BusinessJun 15, 2026
AI’s Infrastructure Race Has a Bottleneck Problem, and Industrial Stocks Are Cashing In
Rhea Lobo
The AI data center boom is writing checks that the construction industry can’t cash fast enough. Four major hyperscalers raised their combined AI capex to $750B for 2026, with spending on track to top $1T next year. As that gap widens, the companies supplying the equipment needed to build new facilities are emerging as some of the biggest beneficiaries.
Follow the power cables: Industrial suppliers are capturing an outsized share of the spending boom. Cooling systems, grid equipment, and power generation have become critical pieces of every new data center project. Modine Manufacturing grew AI data center sales 73% in fiscal 2026, while Quanta Services ended March with a record $48.5B backlog. Across the sector, expanding backlogs and higher sales point to sustained investment as hyperscalers continue adding capacity.
According to JPMorgan, more than 60% of the planned 2027 data center capacity isn’t yet under construction, while another 7% has already been delayed. Supply-chain bottlenecks, permitting hurdles, and power shortages continue to slow projects. Alphabet took a different route, acquiring wind and solar developer Intersect for $4.75B this year to secure power access, making it the only major tech company to own a power developer outright.
Policy vacuum: Construction delays aren’t the only issue. The Federal Data Center Enhancement Act is set to expire this September with no replacement in place. A General Services Administration employee told Wired, “Never in the history of data center policies has a policy expired without another one having been painstakingly worked on for three years behind the scenes.” Industrial suppliers remain well-positioned, but policy uncertainty now joins power shortages and supply-chain bottlenecks as another challenge for new data center projects.
