Industrial Production July 2026: Resilient, Steady, Hopeful
Industrial production shows the benefits of the AI boom and related multiplier effects, but consumer goods soften.
Industrial production grew again in July with a modest +0.2% MoM and +1.1% YoY with business equipment up impressively YoY at +6.6%. The Materials market group was higher (+1.0%), but Consumer Goods was off by -1.8% as Autos and Appliances/Furniture both weakened. The non-consumer categories carried the day on the top line.
Manufacturing capacity utilization ticked very slightly higher MoM to 76.0%, which is well below the 1972-2025 average of 78.1%. Durables ticked higher on the back of Metals (primary and fabricated), Machinery, and Aerospace/Transportation but with Motor Vehicles was lower. Nondurables dipped MoM.
Basic multiplier effects come into play for industrial production with the “AI World” helping to solidify a range of industrial metrics during July even as we see other major economic drivers such as housing starts showing very negative trends (released on same day).
The materials, components, and finished goods demand that tie into the AI ecosystem drive a lot of economic activity independent of the adverse interest rate moves, but the consumer sector will still need to be the main driver of GDP growth into late 2026 and 2027.
The above chart and text that follows is somewhat of a replay of the history of the industrial production cycles across time from 1967 through today. We have covered much of this in the past, but we roll the historical recap forward for convenience below. We get new readers and students all the time, so we use it as a mini history lesson. For those used to this history, I added some new angles in this version but otherwise you can skip to the July IP and Capacity Utilization charts further below in the commentary.
Here’s a topical summary of the history sections:
A look at the top-down utilization numbers…
A brief cyclical lookback on replay…
Bring on the tech boom and excess…
The credit crisis arrived in the summer of 2007 and peaked in late 2008…
COVID, ZIRP, and a license to print risk…
Time for the new age transition to AI plus some old school crises…
As an “old guy” who covered such industries as autos and steel credits back in the 1980s (and worked on a few audits of resources and energy companies), we have seen plenty of cyclical stress and secular shifts. I came of working age in the first stagflation wave and oil spikes of the 70s and early 1980s, and slices of the “old” are still evident these days. Cyclical change and secular trends always get mixed in, and the evolution, restructuring, or collapse of industries are not new.
The manufacturing sector is today wrestling with tariff waves in a world with global supplier chains. Unwinding supplier chains built over decades using economic reasoning (comparative advantage, serving multiple global end markets, labor efficiency etc.) cannot be unwound or erase. Adjustment challenges are exacerbated when the decisions are erratic (as today), reversed in part or whole on an ad hoc basis, and/or at times based on rationales that often lack facts (e.g., who write the check for tariffs) or a conceptual foundation (is a trade deficit a subsidy or a de facto check written to another country?).
Oil spikes are not new. The US was slammed by them in the 4Q73 to 1Q25 recession and again in 1979 on the way into the 1980-1982 double dip that saw Volcker’s “mother of all tightening cycles.” Rapid increases in oil prices usually translate into yield curve punishment on the long end and occasionally recessions as seen in 4Q73-1Q25 recession, again in the Volcker double dip, and in a brief downturn starting in the summer of 1990 on the way into that Middle East oil adventure after the Aug 1990 invasion of Kuwait by Iraq (see Business Cycles: The Recession Dating Game 10-10-22).
A look at the top-down production and utilization numbers…
For July 2026, the “business equipment” line of industrial production is doing quite well at +6.6% YoY with “information processing” up by +8.9%. Consumer durables are down -1.6% YoY with the exception of the “home electronics” line at+9.3% (Table 1 of the release). Defense and space equipment is up by +6.8%, but construction supplies only +2.4%. Within Industry groups, “computer and electronics” was +9.9% YoY. Overall, durables manufacturing in industrial production was +3.9% and nondurables posted -1.6%.
