By Bryan J. Kaus
The bad economist sees only the direct consequences of a proposed course; the good economist looks also at the longer and indirect consequences.
Henry Hazlitt, Economics in One Lesson (1946)
Technology companies tried to buy all the available electrical equipment for their data center developments. Suppliers pushed back because they had commitments to other projects.
That account appears in the Chicago Federal Reserve district’s October Beige Book, relayed by an equipment distributor.
I keep coming back to the customers on the other side of those commitments.
They did not need to be competing with a technology company for revenue to find themselves competing with one for equipment. Their projects could have sound economics, willing customers and adequate financing, and still depend on whether someone else had already reserved what they needed.
We usually define competitors by what they sell. In a constrained market, it is also worth asking what they buy.
The latest manufacturing and inflation releases point to a broader industrial-investment question. When capital concentrates in projects that draw on shared equipment and skills, how does that change what other businesses can build, when they can build it, and what they must commit to earn a return? Data centers make that competition unusually visible today. The implications extend across the businesses that depend on those same inputs.
The expansion is real. Its breadth is another question.
A Kansas City Fed bulletin published October 13 puts numbers around the unevenness. Inflation-adjusted factory construction spending averaged about 70% more in 2022 through 2025 than in 2015 through 2019, although it has since fallen 37% from its November 2024 peak. Since December 2019, selected high-tech manufacturing capacity has risen 72%. Those industries represent roughly 3% of manufacturing output; total manufacturing capacity is up about 2%. Auto capacity rose 14%, while capacity outside autos and selected high tech fell about 1%.
These are manufacturing measures, not a complete accounting of computing services or the eventual productivity effects of the technology being built. High-tech capacity indexes also reflect improved capability, rather than simply counting additional factories. The evidence supports a narrower conclusion: a substantial investment cycle has not yet produced a broad expansion of manufacturing capacity.
That does not mean the rest of industry has stopped growing. The October 16 industrial production release showed manufacturing output rising 0.3% in September, with gains in machinery, primary metals, electrical equipment and chemicals. The Dallas district reported robust, broad-based manufacturing growth, well beyond semiconductors.
A monthly improvement in output and a longer-term shortfall in capacity can coexist. So can spare capacity across manufacturing as a whole and acute shortages in particular products. An idle production line is little help to a customer waiting for equipment it cannot make.
That is why I hesitate over both sweeping versions of the story: that American manufacturing has returned, or that nothing meaningful is happening. The useful work begins below the aggregate, where the assets, products, customers and constraints become visible.
Competition crosses industry lines
The Kansas City research identifies a particular pressure point. Utilization in electrical equipment, appliances and components reached 87%, its highest since 2008, while that broad category’s capacity remained about 9% below 2019. Transformer and switchgear prices have nearly doubled. Estimated import volumes are substantially higher, though the increase depends on the price adjustment used.
The Beige Book adds operating texture. Kansas City manufacturers described military and data center demand crowding out other production. A Mountain West contractor reported that competition from data center projects was raising pay for electricians and other skilled trades.
Those accounts suggest a competitive map that looks different from the one in an industry presentation. A manufacturer planning an expansion, a utility replacing equipment and a data center developer may serve entirely different markets. Yet each can depend on the same qualified suppliers, installation crews and delivery slots.
The transmission is practical. A contractor changes which work it bids. A supplier reserves its next production run. A customer accepts a longer delivery schedule or pays more to hold the original one. A project that cleared its investment hurdle six months ago may need to be underwritten again.
That creates opportunity as well as pressure. A supplier able to deliver reliably may become more valuable to customers. A service business that can extend equipment life, improve throughput or shorten commissioning may help relieve a constraint without waiting for an entirely new factory.
But the opportunity is specific. A busy order book does not tell us whether a business has dependable earnings, adequate cash or an advantage that will survive the next round of capacity additions.
A cost increase can become a capacity problem
September’s consumer price index rose 0.6% for the month and 3.7% over the year, with gasoline accounting for more than half the monthly increase. That energy shock has its own causes. It should not be folded into an explanation about data centers simply because both can raise a company’s costs.
What matters here is how separate pressures can compound. A business paying more to secure equipment or retain an installation crew may also face a larger freight bill, without receiving any additional orders of its own.
The Richmond Beige Book contains a consequence worth considering: trucking contacts described fuel costs squeezing maintenance spending and delaying fleet purchases. A cost increase can initially appear as lower margin, then become a deferred investment that leaves the business less efficient or reliable.
For a manufacturer, a similar choice might involve postponing a plant improvement to preserve cash for more expensive inputs. The company gets through the immediate pressure, but the improvement that could have reduced its next round of costs remains undone. That is one way a temporary squeeze can leave a longer operating disadvantage.
A backlog is a set of obligations
Consider a hypothetical equipment supplier offered a large order from a well-funded developer. Meeting it requires more inventory, additional people and an expansion of the plant. The revenue opportunity is attractive. Whether the commitment is attractive depends on details that rarely fit into a headline backlog number.
Does the customer provide deposits or progress payments? Can it defer delivery after the supplier has bought materials? Who carries changes in input costs? If the project loses a permit or grid connection, what happens to the order? Can the additional equipment and workforce serve another customer?
A supplier can be right about the market and still accept the wrong obligations. It can also use a strong market to establish relationships, develop skills and improve processes that leave it more capable when conditions change. The difference will often be decided while orders are plentiful, even if it becomes visible only later.
For a company facing higher costs without a comparable increase in demand, the priorities may look different. It needs to understand which inputs are genuinely scarce, where procurement commitments are justified and how quickly customer prices can adjust. Buying early can secure a production slot; buying indiscriminately can consume the liquidity needed to operate through a delay.
Neither situation lends itself to a universal instruction to expand, stockpile, hedge or wait. Good judgment requires matching the commitment to the exposure.
That is where an operator’s view adds something to the market narrative. Demand matters, but so do the obligations accepted to serve it, the cash required along the way and the alternatives available if the sequence changes.
The supply response has its own clock
There is already investment aimed at relieving these constraints. Eaton is converting a South Carolina plant into a $340 million transformer facility, with production expected in 2027. Hitachi Energy announced a $528 million Mississippi transformer facility in September, with production scheduled for 2029. These are company plans, not completed output, but they make clear that suppliers are responding.
For customers planning projects today, the relevant question is when that response becomes equipment they can actually receive, install and operate. A manufacturing expansion scheduled for 2029 cannot resolve a missing delivery slot in 2026. Imports may bridge part of the interval, provided the equipment meets the project’s requirements and can arrive when needed.
The customer’s project has a schedule too. Texas’s September 21 directive pausing TCEQ environmental permits for data center projects pending ERCOT’s grid review illustrates how an approval can change the sequence after commercial commitments have been made. Equipment may be ready before the site can accept it, or the site may be ready while equipment remains unavailable.
That is why I would reconcile supplier delivery dates, project approvals, installation capacity and payment obligations in the same operating review. Each may look manageable on its own. A mismatch between them can leave one party funding inventory while another waits to earn revenue.
The Point Taken
For the businesses on the other side of the Chicago distributor’s commitments, the issue is whether their own projects can proceed on the terms they expected.
That is the part of an investment boom I think deserves more attention. Growth changes the conditions around it before the eventual benefits are fully visible. Some businesses gain orders, others gain costs, and many experience both. What follows depends on the capacity they can access, the terms they accept and the improvements they make while the market gives them room.
A useful question for the next operating review might be: whose growth is changing the assumptions underneath our plan?
The answer may sit well outside the industry we thought we were competing in.
© 2026 23.5 Strategies; The Point Taken™