U.S. reindustrialization and AI reinforce each other while competing for shared resources

Scarcity and structural demand are not the same thing. Scarcity emerges across physical, administrative, execution and financial constraints. Physical pressure is most visible in electricity supply, grid capacity, equipment, serviced land and, in some locations, water. Permitting, local approvals and interconnection processes can delay delivery, while shortages of labor, equipment and specialist construction capability add execution risk. Financing remains available for credible projects, but investors are placing greater weight on sponsor strength, counterparty quality, contracted revenues and the assumptions supporting development and exit. The asset management imperative is to convert structural demand into durable income by controlling or resolving the bottleneck.

These constraints rarely occur in isolation. Land may be available without power; a project may have prospective power but no firm connection timetable; or an approved scheme may stall on equipment, labor or financing. Reindustrialization depends on assembling power, water, specialized equipment, labor, transport links and supplier capacity around individual manufacturing locations. In AI infrastructure, large campuses require substantial and reliable power alongside transmission, cooling and connectivity, while more distributed facilities depend on metropolitan power and network capacity. In both cases, locations where the necessary inputs are already secured are harder to reproduce than nominal development capacity.

These overlapping demands are contributing to longer development timelines and, in some cases, data-center project cancellations. Wood Mackenzie estimates that U.S. power transformers are in a 30% supply deficit, with competition for available production materially increasing lead times and prices.^1,2 Utilities are reportedly ordering some transformers three to four years in advance. A skilled-labor shortage is also testing U.S. data center construction, compounded by an aging construction workforce and too few new entrants to the trades.³ Limited domestic supply has increased reliance on foreign-made electrical components, creating greater exposure to tariffs and other trade restrictions.^4,5

Comparable demands on physical capacity can receive markedly different policy treatment. In January 2026, New York celebrated the groundbreaking of Micron's planned $100 billion semiconductor complex in Clay, expected to support nearly 50,000 jobs across the state.¹⁰ Six months later, the State paused environmental permits for new hyperscale data centers for up to one year while developing protections for ratepayers, the grid and local communities.^7,9 Both require substantial power and enabling infrastructure, and the Micron facility also supplies a critical input to the AI economy. While the fab is associated with employment, supply-chain resilience and community investment, the hyperscale data centers are being assessed against their implications for power systems, consumers and localities. The contrast shows that policy can allocate scarce capacity according to perceived public benefit as well as market demand.

Taken together, these dynamics show how scarcity filters broad demand into deliverable capacity. Projects with credible access to power, equipment, labor and approvals are better placed to advance; those still dependent on future grid investment, supply-chain relief or local consent retain more development risk. Value is more likely to accrue to existing or deliverable capacity, and to enabling infrastructure capable of resolving a binding constraint, than to projects that remain exposed to it.

Reindustrialization creates demand for advanced-manufacturing plants, semiconductor and defense facilities, together with the transport and logistics serving them. The AI buildout creates demand for data centers, fiber, towers, powered shells and serviced land. Semiconductor facilities span both buildouts, while both cycles depend on power systems, water, electrical equipment, skilled labor and long-duration capital. The strongest opportunities lie where assets control or resolve constraints shared by several sources of demand.

Power, utilities and the energy transition

Power is foundational: data centers, advanced manufacturing and wider electrification can expand only as quickly as generation, transmission, distribution and storage can support them. Within this system, contracted renewables offer a different return profile from merchant generation. Battery storage can make intermittent supply more usable and is increasingly attracting institutional capital as revenue structures and operating experience mature. Natural-gas generation can fill gaps when other supply is unavailable, though new construction faces equipment constraints, long lead times and local delivery bottlenecks. Domestic natural gas abundance offers some insulation from imported-fuel-price shocks, but does not resolve these constraints. Utilities and electricity networks offer investors exposure to the substantial capital investment required in grids and distribution to meet rising demand. Returns nonetheless depend on regulatory cost recovery and remain exposed to affordability concerns.

Digital infrastructure

Demand visibility is comparatively strong, and preleasing can reduce occupancy risk, but it does not resolve power availability, construction execution, elevated entry valuations or reliance on capital growth and exit assumptions.

Land awaiting an interconnection agreement carries a different risk profile from an operating facility with secured power, installed equipment and contracted revenues. This distinction echoes and extends the Powered Land investment thesis: as grid constraints intensify, value increasingly rests not in the building alone but in secured power, interconnection rights and permits. Caution toward crowded hyperscale exposure is warranted because of the dispersion in risk and valuation, rather than weaker conviction in underlying data demand. Rising affordability concerns and local community backlash – including project cancellations and proposed moratoriums in some U.S. states – add a further layer of deliverability risk beyond the balance sheet.

Transport and supply-chain infrastructure

Manufacturing must move physical inputs and finished goods, favoring proximity to ports, freight rail, highways and intermodal nodes. Once operational, data centers move information rather than goods, making access to power and fiber more important than freight connectivity. The direct transport opportunity is therefore more closely tied to reindustrialization than to the AI buildout, particularly through irreplaceable nodes serving complex or mission-critical flows. Ports and freight assets nevertheless remain exposed to trade disruption and changing supply chains.