The capacity utilization trends are covered in the tables further below, but it is always good to keep in mind that there are two moving parts: capacity growth and the rate of utilization. Over/under expansion relative to demand can shape pricing power (e.g. semiconductors on a global scale).
The current 76.0% for total manufacturing utilization is below the long-term median of 78.2% and slightly below the post-COVID median of 76.2%. The industrial production and capacity utilization numbers are among many metrics that take the label of “greatest economy in history” off the table. Whether GDP growth or payroll trends, the current economy is a bottom quartile economy when the growth of the 1980s and 1990s are factored in (see Presidential GDP Dance Off: Clinton vs. Trump 7-27-24, Presidential GDP Dance Off: Reagan vs. Trump 7-27-24).
The wildcard for this current cycle is the scale of the fixed investment that is now ramping up around the AI ecosystem. The multiplier effects cut across the full array of products and services and extend across many major trade partners with Taiwan at the top of the list (see Taiwan: Stakes are High, US Awareness is Low 5-17-26, US-Taiwan Trade: Risks Behind the Curtain 2-1-26). We would hope that someone in the White House would realize that the trade deficit with Taiwan is proof that such metrics can reflect strength (just look at the stock market and AI ecosystem valuations). Everyone is afraid to tell Trump. He might learn something.
The consumer sector and PCE growth have been unimpressive in 2026 with weakness in 1Q26 and a recovery in 2Q26, but fixed investment is now soaring (see 2Q26 GDP: Good Underlying Numbers Despite Headline 1.5% 7-30-26, GDP 1Q26 Final: PCE Growth Plunge 6-25-26). If you like comparing numbers (not a Washington thing) as opposed to spewing partisan adjectives, Trump 1.0 did not even frame up all that well vs. Carter and Biden – let alone the 1980s and 1990s (Gut Checking Trump GDP Record 3-5-25, Trump’s “Greatest Economy in History”: Not Even Close 3-5-25).
Trump 2.0 is off to a slow start on GDP growth (annual GDP growth in 2025 was well below 2024) and payroll additions pale in comparison with the earlier glory years of the 80s under Reagan and the 90s under Clinton and even Biden. That clear conclusion requires more numbers and less adjectives than we hear today.
A brief cyclical lookback on replay…
The chart at the top shows the material swings across expansion booms and busts. That includes the aftermath of some highly damaging oil spikes that brought on cyclical weakness and later saw capacity utilization plunge. Gasoline and home heating bills can damage household cash flow subject to the consumer financial profile. There were a lot more structural changes underway in those cycles than just oil and gas price pressures.
The pace of industry restructuring was a major factor in the 1980-1982 double dip as well as during the recession of late 1973 to early 1975. The Arab Oil Embargo set off chain reactions that turned the post-1973 period into an ebb-and-flow of macro nightmares for inflation and the consumer. Currency trends and labor costs were going through major changes across many years into the 1980s.
The May 1975 low of 71.6% followed the first bout of severe stagflation that unfolded after the Arab Oil Embargo. The 1980-1982 double dip recession came after the 1979 inflation spike took off with the Iranian oil crisis. The themes from Middle East wars or OPEC clashes (Iran vs. Iraq in the 1980s, Iraq vs. Kuwait/US in 1990) are being revisited today. There is room for the conflict to get much worse with a wide range of potential geopolitical outcomes.
Capacity utilization saw a 68.7% low in Dec 1982 as the economy was just coming off the recession trough. The first Gulf War came after the August 1990 invasion of Kuwait with capacity utilization hitting a low in March 1991 at 77.2%, which is above where capacity utilization stands now in 2026. That takes some of the edge off the “greatest economy in history” hype.
Bring on the tech boom and excess…
The age of tech and credit excess in the late 1990s brought some new swoons with 71.8% in Nov 2001 after the TMT meltdown and HY default wave rolled in. The Utilities and Power sectors and Telecom capex all took a beating in that time frame.
