Data Center Valuation in an AI-Driven Era: Bridging the Gap Between Digital Infrastructure and Real Estate Life Cycles (Part 2)

The rapid expansion of data center development has revealed a structural tension that is now central to valuation: the widening gap between the life cycle of digital infrastructure and that of the underlying real estate. While data center shells and land retain economic lives measured in decades, the critical systems housed within them—power, cooling, networking and computing—are evolving in dramatically shorter timelines. This divergence is reshaping how investors, lenders, and valuation professionals assess risk, cash flow durability, and long-term value. 

As the second installment in SitusAMC’s four-part series on data center valuation trends, this article examines how that life-cycle mismatch manifests in underwriting and appraisal—and the contractual, financial and structural mechanisms emerging to manage it. We focus on the challenges valuation professionals must now address, the market responses taking shape and the implications for income durability and exit pricing. 

The Valuation Challenge: When Infrastructure Ages Faster Than Real Estate 

Few asset classes face the pace of technological change confronting data centers today. Advances in chip architecture, rack density, cooling methods and power delivery are compressing useful lives for core infrastructure. Facilities designed for yesterday’s workloads may struggle to compete as AI-intensive deployments demand radically different configurations. 

S&P Global Ratings has cautioned that the “winning odds” for legacy data centers are becoming less certain, as technological change increases the risk that older facilities may become functionally obsolete well before the end of their physical lives. AI workloads require exponentially more power and cooling, increasing the likelihood that older assets will face value impairment without continuous reinvestment. For appraisers, this raises fundamental questions: 

• What is the remaining useful life of the current improvements, based on functionality and/or current market standards? 
• How much capital reinvestment should be assumed to maintain competitiveness? 
• At what point does retrofitting become unsupportive relative to redevelopment? 

The answers vary widely by asset quality, location and tenant profile—but measuring underlying risk plays a key role in the valuation of data centers into the future. 

Power Timelines, Interconnection Risk and Front-Loaded Capital 

The life-cycle gap is also apparent with power infrastructure. Securing power through power purchase agreements (PPAs) and grid interconnections can take 12 to 36 months—often longer than vertical construction itself. Yet capital expenditures are incurred well before revenue begins. This creates a valuation challenge at every stage: 

• Development risk: capital deployed long before future (speculative) leasing 
• Underwriting risk: assumptions around timing, rent escalations and utilization 
• Appraisal risk: discount rates increasingly sensitive to delays 

Hyperscale data centers — known for their enormous size, cutting-edge infrastructure, built-in scalability and capacity to power massive amounts of data—now require hundreds of millions, if not billions, of dollars in upfront spending for power, cooling and interconnection—often years ahead of meaningful cash flow. Each phase introduces uncertainty, and each delay compounds execution risk. From a valuation perspective, this front-loaded investment profile magnifies the consequences of timing misalignment. A modest delay in power delivery can materially impact expected returns, timing of those returns and therefore, lender appetite. 

“Pull-Forward” Demand and Structural Credit Risk 

The surge in AI-driven development has also raised questions about demand timing. Moody’s and other credit analysts have highlighted the inherent risk of “pull-forward” capacity—in which developers build next-decade infrastructure today to secure land, power and equipment before shortages worsen and pricing increases. While most market participants do not foresee a near-term bubble, the sheer scale of capacity under development makes risk assessment essential.  

Even hyperscalers are signaling caution. Reuters reported in mid-2025 that Microsoft slowed the pace of new data center leasing, citing internal supply-chain checks—despite reaffirming its plan to invest roughly $80 billion in AI and cloud infrastructure. Since then, Microsoft has addressed this pullback by stating it was forging ahead with AI investments and data center developments, but being diligent about leasing pace, just as in other investment segments. 

This movement underscores the importance of disciplined absorption assumptions. If buildouts continue to outpace realized demand—even temporarily—valuation professionals must carefully assess lease-up timing, renewal probability and exit liquidity. 

Operational Performance and Lease Economics 

Modern data center leases increasingly resemble operating contracts rather than traditional net leases. Service-level agreements (SLAs) tied to uptime, power delivery and cooling performance are now standard, particularly for hyperscale and AI tenants. Failure to meet performance thresholds can trigger rent abatements, credits or termination rights, introducing operational-based cash-flow volatility not particularly common for most commercial real estate asset classes. 

As a result, lenders and investors are underwriting not only tenant credit, but also operational reliability. For valuation professionals, income durability must now rely on performance covenants, revenue floors, cure provisions and exposure to SLA-driven deductions. Lease analysis has become inseparable from operational risk assessment. 

Market Responses: Structuring Around the Life-Cycle Gap 

In response to the growing mismatch between infrastructure life cycles and real estate durability, market participants are increasingly relying on contractual and structural solutions rather than physical design alone. Longer PPAs and lease terms, early termination rights, take-or-pay power structures and vertical integration are all being used to reallocate risk across tenants, owners, utilities and capital providers. Together, these mechanisms are reshaping how income durability, execution risk, and long-term value are underwritten. 

Longer PPAs and Extended Lease Terms 

One of the most direct responses to life-cycle mismatch has been the extension of lease terms, particularly for hyperscalers. Longer, 10 year+ lease commitments are increasingly common, aligning revenue visibility with the economic life of power and cooling systems. But this raises the question of whether a 10- to 20-year legal commitment can truly predict and cover unforeseen changes that may occur as a result of ever-changing technology and a rapidly evolving industry. Power purchase agreements have also grown more complex, reflecting political, regulatory and Environmental, Social and Governance (ESG) considerations. Hyperscalers such as Microsoft, Amazon, Google, and Meta are using long-term PPAs not only to secure power, but to hedge price volatility and meet sustainability mandates. 

