AI's Real Estate Footprint: A Structural Shift for Data Centers and Offices

The catalyst for this shift is clear: a global AI arms race. This competition is not just about algorithms; it is a physical and financial sprint to build the underlying infrastructure. The scale of the investment is staggering. Preliminary data shows that announced foreign direct investment in data centres exceeded an estimated $270 billion in 2025, making the sector a top recipient of new, from-scratch foreign investment worldwide. This surge helped push global FDI up 14% for the year, though the headline figure is somewhat inflated by flows through major financial centres.

The core of this investment is power. Goldman Sachs Research forecasts that global power demand from data centers will increase 50% by 2027 and by as much as 165% by the end of the decade compared to 2023. This isn't a distant projection. The current power usage of the global data center market is around 55 gigawatts, with AI workloads already accounting for 14%. By 2027, that demand is projected to reach 84 gigawatts, with AI's share nearly tripling to 27%. This creates a fundamental constraint: the ability to secure reliable, high-capacity power is becoming a primary factor in where data centres-and the companies that need them-can be built.

This physical build-out is mirrored in the commercial real estate market. The demand for office space from AI firms is explosive. In Manhattan alone, AI firms leased 486,000 sq ft of office space in the first nine months of 2025, a figure that already eclipses the 414,000 sq ft leased for the entire previous year. This rapid scaling is a direct result of the capital being deployed into the sector, as companies race to hire and establish physical presences to support their AI operations.

The bottom line is a dual demand creating a structural catalyst. On one side, there is a massive, capital-intensive build-out of data centres to house the AI compute. On the other, there is a surge in demand for office space to house the people building and managing it. This is not a cyclical trend but a fundamental reconfiguration of investment flows and real estate needs, driven by the relentless pursuit of technological advantage.

The Data Center Revolution: Power as the New Metric

The economics of data centers are undergoing a radical transformation. The asset class is no longer being valued on traditional real estate metrics like rent per square foot. Instead, it is being underwritten by a new, fundamental unit: megawatts. This shift from "sticks and bricks" to power-centric infrastructure is the defining characteristic of the AI era.

The scale of institutional capital flowing into this new paradigm is staggering. Between January and August of 2025, there were 42 data center transactions with an aggregate value of nearly $13 billion. That pace, on top of a $52 billion market in 2024, signals a swift institutional embrace of digital infrastructure. The underwriting logic has changed accordingly. Investors now prioritize secured power capacity via power purchase agreements (PPAs), guaranteeing payment regardless of a tenant's usage. This focus has moved the center of gravity in valuation from tenant diversification to power assurance.

The physical footprint of this new asset class is unprecedented. While the largest existing U.S. data centers draw less than 500 megawatts, the next generation is planned at a scale that dwarfs traditional power plants. The largest projects aim for capacities of up to 2,000 MW, or 2 gigawatts. Even more ambitious are early-stage, 50,000-acre campuses that could consume 5 gigawatts-more than the capacity of the largest U.S. nuclear or gas plants. This creates a fundamental tension: the AI infrastructure build-out is racing ahead, but the physical grid is not.

This power-centric model faces severe constraints. The nation's electrical grid, largely built in the 1960s and 1970s, is near capacity and aging. As a result, there is currently a seven-year wait on some requests for connection to the grid. This grid stress is the leading challenge for development, cited as very or extremely challenging by 72% of surveyed executives. Compounding the problem, supply chain issues are constraining energy companies and hyperscalers, with costs for key components having jumped 40% over five years. The result is a high-stakes race between AI ambition and physical infrastructure, where the ability to secure power is the ultimate gatekeeper.

Industrial/Logistics Synergy: The Prologis Play

The most compelling structural opportunity lies at the intersection of two powerful trends: the industrial real estate sector's deep expertise in logistics and power management, and the explosive demand for AI infrastructure. This convergence is exemplified by Prologis, the world's largest industrial landlord, which has made a decisive strategic pivot. The company plans to spend $8bn on data‑centre projects to meet AI's computing needs, aiming to develop up to 100 new facilities by 2029. This move is not a sideline bet; it is a direct leveraging of its core industrial platform to capture the AI boom.

