Powering Data Centers

Megawatts to Megabytes: Orrick’s Guide to Developing, Financing & Powering Data Centers
8 minute read | November.20.2025

How can data center developers secure reliable power when U.S. demand is growing 23% annually and interconnection delays now stretch to 5+ years? This guide examines practical strategies for powering data centers amid grid constraints and clean energy requirements.

The guide covers power generation options from nuclear and geothermal to natural gas and fuel cells, grid interconnection strategies including co-located generation and replacement rights, and contracting mechanisms such as Clean Transition Tariffs, behind-the-meter PPAs, and virtual PPAs. It also addresses key regulatory requirements including FERC interconnection rules, state supply restrictions and market-based rate authority.

Read this excerpt from our guide Megawatts to Megabytes below or download the full PDF here 

Contents

Introduction

According to McKinsey’s analysis, by 2030, U.S. data center power demand alone is expected to rise by 400 TWh, growing at 23% annually, and could represent 30%–40% of new net demand. This demand would require $500 billion in infrastructure investment. The size and scale of data centers are constantly expanding, with “hyperscale” data centers that require at least 100 MW or more of capacity accounting for around 41% of worldwide data center capacity. Hyperscale data center capacity is projected to exceed 60% by 2029 (Synergy Research Group). The rapid expansion of data centers is driving unprecedented energy demands, necessitating very significant strategic investments and innovative utility tariff designs to help streamline regulatory processes and ensure that the rapid growth can be met by dedicated clean resources.

To meet their customers’ needs, data center operators and users require 24/7 energy deliverability, fast ramping capabilities for peak demand periods, and high redundancy. With their unique power demands, the rapid growth of data centers is raising concerns among utilities, particularly in states lacking customer choice, as new resources are needed to meet this demand. Specifically, utilities cite concerns about integrating these new loads into their systems while maintaining reliability, affordability, and sustainability. They fear price increases for non-benefiting customers and the potential of stranded assets due to technological changes or a drop in data center demand. Moreover, with many data center users pursuing low-carbon or carbon-free goals, there is an added challenge in meeting data center growth with qualifying resources. To achieve these multifaceted goals, data center-focused clean utility tariffs may facilitate the expedited development of dedicated clean energy resources, equitable financing mechanisms, and appropriate cost allocation structures.

^ top

Chapter 1.

Key Considerations for Data Center Developers and Grid Capacity

In this chapter: Data center grid interconnection challenges including transmission infrastructure delays that now extend into the decade-long range in some jurisdictions, and developer options for addressing capacity constraints such as onsite generation, islanded sites and relocating to regions with available capacity.

Data centers are among the most energy-intensive infrastructure assets—and demand not just large volumes of power, but near-perfect reliability. The industry standard of “Five 9s” (99.999% availability) permits only about five minutes of downtime per year, making fully firm, uninterrupted grid supply essential. Most developers pursue grid interconnection capacity sufficient to always meet peak demand, typically through utility upgrades or direct transmission investment. Depending on region and scale, this process can take months or years for hyperscale projects. But that model is rapidly becoming unworkable.

“Data centers are among the most energy-intensive infrastructure assets – and demand not just large volumes of power, but near-perfect reliability.”

A System Under Strain

The explosive growth of data center demand, combined with the broader electrification of transport and industry, is pushing transmission infrastructure to its limits. In many jurisdictions, delays for new connections now stretch into the decade-long range. The UK, for example, has seen 10+-year wait times in some regions. In other parts of Europe, developers must secure new dispatchable generation or storage to qualify for interconnection capacity.

Many governments recognize the urgency. Yet policy reform and grid expansion are slow-moving by nature. Many developers simply cannot wait—especially given the speed at which AI and digital services are expanding.

Data Center Developer Dilemmas: What Are the Options?

Faced with grid constraints, data center developers are increasingly forced to choose between:

  • Waiting for long-delayed transmission upgrades
  • Building onsite generation to compensate for non-firm grid supply
  • Constructing fully islanded sites with 24/7 self-generation
  • Relocating to regions where capacity is still available
  • Acquiring grid access on the secondary market, such as purchasing decommissioned industrial land with legacy connections

For developers committed to proximity—whether for latency, fiber infrastructure, or customer requirements—onsite generation often becomes the most attractive option. But it brings trade-offs and complexity.

^ top

“The winners in this space will be those who can navigate the technical, regulatory and reputational complexity—and bring resilient, scalable power solutions to the front lines of the digital economy.”

Chapter 2.

Grid Interconnection Strategies: Co-Located Generation vs. Network Load Service

In this chapter: Comparing direct grid interconnection as network load versus pairing data centers with behind-the-meter generation, including FERC interconnection reforms, surplus interconnection rights, replacement interconnection rights and energy park concepts that integrate multiple generate assets behind one point of connection.

Grid capacity constraints are redefining data center development. While firm grid interconnections remain the ideal, constrained transmission infrastructure is forcing developers to consider complex alternatives. Similar to co-generation facilities previously popular for industrial manufacturing sites, onsite gas generation may be a necessary first step in some cases, but it must be carefully structured and future-proofed to align with both commercial objectives and carbon reduction imperatives.

The winners in this space will be those who can navigate technical, regulatory, and reputational complexity— and bring resilient, scalable power solutions to the front lines of the digital economy.

Connecting hyperscale data center load to the interstate transmission grid is an increasingly complex, costly, and time-consuming process. Developers must carefully weigh the trade-offs between direct interconnection as a “network load” and pairing with behind-the-meter generation. Each path offers unique benefits and risks across timing, reliability, regulatory treatment, and cost recovery.

Network Load Service: Direct Interconnection Without Co-Located Generation

In the traditional model, a data center connects directly to the transmission grid and is designated by the utility as a network load. The transmission provider studies the service request and, if upgrades are needed, finances them upfront and recovers costs from data centers through its transmission tariff. The data center developers may be required to post security for the upgrades, which is typically refunded upon energization.