We recently saw a market commentary comparing and contrasting the tech bubble of the 1999-2002 swing with today’s valuations in the AI ecosystem. That writer had referred to the tech bubble of 1999 as NOT a case of balance sheet pressures as his theory was simply about valuation metrics. We are not sure which tech bubble he was referring to during 1999, but he missed 2 of the 3 letters in “TMT”. He was using internet stocks as his focus. The internet stock bubble was very different from the debt-fueled telecom (T) and media (M) credit cycle and market implosion.
The telecom-related, debt-funded capex and balance sheet stress of 2000-2002 was ugly with many cash flow bleeders heading to Chapter 11 and distressed debt restructuring that also brought industry consolidation. The wave of fraud (Enron, WorldCom, and more) also hit confidence levels. The credit stress was much higher in 2001 and the counterparty credit quality was much worse in 2000-2002 than what we see today. Private credit gets tossed around but 2000-2002 was an underwriting quality tragedy. That sets a high bar for “worst practices.”
The current fear of revisiting credit stress ties into the worry that AI models (revenue, cash flow) will fail to be realized. That does not change the fact that 2026-2027 will remain highly dependent on raising much more debt to “finish the job” that is being built into revenue growth and earnings models. That is anything but assured. If the debt markets balk, the stock valuations will see a lot of pain.
The credit crisis arrived in the summer of 2007 and peaked in late 2008…
The next crisis in the manufacturing sector plotted in the above chart came with the post-credit crisis low of 63.4% at the recession trough in June 2009. Mortgages and derivative excess caused a very real systemic panic and deep risk aversion from the Lehman Sept 2008 period and well into 2009. Banks and brokers collapsed and the whole Ghostbuster scenario (“cats and dogs living together”) started to play out.
Fears of cascading counterparty collapses rose with the bank system crisis as did worries around mass credit contraction hitting a broad swath of businesses. That in turn led to bailouts by the UST and protracted action by the Fed (see Wild Transition Year: The Chaos of 2007 11-1-22).
That period in late 2008 and into 2009 was famous for the bankruptcy and bailout of GM and Chrysler and how that reverberated across supplier chains from materials to components. TARP and late 2008 had already hooked up the “bailout infusion.” The credit contraction threat was very real, and the recovery in payrolls was slow to unfold with the consumer hammered by housing sector pain (Autos: War Stories & Anecdotes 9-22-23).
COVID, ZIRP, and a license to print risk…
COVID edged out the June 2009 utilization bottom with a record low of 62.6% in April 2020 and a wild 2-month recession. That pandemic period saw ZIRP, supply-demand imbalances and later the Russian invasion of Ukraine and ensuing oil spike that set off inflation and a tightening cycle that still did not trigger a 2022 recession. The Fed should get some credit for navigating that timeline even if late to the party.
In the context of manufacturing, ZIRP brought a wave of refinancing and extension of liabilities with low cost of capital encouraging investment and pent-up demand that rewarded rebounding inventories and capex. Consumer financing was cheap, and captive finance companies and the finance sector were ready and willing to lend.
Time for the new age transition to AI plus some old school crises…
As we detail below in the recession histories, the idea of what is low or high for capacity utilization in the context of corporate sector profitability is not as direct as one might think. Breakeven volumes, cost structures and pricing power are key drivers of profits.
The tariff excess as a core Trump policy only makes that unit cost analysis more complicated as we head into a new round of energy-related cost pressures in late 2026 and 2027. Material costs (notably petrochemicals, base metals), operating costs for companies, and consumer discretionary household cash flow headwinds make for a tricky set of moving parts shaped by erratic geopolitical decisions.
The AI upside as well as the fallout and related fears is a new variable in the mix that cuts across both the services and goods sectors. What AI will mean for productivity improvement vs. payroll contraction will be an ongoing test of theories. Even if AI proves true to the valuation promises and brings a “revolution” to the economy, many will also pay the price.