The synergy is logical and powerful. Industrial real estate has long been about securing land, managing complex utility connections, and building large-scale, power-intensive facilities-precisely the skills required for next-generation data centers. Prologis brings a proven track record in sourcing and developing large parcels of land, a critical asset for campuses that can consume gigawatts. Its existing relationships with energy providers and its operational experience in managing high-utility loads provide a natural advantage in navigating the power-centric underwriting model that now dominates the sector. As CEO Hamid Moghadam notes, meeting these needs will require a diverse mix of energy sources, a challenge where industrial developers' logistical expertise is invaluable.

This strategy signals a broader trend. Industrial REITs, with their established infrastructure and land portfolios, are uniquely positioned to become key players in the digital infrastructure build-out. They are not entering a new market from scratch; they are applying their existing capabilities to a new, high-growth asset class. This creates a virtuous cycle: the capital and operational know-how from industrial development can accelerate the construction of the power-hungry data centers that AI depends on. In turn, the success of these projects reinforces the value of the industrial real estate model, which is now inextricably linked to the AI supply chain.

The bottom line is a structural re-rating of industrial assets. For Prologis and its peers, the $8 billion commitment is a bet that their industrial expertise is the optimal platform for the AI infrastructure race. It transforms their portfolio from a collection of warehouses into a foundational layer for the digital economy, directly linking their growth to the relentless expansion of AI compute.

The Office Market Reinvigoration: A Selective Recovery

The narrative of an AI-driven office market recovery is gaining traction, but it is a story of selective strength against a backdrop of deep-seated weakness. While the sector's overall vacancy rate remains high at around 35%, a distinct leasing flurry is underway, driven by the capital and hiring needs of AI firms. This demand is not a broad-based resurgence but a concentrated catalyst that may offer a lifeline to specific portfolios.

The evidence points to a surge in activity, particularly in tech hubs. In San Francisco, for instance, AI companies have leased around 500,000 square feet this year, following a similar pace last year. This has spurred optimism among local landlords, with leasing growth coming "from AI companies," according to one analyst. The trend is mirrored in Manhattan, where AI firms leased 486,000 sq ft in the first nine months of 2025. This pattern echoes past tech cycles, with observers drawing a parallel to the mobile app boom of the 2010s, which fueled massive office leasing outside major hubs.

Yet the sustainability of this recovery is questionable. The demand is bifurcated and often tied to short-term needs. Larger, established AI firms seek premium class A space, while startups prioritize flexibility and immediate occupancy in class B buildings. This creates a dynamic where tours in some portfolios have increased 60% year-over-year, but it does not necessarily signal a shift in the broader market's risk profile. The fundamental headwinds-high vacancies, ongoing tech layoffs, and entrenched remote work patterns-remain largely unaddressed by this niche demand.

The bottom line is that AI is a selective catalyst, not a cure-all. It is providing a powerful, near-term boost to certain office markets and landlords, validating the thesis that AI companies are "in the office every day of the week." However, this demand is concentrated among high-growth firms and does not yet reflect a broad reversion to pre-pandemic occupancy levels. For the office sector to see a true, durable recovery, this AI-driven leasing must translate into longer-term commitments and broader economic confidence. Until then, the office market's revival will remain a story of pockets of strength within a larger, still-challenged landscape.

The Investment Play: Synthesizing the Opportunities

The structural shifts we've outlined create a clear, multi-tiered investment framework. The primary opportunity is in data centers, where the old rules have been rewritten. Valuation is no longer about rent per square foot but about secured megawatts. The underwriting logic has shifted to power-centric models, with investors normalizing income around dollars per kilowatt per month. This creates a new set of winners and losers. The most compelling plays are in the power and grid partnerships that enable development. The leading challenge for this build-out is grid stress, with a seven-year wait on some connection requests. This constraint elevates the value of any asset or partnership that can navigate this bottleneck, whether it's a developer with direct utility relationships or a REIT that can secure long-term power purchase agreements.