This approach often has shorter study timelines than those involving new generation, but also presents notable downsides:

  • No dedicated power generation, leaving the data center fully reliant on grid conditions.
  • Curtailment risk during transmission outages or congestion.
  • Potentially higher wholesale energy costs, especially if located far from generation centers.

Co-Located Generation with Grid Interconnection

Co-located generation involves pairing a data center with a new or existing power plant. Co-located generation provides enhanced reliability and may accelerate energization by enabling the data center to interconnect using the generator’s existing or pending interconnection agreement. This structure can shield the data center from curtailment and offer greater control over energy sourcing. However, it brings significant regulatory and practical complexity.

Interconnection of new generation is significantly delayed across many regional transmission organizations (RTOs) due to study backlogs.

The standard timeline for completing an interconnection agreement is approximately three years from the initial request to the final agreement, with delays stretching to more than six years. 

RECENT REFORMS

In 2023, the Federal Energy Regulatory Commission (FERC) introduced a sweeping overhaul of the interconnection process to address these delays:

  • Transition from "first-come, first-served" to a "first-ready, first-served" model
  • Projects must demonstrate 90% site control and post a commercial readiness deposit equal to 10% of expected network upgrade costs
  • FERC began imposing penalties on transmission providers for missed study deadlines
  • While implementation of those reforms is underway, the benefits may take years to fully materialize.

Accelerated Paths: Surplus and Replacement Rights

To bypass long interconnection queues, data center developers are exploring alternative interconnection mechanisms:

  • Surplus interconnection rights: This approach involves leveraging underutilized capacity from existing generators. For example, where an aging coal plant operates at only 20% capacity, a developer may co-locate a new solar plant to supply the remaining interconnection capacity. By attaching to an existing interconnection agreement, the solar plant skips the interconnection queue. Although this approach may offer lower costs and faster approval, the surplus interconnection is subordinate to the primary interconnection—if the host loses its rights, so does the surplus project.
  • Replacement interconnection rights: Under this approach, a proposed generator can inherit interconnection rights from a retiring facility, provided it does not exceed the prior injection level. This approach may avoid the need for full studies and allow for fast-track development. However, it may require joint ventures or transition agreements with the retiring asset owner to prevent queue jumping.
  • Energy Parks: This approach involves integrating multiple generation assets, storage solutions, and co-located loads behind one point of interconnection as a form of large-scale microgrid. Energy parks offer data centers cost savings and faster energization; however, integrating energy parks into the grid may require changes to current market rules, and they may be difficult to finance given their size and the large number of parties and technologies involved. 

^ top

Chapter 3.

Selection of a Power Technology

In this chapter: Evaluating power generation technologies for data centers including natural gas combined-cycle plants, large-scale nuclear and small modular reactors, geothermal baseload power, solar and wind with battery storage, fuel cells, and grid-enhancing technologies such as dynamic line rating.

The energy sources and technologies available to power data centers present several important considerations, including reliability, scalability, cost, regulatory compliance, and environmental impact, among others. Selecting a suitable power technology requires an assessment of the relative benefits and drawbacks of each option.

We note that, while not specifically addressed in this report, power facilities constructed to serve data centers will require their own financing. Project financings and, to the extent applicable, tax credit monetization transactions are often the predominant financing structure for such projects. Data center developers will need to consider this as they source their power supply.

Natural Gas

Natural gas-fired power plants are a critical tool for ensuring reliable, dispatchable, large-scale energy delivery—especially for hyperscale data centers that require hundreds of megawatts of capacity with near-perfect uptime. For decades, natural gas has served as a cornerstone of industrial power systems, and it remains one of the few technologies capable of balancing scalability, geographic flexibility, and dispatchability.

For developers and investors looking to meet immediate data center power needs, natural gas offers a distinct combination of advantages:

  • Reliability and Dispatchability: Natural gas plants provide stable, on-demand baseload and peaking capacity, essential for mission-critical data operations.
  • Siting Flexibility: Plants can be developed near data center campuses in industrial zones, avoiding lengthy transmission buildouts or congested grid interconnections.
  • Scalability: Natural gas facilities can be built to match the 50–500+ MW scale now typical of AI and hyperscale campuses.
  • Lower Emissions Compared to Legacy Fuels: Compared to diesel or coal, natural gas emits significantly less CO₂, NOₓ, SO₂, and particulates—making it a cleaner option for near-term deployments.
  • Combined Heat and Power (CHP): Recovered thermal energy can reduce a data center’s cooling energy demand, increase total system efficiency, and improve project economics.

Developers and operators are also leveraging next-generation gas technologies and efficiency upgrades to reduce emissions and increase performance. Today’s plants can be equipped with:

  • Combined-Cycle Gas Turbines (CCGT), which improve efficiencies up to 60% by capturing and reusing waste heat.
  • Selective Catalytic Reduction and low-NOx burners to reduce air pollutants.
  • Carbon Capture and Sequestration (CCS) systems to mitigate CO₂ emissions at the point of combustion—an increasingly important (but expensive) feature for regulatory compliance and ESG performance.

These innovations allow developers to deploy natural gas solutions that meet current emissions requirements, and they’re also better positioned for a future energy mix that includes hydrogen blending, renewable fuels and low-carbon operational mandates.

ONSITE GAS: A VIABLE BUT COMPLICATED OPTION

Natural gas-fired generation is often the best onsite solution for firm, dispatchable power. However, while cleaner than coal or oil, it still produces emissions and may conflict with publicly stated carbon goals, exposing developers to reputational and regulatory risks.

How gas is deployed significantly affects its classification:

  • Direct supply to the data center typically results in scop 2 GHG emissions, indirect emissions as the result of purchased energy, which can be offset using renewable energy certificates (RECs or Guarantees of Origin), particularly if the generator is now owned or operated by the data center itself.

