US markets

30 July 2026, 21:18 UTC Chief Investment Office GWM Investment Research

The investment implications of data center regulation US markets Authors: Kurt Reiman, Head of Fixed Income, Americas, UBS Financial Services Inc. (UBS FS); Jeannine Lennon, Municipal Strategist, CIO Americas, UBS Financial Services Inc. (UBS FS); James Dobson, CFA, CIO Equity Strategist, US Energy & Utilities, UBS Financial Services Inc. (UBS FS); Barry McAlinden, CFA, Fixed Income Strategist, CIO Americas, UBS Financial Services Inc. (UBS FS); Nathaniel Gabriel, CFA, CIO Equity Strategist, Materials & Industrial, UBS Financial Services Inc. (UBS FS)

• AI data-center development is unlikely to stop, but the rules of engagement are changing. Local opposition, ratepayer concerns, and permitting scrutiny are shifting the debate from whether projects should be built to how costs, benefits, and community impacts should be allocated.

• The policy response is likely to reshape the geography and economics of the buildout. Projects may increasingly favor jurisdictions with abundant power, clearer permitting rules, lower population density, and stronger frameworks for assigning infrastructure Source: Getty Images costs to developers rather than households.

• Data centers should be viewed as part of the broader AI industrial ecosystem. Restrictions that materially slow deployment could weaken demand for semiconductors, networking gear, electrical equipment, power infrastructure, and other inputs, not to mention US productivity.

Data center moratoriums The current backlash is not simply a story of local resistance to growth. AI data centers are another highly Despite the growing opposition, we expect the AI data- energy-intensive technology, similar in some respects to center buildout to continue in the US. More importantly, semiconductor fabrication (or fabs), steel, aluminum, the rapid buildout will likely force a faster evolution and other strategic industries. The difference is that AI in development terms: more deliberate siting, clearer is evolving unusually quickly and is widely perceived as cost allocation, fewer automatic incentives, and greater threatening, which may be amplifying opposition relative responsibility for hyperscalers and other AI data center to other industrial projects with similar resource demands. developers to make the benefits and trade-offs of artificial intelligence (AI) infrastructure more tangible to That distinction matters. Opposition that materially host communities. In this report, we use hyperscalers slows data-center deployment would not be limited to to refer primarily to the large cloud service providers the facilities themselves. It could also weaken demand developing the extensive computing infrastructure required for the products used to build data centers, including for AI, while recognizing that the broader data-center chips, routers, switches, transformers, cooling systems, infrastructure ecosystem includes other major technology and electrical equipment, reducing the industrial and companies. productivity benefits that would accrue from AI adoption.

This report has been prepared by UBS Financial Services Inc. (UBS FS). Please see important disclaimers and disclosures at the end of the document.

Strategic context: Why AI What counts as hyperscale? infrastructure matters A hyperscale data center is a large-scale, highly automated computing facility designed to deliver massive cloud, Investment in artificial intelligence computing power is AI, and data-processing capacity through a distributed, likely to reach 3% of US economic output in 2027, much software-defined architecture. Typically housing 5,000 of it financing the buildout of AI infrastructure. We assume or more servers, these facilities often span hundreds of that neither capital market funding channels nor end- thousands to several million square feet and consume tens user demand for artificial intelligence will constrain this to hundreds of megawatts of power, with some of the investment. newest hyperscale campuses designed for 1,000-plus MW of electrical capacity. To put that in context, 1,000 MW is Debt raises by the largest hyperscalers have attracted…