Virginia holds roughly a third of the world's hyperscale data centers. They already draw a quarter of the state's electricity — and the Clean Economy Act comes due in 2045.
Executive Summary
This report presents a policy overview aimed at safeguarding Virginia's clean energy goals in the face of growing electricity demand from data centers. Under the Virginia Clean Economy Act (VCEA), the Commonwealth has committed to 100% renewable energy generation by 2045 for Dominion Energy and 2050 for Appalachian Power (Virginia Energy, n.d.). However, the rapid expansion of data centers — currently consuming 25% of the state's electricity and projected to drive a 300% increase in demand by 2045 — poses a significant threat to grid reliability and the achievement of these targets (Boehmer, 2024; Halper, 2024). Without intervention, Virginia could face a 20 GW energy supply gap by 2035, undermining both energy stability and environmental objectives (Appendix A). After evaluating four options based on their potential to reduce grid strain, political feasibility, and cost favorability, this report recommends the implementation of a PUE-Based Tax Credits. This approach would offer a 30% credit to qualifying data centers and, if adopted by just half of Virginia's centers, could reduce energy demand by between 40 and 70 GW by 2050. Such a reduction would put Virginia well on its way to restoring energy feasibility and closing the projected supply gap. Taking proactive steps now is essential to ensure the Commonwealth can meet its clean energy targets without compromising economic growth and resilience.
Overview and Problem Statement
Virginia is at a pivotal crossroads in its energy transition. With the passage of the Virginia Clean Economy Act (VCEA), the state committed to achieving 100% renewable energy generation by 2045 for Dominion Energy and by 2050 for Appalachian Power (Virginia Energy, n.d.). This legislation positions Virginia as a leader in clean energy policy in the Southeast. However, the rapid expansion of data centers threatens to derail this progress. Home to over 135 hyperscale data centers, approximately 35% of the global total, Virginia has become the backbone of the global internet (Virginia Economic Development Partnership, n.d.). This status is driven by a robust fiber infrastructure, favorable tax policies, affordable electricity, and competitive construction costs. Yet these same advantages have created a paradox: the data center sector now consumes 25% of Virginia's electricity (Boehmer, 2024), and demand is expected to increase by 300% by 2045 (Halper, 2024). Based on Virginia's current power supply, compliance with the VCEA is projected to result in an energy supply gap of 9.2 gigawatts (GW) by 2030, growing to 20 GW by 2035, driven by surging data center electricity demand (see Appendix A). This growth threatens the state's ability to meet its VCEA-mandated 100% renewable energy targets, risking both grid reliability and climate commitments.
While some policies have been introduced to improve energy efficiency, the Joint Legislative Audit and Review Commission (JLARC) notes that Virginia's regulatory approach has prioritized economic competitiveness over environmental rigor (JLARC, 2024). Political polarization around environmental regulation further complicates the landscape, with resistance to stricter mandates stemming from concerns about deterring technology investment. At the same time, it is increasingly clear that the market alone is unlikely to resolve the supply crisis. Dominion Energy, Virginia's largest utility, has already sought regulatory approval to bypass VCEA targets by building a new fossil fuel plant in Chesterfield, which would emit 2 million metric tons of carbon dioxide (CO2) annually (Bolster, 2023). This proposal not only undermines clean energy goals but raises environmental justice concerns, as the plant is sited in a low-income, majority Black and Brown community.
Given these compounding pressures, the Virginia Department of Energy (VDOE) has requested policy guidance on how to align continued data center growth with the state's clean energy mandate. This analysis will assess the trade-offs and feasibility of various strategies, ranging from efficiency regulations to renewable energy infrastructure investments, and culminate in a recommendation that ensures Virginia's energy system remains both resilient and decarbonized in the face of escalating demand.
Literature Review
This literature review examines the leading technological and policy solutions proposed to reduce energy demand from data centers. The literature clusters into three major categories: clean energy integration, energy conservation strategies, and global policy interventions. Clean energy integration focuses on transitioning data centers away from fossil fuels toward solar, wind, and other renewables. Energy conservation strategies emphasize efficiency improvements in design, cooling, and workload management. Finally, a growing body of work explores how policy tools, ranging from moratoriums and mandates to market-based incentives, can accelerate decarbonization.