    However, direct linkage increases public visibility and reputational exposure, especially for sustainability-minded operators.
  • Balancing interconnection needs, carbon reporting and public commitments becomes a strategic challenge. Solutions will depend on a developer's risk appetite, sustainability goals, and investor expectations.

Despite its strengths, natural gas power presents several challenges that must be addressed through proactive planning and risk mitigation:

  • Equipment Procurement Lead Times: Global supply constraints and high demand have led to multiyear lead times for major components like gas turbines and heat recovery steam generators (HRSGs). Developers may need to secure procurement contracts early in project cycles.
  • Fuel Supply and Infrastructure: Viable projects require access to high-pressure gas lines, compression infrastructure, and contingency fuel strategies. Siting must be coordinated with pipeline networks and local permitting authorities.
  • Commodity Price Risk Management: Data center developers looking to utilize natural gas power solutions may need to bear (at least partially) commodity price risk of natural gas and may consider hedging and other contractual solutions to mitigate the short- and long-term risks.
  • Environmental and Regulatory Risks: While natural gas-fired generators emit fewer CO emissions (especially as compared with diesel backup generators that are often used in data center power systems), they are not considered a long-term clean energy solution unless paired with CCS or renewable fuels. Additionally, incidental releases or leaks of natural gas in transportation emit methane (CH4), which is a more potent greenhouse gas compared to CO2. Projects may face regulatory scrutiny or declining policy support in jurisdictions with aggressive decarbonization targets.

FUTURE-PROOFING: FROM GAS TO ZERO-CARBON

To mitigate long-term risks, developers considering natural gas must look ahead. Projects should incorporate clear pathways for transitioning to lower-carbon fuels like green hydrogen or implementing CCS as soon as technology and economics allow. Structuring these projects to evolve with policy, technology, and investor requirements may become critical to preserving optionality and the social license to operate.

Nuclear

Nuclear power has provided stable baseload power to the grid for decades and presents a compelling option for data centers seeking stable carbon-free energy solutions. Nuclear power, with its ability to provide consistent baseload power, offers significant advantages in meeting demand for clean, reliable baseload power.

Large-Scale Nuclear Power

Traditional nuclear power plants have long been recognized for their capacity to generate substantial amounts of electricity with minimal carbon emissions. A single traditional nuclear reactor typically generates approximately one gigawatt of electricity with availability of over 90%. These facilities are a natural fit for hyperscale data centers, which often require hundreds of megawatts to operate efficiently. However, the substantial capital expenditure and construction time associated with building and permitting new traditional nuclear power plants is an obstacle to development.

In the last 20 years, only three new traditional nuclear power plants have been built and commissioned in the United States: Watts Bar Unit 2 (2016), Vogtle Unit 3 (2023), and Vogtle Unit 4 (2024). As a result, data centers primarily focus on entering power purchase agreements that support recommissioning nuclear power plants previously decommissioned or extending the life of currently operating nuclear power plants. Given the limited number of options available for recommissioning and extensions, many data centers are instead focusing on small modular reactors for future development plans.

Small Modular Reactors (SMRs)

SMRs represent an innovative approach to nuclear energy, offering a more flexible and scalable solution for data centers. These reactors are smaller in scale, ranging from 1 to 10 MW (also known as microreactors) to approximately 350 MW of output. Their smaller size results in lower capital expenditures than traditional nuclear power reactors. They may also be deployed in a modular fashion, standardizing construction and operation across multiple projects and allowing a single site to host multiple units built together or sequentially. The proposed designs of SMRs incorporate advanced safety features including passive cooling systems and the use of advanced nuclear fuels designed to avoid reactor malfunctions and core compromises. These advanced safety features mean that SMRs can have smaller physical footprints compared to traditional nuclear power plants, which require large exclusion zones under current regulatory standards. This allows SMRs to be located closer to data centers and supports future behind-the-meter deployment.

SMRs deploy existing fission reaction technology but are generally considered first-of-a-kind from a regulatory, financing, and market perspective. In recent years, developers have been forced to navigate challenging and complex regulatory landscapes. However, nuclear power development currently benefits from bipartisan political support as a carbon-free resource with substantial economic benefits. In fact, Congress recently enacted the ADVANCE Act, which directs the U.S. Nuclear Regulatory Commission to reduce licensing application fees and expedite the licensing process. The current U.S. administration also enacted several executive orders to streamline the review and approval process for new nuclear reactors and provide other forms of government support. While the present regulatory scheme is a challenge to SMR development, the evolving political landscape is expected to reduce regulatory burdens and support SMR innovation.

Geothermal

Geothermal power—unlike many other renewable energy sources—can serve as a baseload resource with 24/7 firm availability. This makes it uniquely suited for powering critical infrastructure such as data centers, which require constant, reliable electricity regardless of weather or time of day.

According to the U.S. Department of Energy, geothermal energy could supply up to 120 GW of generation capacity in the U.S. by 2050—enough to meet over 16% of projected national electricity demand. This long-term vision reflects not only the environmental benefits of geothermal energy, but also its role in stabilizing grids with a growing share of variable wind and solar generation.

Geothermal is gaining traction in regions where traditional renewables face siting, land use, or intermittency challenges—particularly in the Asia-Pacific region, where population density and grid reliability constraints complicate solar and wind deployment. Unlike solar and wind farms, geothermal plants have a compact footprint and minimal visual impact, making them more compatible with urban or industrial zones.

However, geothermal energy has historically been underutilized due to high upfront capital costs and geological limitations. Traditional geothermal technologies are economically viable only where high-temperature resources are easily accessible near the surface. That landscape is now shifting rapidly due to technological innovations. Advances in drilling and subsurface engineering—adapted from the oil and gas industry—are unlocking deeper and more complex geothermal reservoirs. Recent breakthroughs include:

  • Enhanced Geothermal Systems (EGS): Using techniques such as hydraulic stimulation to create artificial reservoirs in otherwise dry hot rock.
  • Horizontal Drilling: Increasing reservoir contact and thermal output.
  • Closed-Loop and Modular Geothermal: Innovative systems that do not require water-intensive open reservoirs.

