The Manufacturing Recovery Is Real. It Barely Weighs Anything.

Freight rates are up double digits. Manufacturing sentiment is expanding. Trucking stocks trade at record multiples on a declared cycle inflection.

Volumes are falling.

All of these are true at the same time, and the reconciliation tells you more about where the economy is actually growing than any single headline.

Start with the contradiction

The Cass Freight Index for June 2026 showed shipments down 4.1% year over year, the worst June reading since 2020. Expenditures over the same period rose 11.2%. (Source: Cass Information Systems, Cass Freight Index Report, June 2026.)

Read those two numbers together. Almost none of the increase in freight spending came from more goods moving. Nearly all of it came from what each shipment costs.

That is a supply story, not a demand story.

The drivers are well documented: English proficiency enforcement, the non domiciled CDL rule, thousands of driver training schools removed or put on notice. One large carrier estimates 250,000 to 400,000 drivers pushed out of the market, concentrated in one way spot freight. Capacity left. Rates rose. Nobody ordered more freight.

Why the manufacturing prints do not rescue it

The obvious objection: ISM manufacturing is expanding, with new orders in the mid 50s. Manufactured goods have to move. Where is the freight?

Six mechanisms explain the gap, and they compound.

ISM counts firms, not tons. It is a diffusion index. A hundred small firms improving slightly and three large firms cutting output sharply still prints above 50.

Manufacturing is roughly 10% of GDP. Truckload and LTL demand is dominated by retail replenishment, imported consumer goods, food, and construction materials. Manufacturing can expand at the margin while the larger block contracts.

Dollars are not tons. This is the structural one. Ten billion dollars of semiconductors fits in a handful of trailers. Ten billion dollars of furniture fills thousands.

Orders are not shipments. ISM new orders lead physical freight by three to six months, longer for capital equipment.

Construction freight does not repeat. The reshoring and data center buildout generates a wave of one time inbound loads. Steel, transformers, switchgear. Once the facility is built, a data center produces electrons, not truckloads.

Private fleets are invisible. Cass and ATA measure for hire freight. Shippers who bought their own trucks during the 2021 capacity crunch are still running them, and that freight no longer appears in the indices.

The concentration finding

Hard industrial production data settles it. Manufacturing output is growing, just not the way the sentiment surveys imply, and not evenly.

Source: Federal Reserve statistical release G.17, Industrial Production and Capacity Utilization, June 2026. High technology is defined by the Fed as semiconductors and related electronic components (NAICS 3344), computers (NAICS 3341), and communications equipment (NAICS 3342). The 11.1% and 15.5% growth rates as calculated by First Trust Advisors from the same release.

A ten to one gap between high tech and the total index. Growth is concentrated in the lightest, highest value goods in the economy, in three NAICS codes that together carry a low single digit share of the index. Everything else is flat, and capacity utilization sits well below its long run average, which is not what a manufacturing recovery looks like.

The manufacturing expansion is real. It is real in exactly the categories that do not move by truck.

Two companies that prove it

Stanley Black and Decker is a large fabricated metals manufacturer. Its Q2 2026 organic revenue grew 3% and it guided full year revenue to flat, with management explicitly framing the plan as capturing share in a flat end market environment. Margin expansion came from productivity, mix, and tariff refunds, not volume. (Source: SWK Q2 2026 earnings release, July 29, 2026.)

That is a manufacturer telling you its end markets are flat while the diffusion index prints in the mid 50s.

But look at the one part of SWK that is growing: Engineered Fastening industrial applications, up 7% organically on strength that management attributed partly to data centers and solar.

Eaton is the same signal at full volume. Q2 2026 sales of $8.5 billion, up 21% with 14% organic growth. Twelve month rolling orders up 41% in Electrical Americas. Data center organic revenue up 65% against an underlying market growing 23%. Full year organic guidance raised twice, from 8% to 12% at the midpoint. (Sources: ETN Q2 2026 earnings release, July 31, 2026, and Q2 2026 earnings call.)

Three companies in three unrelated industries. Same conclusion.

The AI buildout is real, large, and narrow

This is the thing worth internalizing.

Industry estimates put total data center related for hire trucking spend at roughly $7 billion, with something over half of that flatbed. Against a flatbed market on the order of $70 billion, that is a high single digit share, and flatbed itself is a fraction of total US truckload. [VERIFY BEFORE PUBLISHING: these market sizing figures are third party estimates and should be attributed to a named source or removed.]

Flatbed spot rates sit at records with tender rejection rates several times higher than dry van. The buildout is transforming that one corner of the market. It is immaterial to the whole.

Where it lands:

The table above is the author's classification, not sourced data. It reflects which modes and product categories carry data center construction inputs versus which do not.

A tailwind that lifts 6% of one freight mode cannot offset softness across the other 94%.

A note on measurement

One caveat that cuts against the bearish read. Cass has a disclosed bias. Its client base skews to consumer packaged goods, food, automotive, chemicals, and retail. Its truckload linehaul index is dry van only. A boom in flatbed and heavy haul construction freight would barely appear in it.

So Cass is a poor proxy for total US goods movement.

It remains a good proxy for the freight that dry van and LTL carriers actually haul, which is precisely the point. The blind spot in the index and the revenue exposure of the large public LTL carriers overlap almost perfectly.

What this means for the freight names

The 2026 setup across public trucking is a narrative and multiple problem more than an earnings problem.

Carriers spent the summer declaring a cycle inflection. Several delivered genuine beats. Several sold off anyway, which is the market's way of saying the good news was already in the price.

The unresolved question is what happens when the supply led rate spike meets returning capacity. Truck orders have rebounded sharply and net carrier authorities have turned positive again. A rate spike without demand confirmation has a limited shelf life.

Meanwhile the asset light brokers face the opposite problem. Rising spot rates reprice their cost base faster than their contract book, compressing gross margin per load. Rates falling hurts one group. Rates staying high hurts the other. The complex is not directionally uniform.

The test worth watching

If manufacturing strength were genuinely converting into freight demand, weight per shipment would rise across all LTL carriers at once. Instead it is diverging by carrier, which points to share shift rather than end market growth.

That single line item, published monthly, resolves the question faster than any earnings call.

Sources

• Federal Reserve Board, statistical release G.17, Industrial Production and Capacity Utilization, June 2026
• First Trust Advisors, economic commentary on the June 2026 industrial production release, July 17, 2026
• Cass Information Systems, Cass Freight Index Report, June 2026
• Stanley Black and Decker, Q2 2026 earnings release, July 29, 2026
• Eaton Corporation plc, Q2 2026 earnings release and earnings call, July 31, 2026

This article is analysis, not investment advice.