1. Clean Energy Integration
Historically, data centers have relied heavily on fossil fuels such as coal and natural gas (Vaishnawi & Bhuvana, 2024). However, concerns about greenhouse gas (GHG) emissions and grid instability have prompted both scholars and operators to pursue renewable energy solutions, including solar, wind, and nuclear. Renewable infrastructure can be developed on-site or off-site, though off-site delivery can introduce energy transmission losses (Oró et al., 2015).
Solar energy is the most frequently studied renewable in this context. It can be deployed directly on-site (e.g., rooftop or adjacent land) or supplied via off-site photovoltaic (PV) installations. McMullen and Wemhoff (2024) found that implementing 7.2 megawatts (MW) of on-site solar, roughly 15% of a data center's energy load, reduced the center's carbon usage effectiveness (CUE), the ratio of total CO2 emissions caused by total data center energy consumption to the energy consumption of IT equipment, by 13%. Similarly, Amin et al. (2023) modeled a solar microgrid system for a data center in Bangladesh and projected that the configuration, capable of generating 249,219 kilowatt (kW) and storing 398,547 kWh, could reduce CO2 emissions by 652% compared to grid-only electricity. Over the center's lifetime, this setup could also save $21.8 million in energy costs.
While both studies confirm the environmental and economic potential of solar, they diverge in scale and context. McMullen and Wemhoff examine marginal efficiency gains in the U.S., while Amin et al. model transformative reductions in a developing economy. This contrast suggests that the feasibility and impact of clean energy integration are highly context-dependent, warranting localized feasibility assessments.
2. Energy Conservation Mechanisms
2.1 Technological and Design Efficiency
Beyond fuel switching, the literature highlights that internal design improvements can significantly reduce total energy demand. Cooling systems alone account for up to 50% of a typical data center's energy use and generate substantial waste heat (Sovacool et al., 2022). Studies estimate that up to 68% of this heat can be recovered and repurposed (Huang et al., 2020). For example, Davies et al. (2016) found that capturing waste heat from a 3.5 MW London data center and redistributing it via heat pumps could prevent 4,000 tons of CO2 emissions annually while saving over $1 million.
Suboptimal hardware layouts also raise energy consumption. Poor rack placement and airflow management can force cooling systems to work harder (Sovacool et al., 2022). Rong et al. (2016) found that proper room layout and ventilation design, costing 8–10% of initial capital expenditure, typically pays for itself in energy savings within two to three years. These findings emphasize that relatively low-cost design decisions can yield substantial reductions in data center energy demand over time.
2.2 Workload Management
To fully realize the potential of renewable energy, studies emphasize that computational efficiency must complement clean energy deployment. Two key strategies — workload rescheduling and workload migration — can help reduce peak demand and align processing with renewable availability (Huang et al., 2020).
Workload rescheduling involves shifting non-critical computing tasks to times when renewable energy is more abundant. Martin et al. (2012) showed that restricting non-critical tasks to the 9 p.m.–6 a.m. window could reduce energy storage costs by 79% in solar-powered data centers. Even outside of renewable energy supply (fix), Rong et al. (2016) found that using scheduling algorithms could cut total energy use by 10–15%.
Workload migration goes a step further, transferring processing tasks between geographically distributed data centers based on available renewable supply. Chen et al. (2012) modeled solar-powered centers in multiple time zones using 10,000 solar panels each. When using an algorithm that factored in solar availability, cooling needs, and time of day, fossil fuel use dropped by up to 40% compared to a baseline model.
Although these systems show strong potential in simulations, real-world implementation remains limited. Workload migration, in particular, faces several policy and logistical challenges, including cross-border data transfer regulations, infrastructure disparities, and data privacy concerns, that can hinder adoption. In Virginia, where the data center industry is dominated by a fragmented mix of private entities, coordination across companies presents an added barrier. Without a unified governance framework or shared infrastructure standards, it remains unclear how feasible large-scale workload migration would be in such a decentralized environment. Thus, while workload optimization techniques could significantly reduce energy demand, they are not yet viable at scale without substantial regulatory alignment, industry collaboration, and technological support.
3. Policy in Action
While technological strategies such as solar adoption and workload optimization offer significant emissions-reduction potential, these gains will remain limited without complementary policy frameworks that encourage or mandate the implementation of these strategies. The literature identifies three broad policy approaches that have emerged in response to rising data center energy demands: restrictions on expansion, energy-focused mandates, and incentive-based mechanisms.