These advancements should extend the geographic viability of geothermal power into areas previously considered uneconomic or geologically unsuitable.

Another accelerating trend is the co-location of data centers with geothermal resources, particularly in regions with supportive regulatory frameworks and high energy demand. Tech companies are leading the way:

  • Google and Baseload Capital (Taiwan): Google signed a corporate Purchase Power Agreement (PPA) to support a new 10 MW geothermal facility to power its Taiwanese data center and chip manufacturing operations — marking one of the region’s most advanced clean energy integrations.
  • Google and NV Energy + Fervo Energy (Nevada): In the U.S., Google is pursuing an innovative arrangement where up to 115 MW of geothermal energy from Fervo Energy will be used to power its growing data center load in Nevada. Fervo’s use of fiber-optic sensing and horizontal drilling is setting new standards for geothermal monitoring and efficiency.

Beyond electricity generation, geothermal energy can also be used to support direct cooling of data centers using geothermal heat pumps or absorption chillers. These systems can significantly reduce electricity demand for cooling (often 30%–40% of total data center load) as well as water consumption, a growing sustainability and regulatory concern in many regions.

Solar and Wind Resources

Renewable energy resources, such as wind and solar, are commonly used to power data centers. However, data centers may not be able to rely solely on these generation sources for 24/7 uptime due to the inherent intermittency of wind and solar generation.

To achieve reliability, developers may need to supplement wind and solar generation with:

  • Deliveries of conventional power from the market;
  • Diesel backup generators;
  • Battery storage that stores renewable energy to use at times when renewable energy generation is unavailable; or
  • In some cases, hydropower, nuclear or geothermal.

There is also a business case for utilizing existing wind or solar projects as sources of power for data centers. Typically, during periods of negative pricing in ISO regions, renewable energy producers may choose to curtail generation instead of paying the grid operator to generate in the negative price environment. However, by co-locating a data center at the wind or solar site, this excess power can be directed to power the data center load rather than being curtailed due to negative pricing. In this type of co-located arrangement, the data center remains connected to the grid to ensure a continuous power supply when the renewable source is not generating. This approach provides a win-win situation, both optimizing the renewable energy resource and maximizing revenue for the renewable energy producers who otherwise may have curtailed the supply.

Fuel Cells

Fuel cells are a clean energy technology that generate electricity through an electrochemical reaction—typically using hydrogen and oxygen—without combustion. This process results in significantly lower emissions compared to traditional fossil-fuel-based generation, with water and heat as the primary byproducts. As the global demand for digital infrastructure surges, fuel cells are emerging as a scalable, dispatchable power solution for data centers.

In some cases, fuel cells provide continuous, behind-the-meter power directly to data centers, reducing reliance on the grid and avoiding the risks associated with transmission congestion or interconnection delays. They can also serve as an alternative to diesel generators for providing backup power during grid outages, offering faster start-up, lower emissions, and compliance with increasingly strict air quality regulations.

Fuel cells offer high reliability—often with availability factors above 99.9%—making them well-suited to meet the always-on power requirements of hyperscale and co-location data centers. Additionally, data center developers often prioritize fuel cell systems for their relatively fast deployment timelines and easier permitting compared to traditional generation.

This makes them strategically valuable as a “bridge” resource, providing interim power for several months or years while larger permanent grid infrastructure is built or upgraded.

From an investment standpoint, fuel cells can help mitigate the power delivery bottlenecks that have delayed or constrained new data center developments in key U.S. and European markets. In regions where grid interconnections are projected to take 3–5 years or longer, fuel cell installations can enable earlier revenue generation from data center assets—improving project IRRs and unlocking portfolio value.

Major technology firms and data center operators have deployed fuel cells at scale:

  • Equinix has partnered with Bloom Energy to install fuel cell systems at 19 of its U.S. data centers, with a total capacity exceeding 100 MW. These systems operate continuously, offsetting grid power and contributing to Equinix’s long-term sustainability goals.
  • Additional pilot projects by Amazon, Google, and Meta are exploring both natural gas-powered and green hydrogen fuel cell technologies as part of broader efforts to decarbonize data center operations and enhance grid independence. 

Key considerations for data centers looking to deploy fuel cells include:

  • Technology Maturity: While natural gas-based solid oxide fuel cells are commercially mature, hydrogen fuel cells are still gaining traction as green hydrogen becomes more available.
  • Policy and Incentives: Federal and state-level incentives enhance the economics of fuel cell projects, particularly those utilizing low-carbon or renewable fuels.
  • Siting and Permitting: Modular design and lower emissions profiles make fuel cells easier to permit than diesel generator sets or gas turbines—often enabling deployment in urban or constrained areas near data hubs.
  • Revenue Models: Opportunities exist in long-term energy-as-a-service (EaaS) agreements with data center operators, as well as potential grid services revenues through demand response or capacity markets.
  • Environmental Considerations: While hydrogen fuel cells create electricity without emitting any CO2, overall emissions depend on whether the hydrogen is produced using renewable energy (green hydrogen) or natural gas (blue hydrogen).

Energy Storage

The case for utilizing energy storage—including battery energy storage systems (BESS) and other storage technologies—to help manage the power supply equation for data centers is compelling. BESS can provide the following functionalities, depending on ISO/RTO status and state regulatory requirements, and subject to the constraints and challenges described below:

  • Load-Matching: If co-located with solar, gas, or other power generation resources, BESS can potentially be used to help address data center power demand. A BESS with a four-hour duration, however, is not likely to satisfy the 24/7 load-matching goals of hyperscale data centers. Long-duration energy storage (LDES) solutions, although costly, are considered viable alternatives—and multiple LDES technologies and vendors have commercially viable projects already in operation.
  • Time-Shifting: A stand-alone BESS, which is a battery interconnected to the grid without a co-located generation resource, can potentially be used to decrease a data center’s energy costs by arbitraging market prices (charging when local marginal prices (LMPs) are low and discharging when LMPs are high). The viability of this approach depends on whether the data center is in an ISO, whether the BESS is behind or in front of the data center’s revenue meter, and other regulatory considerations.
  • Reliability: A BESS placed behind the data center’s revenue meter can serve a reliability function, including during power failures. However, installing a BESS system is costly, and the relatively short duration of most BESS reduces its attractiveness.