Revolution is a term often associated with a body count (even if jobs and not KIA). We have been getting a taste in the recent SaaS services panic (the “SaaS-pocalypse”). Those SaaS services markets consume a lot of goods and have a lot of employees. Total paychecks drive PCE, which is 68% of GDP. Massive payroll declines in services a real worry. Massive payroll decline in services is a legitimate concern.
The above table breaks out the high-level groupings with total industry and total manufacturing posting a mix of positive and negative variances in July 2026. We include a “delta column” on the right.
Mining and Utilities moved slightly higher for the month, and the rational expectation is that utilities will see more capacity added and more demand. The data center and power supply issue is more pressing and getting more political by the day in an area where there has been working class bipartisan anger around the side effects of such projects on power costs and water threats among other issues. That controversy has spread to Canada where we are seeing more legislative and regulatory battles around the data center capacity plans.
It may be local on a center-by-center basis, but it will find its way into an assessment of data center adequacy at some point for growth, AI ecosystem valuations, revenue models, pricing power, and cost structures. Any sense of faltering optimism around volumes (e.g. international restrictions on AI), pricing power, open source competition, regulatory backlash, or political risks (elections have consequences, etc.) could roil valuations.
Durables vs. Nondurables
We see Durables and Nondurables as the main events to track, and durables firmed up this month, ticking higher as a group. The durables sector is especially important with all of the supplier-to-OEM multiplier effects of tariffs and oil spikes. Some industries will lose and some will win.
The IEEPA SCOTUS setbacks and recent attempts by Trump to counterattack by “creatively” using existing tariff legislation (e.g., Section 338 from the 1930 Smoot Hawley Tariff Act) will face challenges. Trump always goes to “Plan B” and “Plan C” in his attempts to cut Congress out of any tariff decisions.
Trump is hanging his hat on Section 301 and has also invoked “forced labor” rules as a loophole. He recently threatened France with triple digit wine tariffs even without IEEPA. He seems to forget the EU works as a bloc. He also threatened Canada over wildfire smoke tariffs even as the USMCA is up for review. We will get some answers this week (potentially tomorrow) on the Canada 50% threats. The use of a Smoot Hawley 1930 provision is a Hail Mary that so far is unfolding subject to legal challenges.
Looking back from 2025 into early 2026, we see Manufacturing in a relatively narrow range and the same for Durables. Tariffs have not moved the production needle much so far, but that does not tell the profitability story on the tariff cost impact. Most people (outside the White House) know that buyers/importers write the check for the tariff, so that implies a higher breakeven volume or some offsetting actions in other costs or pricing.
Those tariff impacts will only play out over time across working capital cycles. Pricing decisions are more likely to come at a lag given the White House pressures, so there is a lot more to play out in 2026 earnings seasons and in company color ahead of FY 2027 when midterm results will be in hand.
The inflation story has turned into a negative variable for the macro picture in 2026. The “good news, bad news” aspect of a resilient industrial sector is that higher input costs can lead to the exercise of pricing power that flows into inflation. That beats the stagflation alternative but will keep life challenging for the FOMC and “Team Warsh” (if he can assemble a supportive voting team).
The Iran effects eased up in June but are now heading back in a negative direction in July and August. “War is hell” but it is also very unpredictable in how bad the macro effects can be. At the very least it is very bad for budget deficits that infect the UST curve.
The above table breaks out the details for the top 5 Durables lines and the two largest Nondurables lines. We include the capacity utilization deltas in the column on the right. We see more positive moves across all the major durables industries with 4 of 5 higher. “Motor Vehicles and Parts” moved lower. The two major Nondurables lines were both lower with declines in Chemicals and also lower in Food, Beverage and Tobacco.
The above table updates the capacity utilization history for expansions and recessions. As we routinely discuss, the ability of companies to generate solid profits at lower capacity utilization in today’s markets relative to past cycles is tied to both automation and the evolution of low-cost global supplier chains.
