just now
This is a quiet week by the standards of the last month. No CPI. No FOMC. No hyperscaler earnings. The headline catalysts are gone. What remains is something more useful: industrial production and capacity utilization data that will tell us whether the reindustrialization thesis is showing up in the numbers or just in the speeches.
We are more than eighteen months into the Trump administration's second term. Reshoring, onshoring, and industrial capacity expansion have been stated priorities since day one. The CHIPS Act incentives, the TCJA extension into the One Big Beautiful Bill, the tariff regime, and the defense procurement surge have all been oriented toward the same objective: rebuilding domestic manufacturing capacity that was offshored over the past three decades.
The question this week is whether the data confirms the thesis. If industrial production and capacity utilization are trending upward, the reindustrialization is real and investable. If they are flat or declining, the narrative is ahead of the fundamentals.
The Evidence So Far
The evidence from multiple data sources suggests the reindustrialization is real, though unevenly distributed.
Manufacturing output increased 4.6% in Q2 2026, with hours worked rising 2.6%, producing a productivity gain of approximately 1.9%. ISM Manufacturing PMI has been in expansionary territory for eight to nine consecutive quarters. Equipment investment in Q1 GDP grew at 17.2% annualized, the strongest print since the post-pandemic recovery. These are not survey-based sentiment indicators. They are hard output data showing that factories are producing more goods with more workers.
The domestic manufacturing footprint of key AI and defense companies reinforces the picture. Approximately 75% of Intel's foundry capacity is located in the United States. GE Aerospace's manufacturing base is anchored domestically. Raytheon Technologies (RTX) conducts an estimated 55-75% of its production in the US. Even Micron, where only about 2% of current production is domestic, is investing heavily in US-based fabs through CHIPS Act funding.
The GDP composition data tells the same story from the demand side. Data center construction, semiconductor equipment purchases, and defense procurement are all showing up as line items driving GDP growth. The investment economy is not an abstraction. It is physically being built: fabs in Arizona and Ohio, data centers in Virginia and Texas, packaging facilities in Oregon and New Mexico.
XLI holdings have reflected this reality. Earnings per share across major industrial names have grown meaningfully, and forward EPS estimates remain strong. The market is not pricing industrials on hope. It is pricing them on order books that are filling.
The Macro Tension
Industrial firms are facing a genuine dilemma that deserves honest examination.
Higher interest rates make capital investment more expensive. A factory that could have been financed at 3% in 2021 now costs 6-7% to finance. Every percentage point of additional borrowing cost reduces the internal rate of return on the project and extends the payback period. For capital-intensive manufacturing, where a single fab costs $15-20 billion and takes 3-4 years to reach full production, the financing cost is a material component of the total investment.
Simultaneously, the demand tailwind is so strong that waiting for lower rates would mean ceding market share to competitors who invest now. The AI infrastructure buildout, the defense modernization cycle, and the reshoring imperative are all creating demand for domestic manufacturing capacity that did not exist five years ago. A semiconductor company that delays building a US fab because financing is expensive risks losing CHIPS Act subsidies, falling behind on capacity commitments to hyperscaler customers, and watching a competitor capture the demand instead.
The result is that companies are investing despite the cost of capital, not because of it. This is unusual. In most cycles, high rates suppress capital investment. In this cycle, the demand signal is powerful enough to override the rate signal. The question is whether the economics work at current financing costs, and the answer depends on how long rates stay elevated.
The Missed Window and the Policy Response
The optimal time to launch a reindustrialization push was 2020-2021, when interest rates were at zero and the cost of financing new factories was negligible. That window was used instead for consumer stimulus, PPP loans, and enhanced unemployment benefits. The industrial policy came later, after rates had already risen.
This timing mismatch makes the extension of the Tax Cuts and Jobs Act into the One Big Beautiful Bill (OBBB) not just politically convenient but economically necessary. If the government wants companies to build factories at 6-7% financing costs, it needs to offset part of that cost through accelerated depreciation, investment tax credits, and direct subsidies. The TCJA provisions that allow full expensing of capital equipment are the mechanism that makes the math work for manufacturers who would otherwise wait for lower rates.
The CHIPS Act performs the same function specifically for semiconductor manufacturing: direct subsidies that reduce the effective cost of building domestic fabs to a level competitive with the subsidized construction costs available in Taiwan, South Korea, and Japan. Without these offsets, the reindustrialization thesis relies entirely on tariffs to make domestic production cost-competitive, which is a blunter and less reliable tool.
The TGA and the Coming Deployment
The Treasury General Account balance stands at $966.6 billion, approaching $1 trillion. This is the government's undeployed cash reserve, and its eventual deployment into the economy will have direct implications for the industrial buildout.
When the TGA is drawn down through government spending, the cash flows into the banking system as reserves, loosening financial conditions. If that spending is directed toward industrial and infrastructure priorities (defense contracts, CHIPS Act disbursements, infrastructure procurement, reshoring incentives), it flows directly to the companies building domestic manufacturing capacity. The effect is targeted fiscal stimulus that reaches the industrial economy without requiring the Fed to lower rates.
The political incentive to deploy this spending ahead of the midterm elections is obvious. The economic effect, if it flows through industrial channels rather than consumer transfer payments, would validate the reindustrialization thesis with actual capital rather than just policy intention.