Although the potential for utilizing BESS to support data center requirements exists, not many BESS-data center transactions have occurred to date. The relative costliness of BESS (combined with recent uncertainty in U.S. tax credit (now resolved with President Trump’s signing of the “One, Big, Beautiful Bill” on July 4, 2025) and tariff policies), short BESS durations, and regulatory considerations are all contributing factors. In addition, the inconsistency across the U.S. of mature capacity markets, and changing capacity accreditations for BESS have resulted in similarly inconsistent incentives for BESS deployment. As data center power demands increase and U.S. capacity markets develop in the coming years, however, we believe that mature BESS technologies are poised to play a critical role in addressing these demands.

Other Technology Solutions

As demand pressures on the grid increase – due in part to the proliferation of data centers – and obstacles to expanding the transmission network remain significant, grid-enhancing technologies have become critical tools to increase the existing network’s capacity, reliability, and efficiency. These technologies encompass a variety of innovations, such as:

  • Sensor Technologies: By deploying sensing and monitoring devices throughout transmission and distribution systems, utilities can gather real-time data on condition and performance. These devices enable early detection of operational issues, leading to improved reliability and efficiency and reduced maintenance costs. As extreme weather events become more frequent due to climate change, these technologies are increasingly vital for monitoring the stability of transmission and distribution infrastructure.
  • Dynamic Line Rating (DLR): DLR systems are installed along transmission lines and monitor real-time weather conditions and other environmental factors. In response to these conditions, the DLR technology modifies the rating assigned to the line. As a result, DLR technology provides data on a transmission line’s real-time power carrying capacity, as well as forecasted carrying capacity. This granular data helps to maximize existing transmission infrastructure by ensuring energy generators fill remaining capacity.
  • Flexible Alternating Current Transmission Systems (FACTS): FACTS technology improves network efficiency by increasing control over power flow, reducing power loss, and maintaining power quality. Distributed FACTS (D-FACTS) is a scaled-down version of FACTS with the benefits of lower cost and easier installation.
  • System-Wide Load Flexibility: System-wide load flexibility refers to the power grid’s ability to integrate large, controllable energy demands—such as data centers or electric vehicle charging—by allowing these loads to temporarily reduce or shift their electricity usage during periods of grid stress. This approach could enable the addition of substantial new data center loads without necessitating significant infrastructure upgrades, thereby maintaining grid reliability and affordability; however, it requires that data center operators and hyperscalers are willing to lose power supply for around 40 hours each year (0.5% of the time) when power supply is most constrained, which is well above the 99.999% reliability that the industry strives for.

^ top

Chapter 4.

Contracting for Power

In this chapter: Power procurement mechanisms including utility tariff evolution for data center load, green tariffs for renewable energy credits, 24/7 clean utility tariffs for hourly matching, Clean Transition Tariffs that provide dedicated clean resources without cost shifts to other customers, behind-the-meter PPAs, build-transfer arrangements, virtual PPAs, REC transactions and hedging strategies.

To meet the unique energy demands of data centers, there are a variety of contractual arrangements and mechanisms that provide strategic opportunities for securing power while achieving environmental goals.

The Evolution of Utility Tariffs to Serve Data Center Load

Behind-the-meter strategies cannot work for all locations as a result of geographical, technical, and regulatory impediments. As a result, data centers connecting directly onto the grid require solutions to integrate large demand loads into utility tariff frameworks.

Where the existing grid does not have sufficient generation or transmission capacity to support the interconnection of an additional data center, the data center developer will often be required to absorb resulting financial and contractual risk. The utility is often not in a position to provide the tens or hundreds of megawatts of power required to power the data center without the build-out of additional dedicated generation and transmission capacity.

In these circumstances, the data center developer may be required to source and deliver new generation projects to the utility and/or in parallel, to pay the utility to construct expensive and lengthy transmission upgrades—all at the data center developer’s cost and risk, before any tenant is signed up, and before any sleeved or dedicated PPA is entered into between the generator and the utility. To do so, the data center developer will require significant balance sheet support or third-party financing to absorb these risks, creating a number of commercial and legal issues to resolve.

Many of these arrangements between data center load and utilities are negotiated on a bilateral, one-off basis. However, as data centers proliferate, we expect that more utilities will adopt uniform tariffs to provide a more consistent approach to incorporating data center load. Ideal tariff structures balance data center priorities for reliable, affordable, and clean electricity, with the utility’s priority to mitigate the costs and risks related to building large, new energy resources to power data centers.

Green Tariffs

Many data center developers and customers have ambitious clean energy pledges that require they generate on-site carbon-free electricity, or else purchase clean energy attributes—i.e., Renewable Energy Credits (RECs)—which tie the data center’s electricity usage to a renewable energy project’s generation.

Green Tariffs allow data centers to offset their overall electricity consumption with RECs or buy bundled renewable electricity (electricity paired with RECs) from a specific project. Green Tariffs allow data centers to match their overall electricity usage with renewable energy resources, but they do not necessarily match real-time energy consumption with renewable energy generation.

24/7 Clean Utility Tariffs

Some markets are moving toward a more sophisticated clean energy procurement strategy called “24/7 carbon-free energy” or “hourly matching.” This strategy requires that buyers match their electricity usage with carbon-free electricity generation on an hourly basis, typically using resources located on the same grid where the electricity is consumed.

