Barclays says annual AI infrastructure spending by Western hyperscalers and AI labs could exceed $1 trillion, more than $300 billion above current consensus, with upside from sovereign AI and China before peaking in 2028. The firm highlights EPC and infrastructure names such as Quanta, EMCOR, MasTec, AECOM, Sterling Infrastructure, Jacobs, and others as key beneficiaries of the AI data-center and power buildout. The note is constructive for infrastructure-linked stocks, though it is analyst commentary rather than company-specific news.
The key implication is that the AI buildout is no longer a chip-only trade; it is a bottleneck trade shifting downstream into power, interconnect, and field execution. In that regime, EPCs with exposure to electrical, mechanical, and utility scope should enjoy a multi-quarter backlog re-rating, but the best risk/reward likely sits with names that have the greatest mix of scarce labor, mission-critical complexity, and pricing power rather than simple general contractors.
Second-order effects matter more than the headline capex number. As hyperscaler spending accelerates, the real constraint becomes grid interconnection, transformer availability, switchgear lead times, and permitting cadence, which can extend project cycles and inflate change orders. That should favor PWR/EME/STRL-type execution specialists over more commoditized construction players, while also pulling forward demand for adjacent subcontractors and industrial supply chains that are not explicitly called out here.
The market may still be underestimating duration. Even if AI training intensity eventually moderates, the installed base of data centers requires power redundancy, cooling retrofits, and network densification, which supports a longer tail of non-discretionary spend. The contrarian risk is that consensus extrapolates a straight-line boom: if hyperscaler capex pauses for even two quarters, the most cyclical small/mid-cap EPC names could de-rate sharply because their multiples embed perfect execution and uninterrupted order growth.
From a timing perspective, this is a months-to-years setup rather than a day trade. Near term, the best catalyst is any evidence of backlog conversion, margin leverage, or upward guidance from contractors with direct data center exposure; the main reversal trigger would be capex discipline from the hyperscalers or delays caused by grid connection constraints becoming so severe that project starts slip instead of merely getting more expensive.
AI-powered research, real-time alerts, and portfolio analytics for institutional investors.
Request TrialOverall Sentiment
moderately positive
Sentiment Score
0.55
Ticker Sentiment