Low-cost supplier chains are getting severely undermined by tariffs and are now facing supply chain headwinds from the Iran War energy effects and the related supply-demand imbalances.
The “AI story” and what it all means for productivity and headcount will be a big part of the risk analysis ahead. The natural outcome will be less jobs in a service intensive economy such as the US. The irony is that the most immediate impact on jobs has been in tech, but that payroll retrenchment can easily spread.
The above chart details economic expansion and recession averages and lines them up by height as a frame of reference. We see the current level on the left at 76.0%. The fact that the current capacity utilization is below numerous recession averages posted on the right side of the chart also scoffs at the recurring “hottest country” rhetoric.
Trump has dialed that back a bit of late as more objective metrics work against his claims. That is notably the case with GDP growth rates, monthly payroll additions, higher interest rates, sliding housing volumes, and negative real wage growth among others. Calling it a “Golden Age” starts to cross into the realm of absurd (See links at bottom). Record stock markets are a positive and a reasonable bragging right, but the only Presidents since Nixon to see a record high stock market in their term was… all of them.
In the context of the tariff priorities and reshoring ambitions, the slack capacity implied by the chart in historical context underscores that the analysis is a lot more complicated than just “build more plants” and “keep tariffs high” to force reshoring. That is a story told on an industry-by-industry basis. It gets into themes around pricing power, cost structures and sourcing risks (tariffs, etc.). These days the topical questions revolve around “AI winners and losers” in productivity and the relative ability to adjust and engage in tariff cost mitigation strategies.
The multiplier effects are important, but the relocation of supplier chains and OEM assembly relocation and retooling is by definition very slow to unfold given intrinsic lag times, tooling equipment, and contracts in place (including collective bargaining and unions). It is a slow and painful process battling against the reality of automation and downsizing in some industries.
There is also the fact that the cost of relocation has risen with all the tariff effects for materials, components, and qualified labor. Autos are the main relocation stories, but that is generally going to be tied to existing infrastructure and not a strategy of all-new greenfield plants.
Trump vs. Canada
The issues around the USMCA review and a tariff spike levied on Canada could come to a head as soon as this week. Most remember Trump’s threat of “economic coercion” to make Canada the 51st state. That was in Jan 2025. That fat orange genie cannot get put back in the bottle.
The erratic USMCA and Canada negotiations that could unfold effectively make planning much more difficult if not impossible to do with clear economic certainty on trade rules. Trump’s signaling has been that he may terminate the USMCA entirely. That threat was regardless of any legal interpretations on whether his sole-actor decision is even legal.
Along with his recent Section 338 tariffs (Smoot Hawley) and threats of 50% tariffs on Canada, the potential for Canada to fight back is much more realistic this time around. That adds more drama to a policymaking framework that does not lack controversy in the best of times.
Anyone who has studied the Depression era economic collapse would have been hard pressed to predict that Trump would turn to a 1930 Act that has been pilloried by economists and historians for many decades (almost a century). This is now Trump’s “go to” legislation. His Congressional sycophants need to send out a search party for some body parts (brains, spines, and a few other missing anatomical needs)
See also:
Market Commentary: The Curve 8-17-26
Market Commentary: Asset Returns 8-16-26
PPI July: When Lower is Better but Still Bad 8-13-26
CPI July 2026: Slight Breeze, No Chill 8-12-26
The ChatGPT Living Wage Test 8-11-26
Existing Home Sales July 2026: Slow Row 8-11-26
Payroll Deltas July: Mixed Bag, Mediocre Numbers 8-9-26
The Curve: Steeper, Inflation Anxiety Remains 8-2-26
Employment Cost Index June 2026: Inflation > Wage Growth 7-31-26
June 2026 PCE: Inflation, Income, and Outlays 7-30-26
2Q26 GDP: Good Underlying Numbers Despite Headline 1.5% 7-30-26