24/7 clean utility tariffs can facilitate hourly matching by aligning hourly data center load with dedicated, carbon-free energy generation. Unlike traditional green tariffs, 24/7 clean utility tariffs match actual consumption with real-time clean supply, improving both sustainability and reliability. Crucially, these tariffs are designed to reflect the true cost of service without imposing rate distortions that deter regional investment. However, from the utility perspective, it can be a challenge to contract for sufficient renewable generation to deliver 24/7 clean energy to customers, and so the costs of these tariff products can be significantly higher than other tariff offerings.

Clean Transition Tariffs

Clean Transition Tariffs (CTTs), which go by other various names such as Accelerating Clean Energy tariffs, go one step further by offering a structured, transparent financial mechanism for delivering new, dedicated clean resources in direct partnership with data center operators and energy developers.

CTTs are gaining traction in non-Independent Service Operator (ISO) utility regions where utilities function as monopolies, and do not permit data centers to purchase electricity from outside suppliers. CTTs offer a breakthrough path to develop dedicated clean resources without shifting costs to nonparticipating customer classes or triggering prolonged regulatory disputes. This approach bypasses many of the bottlenecks in traditional regulatory pathways by:

  • Isolating cost responsibility to beneficiaries, avoiding cost shifts to other customer classes
  • Reducing regulatory friction by clarifying financial commitments up front
  • Securing long-term certainty through binding agreements and defined risk mitigation mechanisms
  • Well-crafted CTTs emphasize collaboration between stakeholders to ensure that the cost of service aligns with actual demand, to create predictable cost structures, and to establish financial mechanisms to allow for the expedited development of dedicated clean energy resources.

In conjunction with CTTs, innovative contractual and operational mechanisms can expedite data center deployment. These mechanisms may include:

  • Financing structures that share initial investment costs and guarantees between utilities and large customers for new generators developed to serve data centers within the utility’s territory
  • Encouraging data center operators to program behind-the-meter generation and storage assets, where available, to be dispatchable to interconnecting utilities under limited and defined periods, thereby contributing positively to the grid during certain system events
  • Creating incentives or requirements for demand-side management programs at data centers, such as selective load shedding or prioritized load management during defined system events (e.g., prioritizing training modules over latency-sensitive applications at a given data center)
  • Establishing a low-cost standby backup service for data centers primarily served by behind-the-meter generation

NV ENERGY'S CLEAN TRANSITION TARIFF (CTT)

NV Energy’s new CTT illustrates this model in practice:

  1. Available to customers with ≥ 5 MW average hourly load
  2. Rates based on the utility’s Base Tariff General Rate, with credits for generation because power is supplied by a dedicated resource
  3. Long-term agreements matching the asset’s lifespan
  4. Risk protections via liquidated damages and security requirements

This type of structure helps provide developers with bankable certainty, utilities with cost containment and data center customers with reliable, clean energy tailored to their needs.

“Take-or-Pay” Tariff Structures

Unlike the versatile tariff structures discussed above, “Take-or-Pay” tariff structures prioritize mitigating utility risk above all else. This structure requires that data centers pay all or a percentage of their contracted capacity regardless of actual energy use. This structure ensures data centers cover the cost of building new generation resources; however, significant fixed costs deter investment and complicate development.

Behind-The-Meter (BTM) PPAs

“Behind-the-meter” refers to co-located power plants that deliver electricity directly to an energy load without using regulated transmission lines as an intermediary. BTM powering of data centers is an attractive option for corporates and data center developers lacking consistent, affordable, or readily available grid energy. These transactions are typically negotiated in a PPA similar to the thousands of megawatts of PPAs already in existence for rooftop and commercial & industrial solar projects. But BTM data centers that do not have any utility grid connection, even for backup power, are rare, because data center developers must over-build generation to ensure that they have reliable supply at all times, resulting in significant incremental development costs.

BTM PPAs offer data centers many advantages:

  • On-site electricity generation can help ensure reliability by providing a stable source of power disconnected from the wider grid.
  • BTM PPAs may enable data centers to directly source renewable energy, helping them meet sustainability targets and reduce their carbon footprint.
  • BTM generation can be customized to satisfy the specific needs and operational requirements of a data center.

However, certain unique considerations exist:

  1. Because data center loads are significant, the project may require a large percentage of the buyer’s land; site arrangements and allocation of site responsibilities are key, and the contractual interface must be evaluated.
  2. Parties will need to consider the most efficient and desired outcome of the project following a default of either party or termination of the PPA and, in particular, the future power sales arrangement for the generator if the data center shuts down, no direct grid interconnection is available, and the generator becomes “stranded” without a load to serve.
  3. Storage can be co-located with generation to shift energy supply when the renewables facility is generating excess output or grid energy is not sufficient to satisfy data center demand.

Build-Transfer Arrangements

A data center developer may facilitate sourcing of new generation through a build-transfer agreement (BTA).

A BTA is a hybrid acquisition and construction contract, in which the BTA counterparty, typically a project developer, secures and initially owns the land rights, permits, interconnection rights, and all other assets necessary to construct and operate the generation facilities. They’re also responsible for constructing (either directly or via a third party) the facilities. This activity all occurs for a fixed price. Afterwards, on the “closing date” under the BTA, the data center developer, as “buyer,” takes ownership of the project assets, and the seller thereafter remains responsible for achieving final completion of the facility.

For its part, the data center developer is responsible for paying the purchase price under the BTA, typically in installments which may be structured such that the buyer is essentially providing project financing for the late-stage development and construction of the facility. If the seller or its parent is financing the development and construction, installments may include a relatively modest pre-closing closing deposit, a closing payment, and one or more post-closing installments conditioned on achievement of substantial and/or final completion payment, depending on the closing conditions.