This section reviews current examples of how these policies are being implemented in practice, with particular attention to the design, scope, and effectiveness of each approach.
3.1 Moratoriums
In response to the rapidly growing energy demand from data centers, some countries have temporarily halted data center expansion through moratoriums, which are short-term bans or suspensions on new project approvals. These policies are often introduced to give governments time to assess infrastructure constraints, such as power grid capacity, land availability, and water use, and to develop more strategic, long-term regulatory frameworks (Soares et al., 2024).
A comparative study by Soares et al. (2024) analyzes policy responses in Singapore, the Netherlands, Ireland, Germany, the United Kingdom, and Virginia, categorizing them into three broad approaches: moratoriums, long-term adjustment policies, and unrestricted expansion. Among these, three jurisdictions adopted moratoriums to pause or restrict new development. For instance, Dublin imposed a full ban on new data centers until 2028, citing grid capacity concerns and the need to reevaluate energy priorities (Traynor, 2024). Both Singapore and the Netherlands initially implemented moratoriums but later transitioned to long-term adjustment strategies, requiring stricter environmental standards for new project approvals. In Singapore, a three-year moratorium (2019–2022) halted data center approvals while the government developed a new framework focused on energy efficiency and sustainability. After lifting the ban, Singapore introduced tighter performance standards, including minimum renewable energy thresholds for new facilities (Soares et al., 2024). Similarly, Amsterdam enacted a nine-month moratorium on hyperscale data centers exceeding 10 hectares and replaced it with a "No, unless" policy — allowing new projects only if they meet stringent sustainability criteria (Soares et al., 2024).
While these moratoriums reflect a proactive approach to managing rapid data center growth, their economic implications are not fully addressed in the current literature. The study by Soares et al. (2024) does not evaluate potential trade-offs such as lost investment, job creation, or competitiveness. Nor does it provide evidence on the effectiveness of the post-moratorium adjustment policies, leaving questions about their long-term impact on emissions, grid stability, and sustainable development outcomes.
3.2 Energy-Focused Mandates
As a longer-term regulatory framework, some jurisdictions have opted for regulatory mandates focused on reducing energy use and increasing renewable integration. These include both renewable energy requirements and performance-based efficiency standards.
3.2.1 Mandate Renewable Energy
Governments are increasingly implementing policies that require data centers to integrate renewable energy and waste heat reuse into their operations. Waste heat reuse involves capturing excess heat from servers and repurposing it for heating buildings, water systems, or industrial processes, turning a byproduct of digital infrastructure into a valuable energy resource. Cities such as Amsterdam now mandate both increased use of renewable energy and the incorporation of waste heat into local energy grids (Soares et al., 2024). Singapore requires all new data centers to source at least 50% of their electricity from renewable energy (Soares et al., 2024). Germany's Energy Efficiency Act (2023) goes further, requiring 100% renewable energy sourcing for data centers by 2027, along with mandatory waste heat reuse targets that progressively increase through 2028 (Soares et al., 2024). In all three cases, these renewable energy mandates can be met through a mix of imported renewable electricity, power purchase agreements (PPAs), and investments in domestic or international renewable energy projects (Soares et al., 2024).
While these policies reflect a growing commitment to decarbonizing digital infrastructure, they also highlight important challenges. Geographic and global resource disparities mean that some countries are better positioned to produce renewable energy than others. However, allowing renewable energy mandates to be met through international offsets or imported renewables risks merely displacing the problem, placing further strain on the global renewable energy supply without expanding total capacity. If states like Virginia meet renewable mandates through external sourcing, they risk intensifying competition for limited clean energy, undermining the global decarbonization goal of policies like the VCEA. Many countries are seeing their data center electricity usage approach or exceed their national renewable energy capacity (see Appendix B), underscoring the need for localized solutions that expand renewable energy generation alongside data center deployment.