A secondary, more selective opportunity exists in office space. The AI leasing surge is real, but it is concentrated and often tied to short-term needs. The recovery story is not a broad-based cure but a catalyst for specific properties. Landlords with flexible layouts and sustainability credentials in established tech hubs are best positioned to attract these high-growth tenants. The demand is for space that can scale quickly, often with two- to three-year leases. This creates a bifurcated market: a strong niche for agile, well-located assets versus the broader weakness of high-vacancy, inflexible portfolios. The AI-driven office demand may offer a lifeline, but it does not yet signal a durable reversion to pre-pandemic norms.

The most unique hybrid play lies within the industrial sector. Here, the convergence of expertise and capital creates a powerful vertical integration story. Industrial REITs like Prologis are not just building warehouses; they are applying their core strengths in land acquisition, utility management, and large-scale development to the next-generation data center. This is a structural re-rating of industrial assets, transforming them into foundational layers for the digital economy. The $8 billion commitment from Prologis is a bet that its existing logistics platform provides the optimal launchpad for the AI infrastructure race. For investors, this offers a way to capture the data center boom through a more familiar, operationally proven vehicle.

The bottom line is a hierarchy of opportunity. Data centers are the primary growth engine, but success depends on solving the power puzzle. Office space offers a selective recovery play, contingent on location and adaptability. And the industrial sector provides a distinctive hybrid model, where proven operational expertise is being leveraged to capture value in the new digital infrastructure. The investment thesis is clear: look past the bricks and mortar to the megawatts, the grid connections, and the strategic platforms that will define the AI era.

Catalysts, Risks, and What to Watch

The structural shift we've outlined is now in motion, but its success and durability hinge on a handful of forward-looking factors. The primary bottleneck is clear: power infrastructure. The grid's ability to connect and serve these new, massive data centers will determine the pace of the entire build-out. The leading challenge for development is grid stress, with a seven-year wait on some requests for connection. This constraint is not a minor delay; it is a fundamental gatekeeper that will favor developers with strong utility partnerships and the capital to navigate complex permitting. Watch for signs of progress in grid modernization and the formation of strategic alliances between data center builders and energy providers. Any easing of this bottleneck would be a major positive catalyst for the sector.

A parallel risk is emerging on the supply side. The massive capital deployment into data centers, coupled with the projected peak in occupancy, raises the specter of oversupply. Goldman Sachs Research forecasts that occupancy will peak at more than 95% in late 2026, followed by a moderation starting in 2027. This cycle could be prolonged if efficiency gains from new hardware or software reduce the capital expenditure needed per unit of compute. The report notes that efficiency driving lower capex levels could mitigate the risk of long-term market oversupply. Investors should monitor for signs of slowing capex plans from hyperscalers and new entrants, as this would signal a more durable, less cyclical market.

For the office market, the key metric is the quality and longevity of the AI leasing surge. The current demand is concentrated, with tenants often opting for shorter two- to three-year leases with expansion clauses. The recovery story will only gain credibility if this activity translates into longer-term commitments and higher occupancy in secondary markets, not just in Manhattan and San Francisco. The broader office market's high vacancy rate and ongoing tech layoffs remain powerful headwinds. Watch for data showing a broadening of demand beyond the initial tech hubs and a shift toward more permanent tenancies.

The bottom line is a race against two constraints: the physical grid and the economic cycle. Success will go to those who can secure power and build efficiently. For now, the catalysts are clear-grid partnerships and capex discipline-but the risks of oversupply and a fragile office recovery are real. The coming quarters will reveal whether this is a durable structural shift or a cyclical boom.