The AI Cycle Is a Cold War Story
The AI infrastructure buildout is typically discussed as a stock market narrative. It is more accurately understood as an industrial policy and national security program.
The competition between the US and China for AI supremacy is being fought across every layer of the technology stack. Models were the first battleground, but models are now commoditizing: Chinese labs (Moonshot, DeepSeek) are producing competitive outputs, sometimes by distilling Western models. The battleground has shifted to the physical layer: chips, packaging, memory, networking equipment, and the energy to power it all.
The US response has been to weaponize the supply chain. Export controls on advanced lithography equipment prevent China from manufacturing leading-edge chips domestically. Restrictions on HBM exports limit Chinese access to the memory required for AI training. The recent moves to ban Chinese networking and optical components from US data centers extend the restrictions to the connectivity layer. The CXMT story (a Chinese memory manufacturer pricing above Samsung and going public at $480 billion) demonstrates that China is building capability despite the restrictions, but it is doing so at mature nodes and commodity products, not at the frontier.
The investment implication is that the AI buildout is not a discretionary spending decision by corporations seeking profit. It is a strategic imperative backed by government policy, subsidies, and export controls. The spending does not stop because of a recession or a rate hike. It continues because the alternative, ceding technological superiority to a geopolitical rival, is unacceptable to any administration regardless of party.
This is why the current bottleneck-driven pain (high GPU prices, constrained packaging capacity, elevated memory costs) should be understood as temporary. The policy apparatus is oriented toward creating a supply glut in every critical input: domestic chip fabrication, advanced packaging, memory manufacturing, and power generation. When that supply glut arrives, the deflationary effect on AI compute costs will be substantial. Input costs fall. Productivity gains accelerate. The inflationary pressures that currently dominate the macro picture reverse.
The Yield Curve in Context
The current yield curve is steep: the 30-year at 5.29% and the 1-year at 3.08% produce a spread of 1.289%. For comparison, in December 2020, the 30-year was at 1.8% and the 1-year was at zero, producing a spread of 1.8%.
The curve is steeper in absolute terms but narrower in spread terms than the ZIRP era. What matters for the reindustrialization thesis is the direction, not the level. If the supply glut in AI inputs materializes over the next 12-24 months, compute costs fall, productivity rises, and the inflationary impulse from the supply side diminishes. In that scenario, rates have nowhere to go but down, because the Fed's justification for elevated rates (persistent inflation) erodes as productivity-driven deflation takes hold.
The long end of the curve is currently pricing in sustained inflation and elevated government borrowing. If the reindustrialization succeeds in its stated objective of expanding domestic production capacity, the supply side of the economy improves structurally. More domestic supply means less import dependence, less vulnerability to supply chain disruption, and less cost-push inflation from logistics and tariffs. The yield curve, which currently reflects the cost of the transition, would eventually reflect the benefit of the transition: a more productive, more self-sufficient economy with lower structural inflation.
What to Watch This Week
Industrial production (Friday). The headline number tells you whether factory output is expanding. The manufacturing component specifically tells you whether the reshoring narrative is translating to physical production growth. Compare to the 4.6% manufacturing output increase in Q2.
Capacity utilization (Friday). This tells you how much of existing manufacturing capacity is being used. Rising utilization means factories are running fuller, which is a precursor to capital expenditure on new capacity. If utilization is above 80%, companies begin planning expansions because they are running near physical limits.
Housing starts and permits (Tuesday). Not directly related to the industrial thesis but relevant for the broader macro picture. Residential construction has been depressed by high mortgage rates. Any pickup signals that the economy is absorbing higher rates without collapsing.
The Apple-CXMT confrontation. Reports emerged this week that the US government has asked Apple not to purchase memory chips from Chinese manufacturers CXMT and YMTC. This is the supply chain cold war playing out in real time at the consumer electronics layer. Apple's reported interest in Chinese memory was driven by pricing and diversification. The government's intervention signals that national security considerations now override corporate procurement decisions, even for the world's most valuable company. For the reindustrialization thesis, this is a direct validation: the policy apparatus is actively redirecting supply chains toward allied and domestic sources, creating structural demand for non-Chinese memory manufacturing capacity. For Micron specifically, every chip Apple cannot buy from CXMT is a chip it must buy from Micron, SK Hynix, or Samsung.
The Bottom Line
The short-term bearish catalysts that have dominated the narrative (elevated rates, sticky inflation, geopolitical volatility) are real but cyclical. The structural shift underneath them (reindustrialization, AI-driven productivity, domestic capacity expansion) is the force that will define market returns over the next 3-5 years.
Industrial production and capacity utilization data this week will provide another data point on whether the structural shift is tracking. The evidence accumulated so far, from GDP composition to manufacturing output to PMI expansion to corporate capex commitments, suggests that it is. The reindustrialization is not a campaign slogan. It is showing up in factory output, equipment orders, and earnings growth across the industrial sector.
For investors, this week is an opportunity to step back from the headline noise and evaluate positions relative to the structural picture. The companies building domestic manufacturing capacity, supplying the equipment for that buildout, and providing the infrastructure for the AI supercycle are positioned on the right side of a multi-year trend. The near-term cost of that positioning (higher rates, compressed multiples, macro uncertainty) is the price of being early to a trade that the data is increasingly confirming.
This article is for informational purposes only and does not constitute investment advice. Past performance is not indicative of future results. Do your own research.