BTA Advantages

  • The data center developer can shorten the time required to secure generation capacity by entering into a BTA with respect to a generation facility in an advanced stage of development.
  • A BTA provides an opportunity (within limits) to customize the generation facility to better address the data center’s specific requirements, including reliability, ramping capabilities, regulatory compliance, and environmental impact, among others.
  • The BTA structure is versatile and may be used to purchase a behind-the-meter project, a network load project, a project utilizing replacement interconnection rights, or a project interconnected to the same utility network and utilizing CTTs (where available).
  • Through a BTA structure, the data center developer would eventually own the generation facility’s tax credits (investment or production) and could monetize them via self-use, a third-party tax equity arrangement, or a third-party sale.

BTA Disadvantages

  • BTAs require a significant commitment of capital and a willingness on the part of the data center developer to take on some development and construction risk, which may be considerable depending on the BTA terms.
  • BTAs are complex and bespoke, requiring the parties to identify and allocate various risks including development, equipment procurement, construction, financing, regulatory and tax risks, force majeure and change-in-law events, and the parties’ rights and obligations if the BTA is terminated before closing occurs. The complexity can be compounded by introducing a joint venture structure, which may be needed to facilitate third-party tax equity monetization or to enable two or more parties to utilize the generation facility’s capacity.
  • The data center developer will own and operate (or engage a third party to operate) the project, which activities have their own costs and risks.
  • Depending on the nature of the project and the applicable regulatory environment, the project could become an expensive stranded asset if the data center fails or the revenue-generating contracts are breached or terminated prematurely.

Virtual PPAs

Many data center companies have pledged 100% renewable energy use for their operations and have innovated the structures to obtain this goal.

One such innovation is the Virtual Power Purchase Agreement (VPPA), which offers a flexible and impactful way to achieve these objectives by allowing operators to make renewable energy claims by financially supporting renewable energy projects without requiring physical delivery of electricity. Through VPPAs, data centers procure Renewable Energy Certificates (RECs), but unlike a REC-only contract, the VPPA allows the purchaser to facilitate the development of new clean energy resources (sometimes called “additionality”) and potentially benefit from the financial settlement of energy price differences.

The adoption of VPPAs presents several advantages for data center operators:

  • Provides price stability by locking in a fixed rate for renewable energy over the contract term, typically 10 to 20 years.
  • Enhances the operator’s reputation by demonstrating a commitment to sustainability and supporting the transition to a low-carbon economy. This can be particularly beneficial in regions where community and regulatory pressures demand greater environmental responsibility from data centers.
  • Can be tailored to meet specific corporate goals, such as sourcing energy from particular types of renewable projects or geographic locations, allowing data centers to strategically align their energy procurement with broader business objectives.
  • Because VPPAs are financial transactions, they allow data centers to meet their renewable energy goals in regions where resource availability or regulatory or cost hurdles make renewable procurement difficult.

As with all power contracts, Virtual PPAs have some drawbacks:

  • VPPAs typically involve long-term purchase commitments.
  • Since VPPAs do not involve the physical delivery of electricity, renewable energy generated from the contract may not directly reduce the carbon footprint of the buyer’s operations.
  • Because VPPAs often involve setting a fixed price for RECs, the buyer is exposed to market price fluctuations and may end up paying more than the market rate.

In 2024, S&P Global estimated that data centers procured over 17 GW of clean energy through direct third-party power purchase agreements. The VPPA trend will surely continue as one of the most effective tools to navigate the complexities of renewable energy procurement while contributing to the global effort to combat climate change.

REC Transactions

Because of the physical delivery challenges associated with carbon-free or renewable energy and, in turn, the challenges to satisfy data center load, data center providers may choose to purchase RECs via one or more long-term agreements from a renewable energy resource. The owner of such resource can then match the RECs to the data center’s load on a 24/7 basis. It should be noted that it is difficult to achieve actual 24/7 matching from renewable energy alone; however, certain protections can be built into the agreements to ensure that the matching is done at the highest possible rate.

Procuring RECs from multiple renewable resources in the vicinity of a data center and arranging for 24/7 matching is an efficient way to make unique green claims without the hindrance of physical delivery and local market constraints (i.e., not enough carbon-free sources to delivery on a 24/7 basis).

Hedging Transactions

Data centers can benefit from active energy hedging strategies, particularly in ISO-markets. Energy hedging is a financial strategy that allows companies to lock in energy costs using tools like futures, swaps, or options. These strategies enable data centers to ensure 24/7 power available to the data center facility, mitigate pricing risk caused by spikes in power pricing in given intervals (e.g., daytime summer intervals), and much like VPPAs, match utility-delivered power supply with available clean power from renewable energy resources.

Unlike bilateral PPAs (whether physical or virtual) or utility tariffs discussed elsewhere in this paper, hedges are defined by their flexibility. Specifically:

  • Tenor: Can be structured on a short-term basis (including on a day-ahead basis) or for longer terms of 7 to 10 years (although hedges rarely approach the true long-term nature of PPAs).
  • Form: Entered into pursuant to bilateral agreements, through commodities exchanges, or through ISOs. They can be contracted through industry-standard instruments such as the ISDA, the EEI, the WSPP, or through bespoke instruments.
  • Contracting Parties: Providers can include not only utilities, but also independent power producers, energy storage companies, commodity trading desks, and insurance providers.
  • Products: May include energy, capacity attributes, ancillary services, and environmental attributes/ renewable energy credits.
  • Terms: Can include fixed-for-floating swaps (e.g., TB2/4 hedges for BESS projects), call options (e.g., heat rate call options for thermal generation) and put options (whether based on price or on revenue).

While allowing more flexibility than traditional instruments, hedges can come with their own set of unique risks. These may include:

  • Creditworthiness and Liquidity: Unlike public utility-approved instruments that may be included in a utility’s rate base, hedges are ultimately backed by both parties’ balance sheets combined with other forms of credit support (e.g., letters of credit). General problems with a hedge provider’s balance sheet, including as a result of issues with other projects, may affect any given hedge in its portfolio (including one with a data center provider).
  • Market Events: Market- or weather-related events that affect physical assets may have knock-on effects under hedges. The most poignant example was during Winter Storm Uri when a large number of fixed-volume hedges in the Electric Reliability Council of Texas (ERCOT) market ultimately failed.

