bet is priced for perfection.
There is also a cross-asset channel the equity rally obscures. A Prices Index at 71.1 with tariffs cited by nearly one in five negative comments keeps inflation expectations sticky, which caps the Federal Reserve's room to cut rates. Lower-for-longer rates, in turn, raise the discount rate on the very long-duration capex projects - data centers, semiconductors, batteries - that are carrying the manufacturing cycle. The policy that claims credit for the boom is therefore tightening the financial conditions the boom depends on. It is a self-limiting recovery, and the limit is visible in the bond market before it shows up in factory orders.
The Cyclical Call - and the Counter-Thesis
The honest reading is that this is a cyclical upturn layered on a slow structural shift that began before the current administration. The cyclical leg - inventory restocking, AI-linked capex, post-contraction rebound - is mean-reverting by nature and already fully priced into industrial stocks. The structural leg - supply-chain diversification away from single-source China exposure - is genuine but it is diversifying toward Mexico and Southeast Asia, not necessarily toward the American factory floor.
The strongest counter-thesis belongs to the administration and its allies: manufacturing added jobs for the first time in three years, the PMI is at a four-year high, and companies are announcing record domestic investment. On the employment claim, the counter has a point - the sector did stop bleeding jobs in 2026 - but it starts from a base that lost 103,000 positions in the prior twelve months, and July's 12.611 million payrolls remain below the 12.7 million mark from January 2025. A rebound from a self-inflicted trough is not the same as a policy-created boom. And on investment, the counter-thesis cannot answer the capacity gap: if the reshoring wave were real, imports would be falling, not rising 4.6 percent.
The specific signal that would falsify the view presented here is concrete: if the Kearney Reshoring Index turns positive for two consecutive annual readings while manufacturing capacity growth accelerates above 2 percent and payrolls decisively reclaim the 12.7 million level, then the tariff-and-reshoring narrative would be winning on the physical measures, not just the announcement headlines. Until that combination prints, the boom is real - but its engine is AI capex and the inventory cycle, and the tariffs are drag, not fuel.
What Comes Next
Short term, the momentum is self-reinforcing: thin customer inventories and rebuilding backlogs should keep production elevated through the rest of the year, and the July employment index crossing into expansion suggests hiring could finally follow output. Medium term, the constraint is capacity and cost - a Prices Index at 71.1 with tariffs cited by nearly one in five negative comments leaves little room for margin expansion if demand softens. Long term, the structural question is whether announced semiconductor and battery investments convert into built capacity once policy incentives sunset or shift.
Base case: manufacturing growth moderates from the July peak but stays in expansion, with output led by electrical equipment and machinery tied to data-center buildout. Upside case: a sustained drop in tariff uncertainty unlocks the roughly $1.3 trillion utility capex pipeline faster than expected, pulling capacity growth above trend. Downside case: input-cost pressure and a stronger dollar squeeze exporters just as the inventory cycle turns, pushing the PMI back toward 50.
US manufacturing is growing again - and the growth is welcome. But a boom financed by AI-driven capex and inventory restocking, while imports rise and capacity stalls, is not a boom that tariffs built. The policy that claims credit is the same policy raising the cost of every input that makes the recovery possible.
Explore more exclusive insights at nextfin.ai.