Early Termination Rights and Construction Risk 

At the same time, hyperscalers are negotiating early-exit provisions to protect against construction delays and technological uncertainty. These clauses transfer meaningful risk to developers, investors and lenders, particularly where delays stem from factors outside the owner’s control. 

Data center securitization has expanded rapidly, with ABS and single-borrower CMBS issuance exceeding $50 billion over the past five years—bringing lease termination risk directly into structured finance markets. We expect that underwriting and lender perspectives will continue to evolve as a result. 

Take-or-Pay Power Structures and Utility Tariffs 

Utility companies are increasingly requiring take-or-pay provisions for large data center loads, obligating operators to pay for reserved capacity regardless of actual usage. For example, Duke Energy recently announced that new contracts for data centers requiring more than 100 MW of power in its territory will include minimum take-or-pay provisions. This means that operators must pay for a baseline amount of electricity whether they use it or not—a structure designed to help the utility manage the cost and risk of serving large, uncertain loads, and to secure regulatory approval for these agreements. 

These structures protect utilities and ratepayers but increase fixed operating costs for data center owners. In some jurisdictions, public scrutiny has intensified around who ultimately bears the cost of grid expansion. For example, a Harvard analysis highlights how utilities have sought to recover grid expansion costs associated with large technology loads through rate structures that ultimately fall on residential and small business consumers, prompting heightened public and regulatory debate over whether ratepayers should shoulder expenses tied to big tech’s data center growth. From a valuation standpoint, these provisions raise breakeven occupancy levels and increase sensitivity to utilization assumptions. 

Vertical Integration as a Risk Mitigation Strategy 

Vertical integration is emerging as another response to life-cycle mismatch. Developers and investors that control development, power sourcing, construction and operations can reduce revenue ramp-up risk and standardize delivery at scale. 

Regulated utilities with integrated generation and load planning may be best positioned to serve data center demand. For example, utilities with coordinated generation and load planning can more effectively manage the capacity needs of hyperscale and AI-oriented data centers, notes consulting firm Wood Mackenzie. This format positions utilities to offer more reliable service and potentially more predictable cost structures—factors that can improve underwriting assumptions around power delivery and timeline risk. 

Recent examples include vertically integrated development platforms and large infrastructure partnerships designed to align long-lived power assets with shorter technology cycles. AWS (Amazon Web Services) is an example of end-to-end data center development and operation, in which the firm controls power generation and renewable sourcing, network infrastructure and proprietary server and chip design. Through vertical integration, they are able to own and control facilities, design hardware, and operate workloads. This compresses margins for traditional data center landlords and changes underwriting assumptions for hyperscale demand durability. Coreweave offers another model of AI-focused, new model vertical integration. The company has blurred the lines between tenants and operators by anchoring facilities, driving design specs and securing financing with long-term commitments similar to hyperscalers. Google has also been a large proponent of purpose-built data centers, emphasizing bespoke, in-house data center design, reducing its dependence on third parties and prioritizing long term efficiency. 

Assessing the Dynamics: What Valuers Must Now Model 

Across these strategies, the common thread is contractual complexity. Life-cycle risk is increasingly addressed through lease language and operating frameworks rather than physical design alone, requiring valuation professionals to move beyond traditional rent-and-term analysis. 

Technology refresh obligations, density flexibility and capex-sharing provisions are becoming more explicit as operators and tenants attempt to manage the pace of infrastructure obsolescence. Service-level agreements are now central to this transfer of risk. Uptime guarantees tied to power delivery and operational resilience are increasingly supported by specialized insurance solutions designed to cover both physical infrastructure failure and business interruption risk—reflecting the growing financial consequences of outages in AI-driven facilities. 

Power access itself is also being treated as a hedgeable asset. Data center operators are increasingly using long-term power contracts, fixed-price arrangements and structured PPAs to manage exposure to energy price volatility and grid uncertainty, effectively embedding energy risk management directly into lease and underwriting assumptions.
 At the asset-level, monthly recurring charge (MRC) floors are playing a growing role in protecting fixed-cost recovery. MRCs—distinct from one-time non-recurring charges (NRCs)—provide predictable baseline revenue covering space, power and cooling, even as usage fluctuates. For appraisers, these structures improve visibility into downside protection but require careful modeling of utilization and escalations.
 

Conclusion: Valuing Assets Built on Uneven Timelines  

The widening gap between digital infrastructure and real estate life cycles is no longer theoretical—it is shaping lease structures, financing terms and valuation outcomes today. For valuation professionals, the implications are clear: 

• Valuation assumptions must consider technological realities, which may include significant capital costs and/or capital reserves. 
• Income durability depends as much on operational performance as on tenant credit. 
• Execution risk rivals demand risk as a driver of value. 
• Power access has become a hedgeable, contractual instrument. 

As AI continues to reshape computing needs, data center valuation will hinge less on traditional real estate metrics and more on adaptability, reliability and alignment between capital spend and revenue realization. 

SitusAMC is the leading provider of independent commercial real estate valuation and review services, offering valuation management, daily valuation, appraisals and more. For more information on SitusAMC’s Valuation services for data centers, contact Heather Byrnes, Director, at heatherbyrnes@situsamc.com, or visit our website. 

Read Part 1 of our Data Center series here. Up next in the series: How power reliability challenges, power grid constraints and environmental issues are reshaping pricing, feasibility and underwriting across global data center markets.