Advanced manufacturing and enabling real estate

Semiconductor plants, defense facilities and other specialized industrial assets depend on power, water, skilled labor, transport links and supplier networks. This does not create a blanket industrial-property opportunity. Value is more likely to accrue to facilities and sites where this infrastructure is already assembled than to generic warehouses or land still dependent on future utilities and approvals. Water, cooling, treatment and reuse systems may also become investable enabling assets where they unlock digital or industrial capacity.

The two buildouts can also succeed one another on the same development platform. Microsoft's Mount Pleasant data-center campus is being developed within the Wisconsin district created around Foxconn's manufacturing project, where land and public infrastructure had been assembled for industrial development.^12,13,14,15 Microsoft has indicated that maximum electricity demand could rise from 400 MW for Phase 1 to around 900 MW as the campus expands. This illustrates how capacity assembled for one buildout can be redeployed by the other. Rising demand at Mount Pleasant also shows how quickly that redeployment can place renewed pressure on the same local power system.

Identifying scarce, well-positioned assets is only the first step. The harder task is execution – securing inputs, coordinating utilities and public authorities, managing permitting and construction interfaces, and aligning capacity with credible long-term demand. An announced pipeline is not the same as contracted, financed and deliverable capacity, and this gap is where much of the return dispersion within a theme originates. In August, Texas paused new grid-interconnection approvals for data centers pending audits of power and water requirements, ownership, incentives and community impacts.¹⁸ For grid-dependent projects, the directive makes interconnection a material underwriting constraint, as financing depends on credible power access and a firm connection timetable.

Operational performance becomes the center of gravity once assets are built. Translating structural demand into resilient income then depends on locking in contracted, inflation-linked revenues, passing through costs effectively, and managing counterparties well. Community and political stakeholder management is also important, particularly where projects heavily rely on local power, water, housing or public infrastructure. Where infrastructure expansion raises electricity costs for households and local businesses, this can trigger community and regulatory pushback that limits an owner's ability to capture the full value of the scheme. Meta's Richland Parish development provides a possible solution to address this friction. Under an agreement with Entergy Louisiana, Meta will fund new generation and pay the infrastructure costs required to serve the facility over a 20-year term, while contributing $215 million to bill-assistance and energy-efficiency programs.¹⁶ Both companies project savings for customers, although those benefits depend on implementation.

Long-duration buildouts pass through materially different risk phases. Development capital absorbs political, permitting and project-formation risk. Value-add capital assumes construction, commissioning, leasing and operational ramp-up risk. Core capital is better suited once contracted revenues, operating performance and durable demand are sufficiently visible.

Battery storage illustrates this progression. Developer capital initially absorbed construction risk, uncertain revenue stacking and immature market structures. Institutional participation has broadened as operating experience, project economics and contractual visibility have improved. These transitions do not always occur on schedule. Permitting delays, equipment shortages, tenant changes or financing conditions can leave capital exposed to risks it was not intended to bear. Abundant thematic capital has not removed the need for discipline on entry valuation, leverage, contractual protection and exit assumptions.

AI demand can remain strong while investors become more discriminating about the cost and structure of delivering the required capacity. This places greater weight on sponsor strength, construction and completion risk, leverage, the duration and enforceability of contracted revenues, and refinancing and residual-value assumptions. Sponsor differentiation matters as much as revenue-counterparty quality: the largest hyperscalers have materially different balance-sheet resources from leveraged cloud providers, AI laboratories or speculative operators dependent on repeated external funding. Capital duration must therefore align not only with an asset's expected holding period, but with the time required to navigate its highest-risk phases.

As development risk falls and operating visibility improves, ownership may migrate between investors with different return requirements. Recapitalizations, continuation vehicles, transfers between strategies and secondary transactions can facilitate that transition. Liquidity should be considered at underwriting. Transaction size can influence exit routes, with mid-market assets typically providing greater liquidity than large platforms dependent on an IPO or a single sale process. These structures create value only where they preserve the operating plan and governance, and do not transfer risk at a valuation that compromises future returns.

Reindustrialization is intended partly to reduce reliance on foreign supply chains for resources and inputs linked to economic and national-security priorities. However, U.S. domestic capacity remains incomplete across important parts of the value chains supporting both buildouts. Semiconductors are the clearest example. Leading-edge production depends on U.S. design, Taiwanese fabrication, Dutch lithography equipment, Japanese materials, South Korean memory and a wider network of specialized suppliers.¹¹ U.S. fabrication plants can reduce geographic concentration, but they do not immediately recreate the process expertise, engineering talent and supplier density required at the technological frontier.

Diversifying the semiconductor supply base does not remove the risk at the infrastructure level. AI-related facilities remain vulnerable to upstream disruption regardless of where they are located: data centers may possess land and power, but still depend on timely deliveries of advanced computing equipment sourced from the same concentrated global supply base.

U.S. reindustrialization and the AI buildout will sustain demand across power, digital, industrial and enabling infrastructure for years to come. Converting that demand into durable income requires asset-management expertise capable of controlling or resolving binding constraints; otherwise, investors remain exposed to the delays, cost inflation and execution risk that make thematic exposure unreliable on its own.

Investors should favor secured inputs and enforceable timetables over announced pipelines, execution capability over passive thematic exposure, and capital structures that can evolve as risk falls rather than remain locked into a single holding period. The objective is not to avoid risk, but to be adequately compensated for the development, financing, counterparty, regulatory and geopolitical risks actually assumed – including the upstream dependencies that persist even where domestic capacity expands.

References

Important Information