The sophistication of a hedging strategy is really in the mind of the conceiver. Risks to the strategy are creditworthiness, liquidity, and increased financial exposure.

^ top

Chapter 5.

Regulatory Considerations

In this chapter: Federal and state regulatory requirements including FERC’s network load classification for co-located projects, state-level limitations on direct energy supply arrangements, exemptions under the Public Utility Holding Company Act for exempt wholesale generators, and requirements for market-based rate authority from FERC for wholesale power sales.

Network Load Classification for Co-Located Projects

A growing regulatory issue concerns whether data centers with behind-the-meter generation must be designated as “network loads,” requiring them to pay for transmission upgrades even if they rarely draw power from the grid.

In a 2024 decision, the Federal Energy Regulatory Commission (FERC) rejected amendments to an interconnection agreement between PJM and Susquehanna Nuclear, LLC, aimed at supplying an Amazon data center behind-the-meter. Protesters argued that the data center should be classified as a network load and share in transmission upgrade costs. FERC agreed, citing unresolved reliability concerns.

FERC has since opened a “show cause” proceeding to evaluate PJM’s treatment of large co-located loads. FERC’s decision could have major cost and timing implications for developers seeking to co-locate generation with data centers.

As transmission bottlenecks and policy reforms reshape the grid interconnection landscape, data center developers must adopt informed, flexible strategies. Whether pursuing direct interconnection or co-locating with generation, success will depend on navigating evolving FERC rules, interconnection queue mechanics, and the emerging regulatory treatment of hybrid generation-load facilities. Choosing the right path requires a careful balance between cost, control, risk, and speed to market—with increasing need for creative approaches with respect to power supply and utility partnerships.

Limitations on Supply Arrangements

Because electric utilities maintain regulated monopolies in most states, state law determines how easy or difficult it is for data centers to contract dedicated energy resources. Currently, only 13 states have comprehensive retail electric choice programs, allowing generators to sell or transfer electric energy directly to co-located loads, including data centers. Another eight states have limited programs for such arrangements.

If a state does not have a retail choice program and the data center owner is not itself the owner of the on-site generation, the generation owner must “sleeve” sales of electric energy through an intermediary. Typically, these sleeved transactions are accomplished through back-to-back PPAs where, under one agreement, the generation owner makes wholesale sales of electric energy to the intermediary (e.g., the interconnecting utility) and, under the other agreement, the intermediary makes retail sales of electric energy to the data center load.

Exemptions under the Public Utility Holding Company Act

Sleeved transactions are common even in states with robust retail choice programs because they preserve exemptions from federal regulation arising under the Public Utility Holding Company Act (PUHCA). Originally enacted in 1935 and partially repealed in 2005, PUHCA establishes onerous accounting, recordkeeping, and reporting requirements for holding companies of entities that own in-service generation and transmission facilities. These requirements were designed to ensure that holding companies were not subsidizing their utility-related businesses with nonutility-related revenues and vice versa.

There are multiple options for pursuing exemptions and waivers from the PUHCA requirements, but the most common and favored exemption by developers and financing parties alike is to have the generation-owning project company self-certify with FERC as an exempt wholesale generator (EWG). To do this, the project company must affirm that it is engaged exclusively in the business of owning or operating, or both owning and operating, facilities used to sell electric energy at wholesale. Although EWGs can engage in certain incidental activities, they cannot engage in any retail sales or transfers of electric energy unrelated to wholesale electricity sales. An exception to this rule would be the provision of station power to other EWGs that might be located behind the same point of interconnection.

Using a sleeved transaction accomplishes two goals: it facilitates service to co-located loads even in states that do not have retail choice programs, and it preserves EWG status.

In states where it is possible for a generator-owner to sell or transfer electric energy at retail behind the meter to co-located data centers, a generator-owner can seek alternative exemptions from PUHCA but still might be restricted by their financing arrangements. Financing parties typically require that generator-owners secure EWG status to ensure that they will not become subject to the accounting, recordkeeping, and reporting requirements under PUHCA, particularly because the financing party may need to foreclose on their interests on the underlying generating facility. Accordingly, even in states with retail choice, many independent power producers use sleeve transactions to maintain EWG status in accordance with their financing obligations.

Need for Market-Based Rate (MBR) Authority

Any entity that makes wholesale sales of electric energy within the contiguous United States outside of the ERCOT region of Texas must file the rates, terms, and conditions of such sales with FERC. To make wholesale sales of energy at negotiated or market-based rates, including pursuant to a PPA, a seller must obtain MBR authority from FERC. For example, independent power producers that enter into sleeved transactions or that wish to make wholesale sales of any excess generation produced by their projects must obtain MBR authority before making any such sales, including test sales from their projects.

Data center owners do not require MBR authority to purchase energy, but they could require it to the extent they intend to resell any energy purchased under their supply arrangement or if they wish to enter into power marketing activities in wholesale markets.

Obtaining MBR authority from FERC is a relatively simple process for most wholesale sellers. A seller must demonstrate to FERC that it and its affiliates lack, or have adequately mitigated, both horizontal (generation) and vertical (transmission) market power in the relevant wholesale market. In addition, a seller must file with FERC a standard MBR tariff that provides terms, conditions, and any applicable limitations relating to its sales of wholesale energy, capacity, and ancillary services. Once a seller obtains MBR authority, it must file with FERC quarterly reports with summary information about its wholesale sales. Additional filings would be required to reflect increases in generating capacity controlled by the seller and its affiliates. As “public utilities” under the Federal Power Act, entities with MBR authority also must obtain prior FERC authorization for financing arrangements and changes in direct or indirect ownership of the seller.

^ top


Download the full report

Learn more about our Data Centers practice