AIANALYSISPODCAST
Roughly $1.595 trillion in US factory commitments has not reversed 32 straight months of manufacturing payroll contraction, because industrial capital spending moves on a 5–9 year arc and a 1.
Tariff-driven factory investment is real. The factory jobs that were supposed to come with it are not. On the timeline the policy implies, they will not be for years.
In June 2026, the Bureau of Labor Statistics' preliminary manufacturing payrolls figure came in at 12,598,000 workers, essentially unchanged from a year earlier. It is the latest print in a 32-month run of contraction or stagnation, and it came in the same month that S&P Global's flash US Manufacturing PMI, a sentiment gauge where readings above 50 signal expansion, hit 55.7, a 49-month high. Output is expanding. Headcount is not. The contradiction is the point.
The dollar flow is the first half of the explanation. American Industrial Magazine's industry data hub, which compiles BLS, ISM, S&P Global, NAM, and Reshoring Initiative series, counts roughly $1.595 trillion in announced factory commitments in the post-tariff period. The same data set records 82,000 fewer manufacturing jobs than at the prior peak. Reshoring Initiative data show tariff citations by companies announcing new US capacity up 454% in the first quarter of 2025 versus the first quarter of 2024. Companies are citing tariffs as the reason to build. The hiring has not followed.
The mechanism is a clock, not a verdict. In a July 2026 episode of Decoder, Verge editor-in-chief Nilay Patel asked Evan Smith whether the tariffs had brought manufacturing jobs back. Smith said they have not. Altana is a trade-and-supply-chain intelligence firm whose customers include eight of the ten largest global logistics providers, plus government agencies and importers. Smith described the company as an "index bet on global fragmentation," built to track exactly this kind of dislocation. What Altana's customer data shows is the same thing: capital investment without the labor rebound.
The lag has two components. The first is industrial. A semiconductor fab announced today takes roughly nine years to translate into operating payroll; pharmaceutical validation runs five to seven years. The $55 billion in 2025 semiconductor pledges, in other words, is a 2030 payroll story, not a 2026 one. The second is mechanical. US labor productivity in manufacturing rose 1.9% in 2025, the largest annual gain since 2010, which means each new dollar of output is being produced with fewer workers than the prior cycle would have required. The new factories will hire. They will hire fewer than the old ones did.
The downstream signal is already in the data. Manufacturing construction spending is down overall, with electronics and semiconductor fab spend off 44% from its mid-2024 peak. NAM's Q1 2026 outlook reads 75.3% positive on manufacturer sentiment. The 75,000 manufacturing jobs lost between April 2025 and March 2026 is a smaller toll than the 170,000 in the prior twelve-month window, but the trend is still negative. June 2026's factory workforce contraction was the fastest since 2009 outside the pandemic.
What changes the picture is the next capex print and the next two payroll releases. If the May and June 2026 readings break the 32-month contraction, the lag-then-automation thesis loses its central evidence. The trade renegotiation calendar matters too: Smith said USMCA talks were expected to stretch into 2027, and a fresh 50% tariff on Canada has drawn partner pushback. The policy lever is still being turned.
The constructive question is which sectors are still expanding payroll within the flat aggregate, and where the capital investment is actually landing. Smith said Altana recently acquired Cervo AI, a customs-document automation firm, which points to one answer: the labor the tariffs have already created is in trade compliance and supply-chain software, not on the factory floor. The factory is being built. The workforce that runs it is still being designed.