The A.I. Boom Is Creating a New Market for the Physical Economy
Earlier this year, Microsoft president Brad Smith told a reporter that the single biggest obstacle to the company’s data center expansion wasn’t power, land or permits. It was the availability of electricians. Microsoft has reportedly resorted to flying in electrical crews from as far as 75 miles away and temporarily relocating workers to keep projects moving. Oracle, which is building data centers for OpenAI, was reported to have pushed some completion dates from 2027 to 2028, in part because of labor shortages, though the company disputed that characterization and said its projects remained on schedule. data center completion date by a year, citing labor shortages. An industry group representing electrical workers has called the shortage a “life or death” issue for the A.I. buildout.
The headlines around the A.I. boom tend to focus on chips, models and the companies racing to build them. Much less attention goes to the people actually building the buildings. That’s a mistake. This boom is landing hard and fast with trades businesses, and they’re the part of the American economy that fintech spent the last two decades largely ignoring.
The scale of the demand is difficult to overstate. Estimates from the Information Technology and Innovation Foundation estimates that the construction industry faced a shortage of roughly 439,000 workers as of late 2025, with more than 400 data centers under active development. Many of the shortages are concentrated in skilled trades such as electricians and pipe layers. In the Washington, D.C., Maryland and Virginia region, where data center construction is particularly concentrated, IBEW Local 26 has more than doubled its membership since 2018, to roughly 14,700 members. Electrician job growth is outpacing nearly every other construction trade the Bureau of Labor Statistics tracks. A regional electricians’ union in the Washington, D.C., Metropolitan area, one of the country’s densest data center corridors, reportedly doubled its membership just trying to keep pace with demand.
The latest jobs data shows just how much of the A.I. buildout is spilling into the broader physical economy. In September, nonresidential specialty trade contractors, including electricians and heating specialists, added roughly 12,000 jobs even as residential construction employment declined.
When the last company I co-founded, Bread, was sold in 2020, I could have built a consumer finance app, or another in the line of corporate cards for software companies. Instead, I went looking for the businesses that many fintech founders had written off: electricians, HVAC crews, plumbers, roofers, utilities and landscapers. These were people running real companies, with real payroll and real vehicles, who were still managing their books with paper receipts and basic credit cards because nobody had bothered to build them anything better.
For years, this has been a bet on a perhaps unglamorous but foundational part of the economy. In the last eighteen months, that’s changed. The A.I. industry’s physical infrastructure boom has turned the trades into one of the tightest labor and capital markets in the country.
That range is wider than it looks from the outside. The businesses riding this demand shift aren’t only owner-operators with a single truck. Increasingly, they’re operators running massive vehicle fleets. A meaningful piece of that growth is consolidation: private equity has spent the last several years rolling up plumbing, HVAC, electrical and landscaping businesses into multi-state platforms, betting that a famously fragmented industry can scale the way software did. Those roll-ups inherit the same back-office mess as the small operators they acquire, just at ten times the size. Scale compounds the financial infrastructure problem, rather than solving it.
Our team hears a version of this growth story from electrical and mechanical contractors in every market where data centers are being built. One services company won a single data center contract and quadrupled its monthly operating spend. This is A.I.’s impact on the physical economy: more jobs, more spending on fuel and materials and more trucks on the road.
There’s currently a gap not only in skilled workers, but also a shortage of financial tools built for how these businesses actually operate. A contractor landing a data center subcontract isn’t dealing with the cash flow of a corporate office. They’re financing fuel, equipment and payroll for a crew that might triple in size for a single job, against invoices paid on a schedule set by a hyperscaler, not by the contractor. A business can be profitable on paper and still need substantial working capital to get through the period between paying its workers and getting paid by its customer.
That distinction is significant as the customers themselves get bigger. Microsoft, Google, Amazon, Meta and OpenAI are committing enormous sums to the infrastructure needed to support A.I. The contractors building that infrastructure have to finance their side of the expansion long before the revenue from a project reaches their bank accounts.
Horizontal expense-management solutions—including some of the fastest-growing fintech companies—were largely designed for venture-backed software-margin businesses, professional services firms and larger corporations with relatively predictable spending. The physical economy doesn’t work that way. Its expenses are tied to projects, crews, vehicles, materials, fuel and the timing of customer payments.
This is the part of the A.I. story that gets skipped because it doesn’t fit neatly into a narrative about models and chips. But if you want to see where A.I. investment is actually landing in the physical world, look at what’s now required to win the work. An electrical contractor with 12 trucks who once competed for kitchen remodels may now be bidding on substation work. The general contractors awarding those jobs may require greater bonding capacity and working-capital reserves than a kitchen remodel ever demanded. The contractor is still the same business, but the financial requirements of the work have changed.
That creates a strange mismatch. The A.I. companies driving the buildout have some of the most sophisticated financial and technological infrastructure in the world. The businesses physically delivering that infrastructure often do not.
Fintech had two decades to build for these companies and mostly went elsewhere, toward consumer apps and enterprise software, where the markets seemed superficially larger and more obvious. The A.I. infrastructure boom is now doing what a decade of fintech pitch decks never did: making the case, loudly and expensively, that the physical economy isn’t an afterthought. It’s where a meaningful share of this decade’s capital is actually going to get spent.
Microsoft can develop models and build data centers, but it still needs electricians to wire them. Google and Meta can commit billions to infrastructure, but contractors still need workers, trucks, materials and working capital to turn those commitments into physical projects. And the same labor constraint that can slow a hyperscaler’s expansion can become an extraordinary growth opportunity for the businesses supplying the work.
The businesses fintech forgot are finally having their moment. It would help if the financial infrastructure serving them can catch up as quickly as their workload has.