3.2.2 Energy Efficiency Mandates
Governments worldwide are increasingly implementing energy efficiency standards for data centers to mitigate surging power demand. In China, the Eastern Data and Western Computing Project sets Power Usage Effectiveness (PUE), which is a standard metric that compares the total energy consumption of a facility to the energy used specifically for computing tasks, targets at 1.25 in the east and 1.2 in the west, with stricter limits imposed in key urban hubs such as Beijing (1.4), Shanghai (1.3), and Shenzhen (1.4) (Zhang et al., 2024). Similarly, Amsterdam has introduced benchmarks not only for energy efficiency but also for land use optimization in its urban planning requirements (Soares et al., 2024). Germany's Energy Efficiency Act (2023) goes further by mandating continuous energy monitoring, reporting, and transparency measures to hold data centers publicly accountable (Soares et al., 2024).
Some national governments restrict these standards exclusively to new facilities. For instance, Singapore has introduced higher efficiency requirements specifically for new data centers (Ministry of Trade and Industry, 2022; Laforga, 2022), while Ireland's Commission for Regulation of Utilities (CRU) now requires new facilities to meet on-site power generation and efficiency criteria before gaining grid access (CRU, 2021; Government of Ireland, 2022). These policies reflect a clear trend toward proactive energy governance, but raise questions about their effectiveness and enforceability.
Despite ambitious targets, there remains little empirical evidence verifying the outcomes of these regulatory approaches. Many rely on theoretical projections rather than retrospective analyses. For example, the European Union's Ecodesign Directive, which set minimum efficiency and environmental performance standards for servers and storage devices in 2020, was originally projected to yield energy savings of 37.5 TWh per year by 2030 (European Commission, 2016; Koronen et al., 2020). Yet to date, few empirical studies have verified whether these projected savings have materialized.
Simulation studies attempt to bridge this gap, but they also rely heavily on assumptions. One study by Fernández-Cerero et al. (2020) evaluated eight proposed energy-efficiency mandates, such as adaptive artificial intelligence (AI)-based resource management and workload arrival pattern optimization, using cloud computing data from Google and Alibaba. The researchers projected that these policies could reduce CO2 emissions by 11.5 million tons annually in U.S. data centers alone, equivalent to removing 4.79 million combustion-engine vehicles from the roads. They also forecasted 15–60% reductions in energy use across regulated regions. However, such optimistic projections may not account for the full complexity of today's rapidly evolving computing landscape. As Srivathsan et al. (2024) note, power demand is now influenced by an array of variables — including chip architecture, AI model design, workload types, and even programming style — which interact in dynamic and unpredictable ways.
3.3 Incentive-Based Policies
Despite the prevalence of regulatory mandates, research increasingly suggests that voluntary agreements and incentive-based policies, such as tax credits for energy-efficient infrastructure, are often more effective in encouraging private-sector investment than rigid requirements (JLARC, 2024; Badrudeen et al., 2024). These approaches offer greater flexibility for firms and tend to face less resistance, particularly in markets with varying levels of infrastructure readiness and innovation capacity. As a result, they are often more politically feasible and practically implementable. The Netherlands promotes sustainable technology adoption through its Energy Investment Allowance Program, which provides tax exemptions for data centers that invest in energy-efficient infrastructure (Soares et al., 2024). In the United States, the Inflation Reduction Act (2022) similarly incentivized renewable energy investments, such as solar and wind power, that could be applied to data center operations (U.S. Department of Energy, 2023b). In Germany, the recently passed Energy Efficiency Act (2023) mandates that existing data centers achieve a PUE of ≤1.5 by 2027 and ≤1.3 by 2030, and that newly built facilities achieve a PUE of ≤1.2 starting in 2026 (Telyatnykov, 2023).
While these programs are promising in theory, data on take-up rates and measurable outcomes remain limited. In contrast, subsidies — another form of incentive-based policy — have a somewhat stronger empirical foundation. One particularly illustrative case is China's 2015 national green data center pilot policy, analyzed by Liu and Yang (2023). Their study provides valuable quantitative insight into the real-world outcomes of a subsidy-based, incentive-driven reform, employing a quasi-natural experiment and difference-in-differences methodology to evaluate the policy's impact on green innovation. Unlike broader voluntary programs or tax credit schemes, China's approach was a top-down, pilot-based initiative in which selected cities received direct financial subsidies and regulatory support. These incentives effectively reduced the financial risk of adopting green technologies, spurring firm-level investment in energy-efficient infrastructure. The initiative ultimately led to the establishment of three national batches of green data centers — in 2018, 2020, and 2022 — covering nearly every province in China (Liu & Yang, 2023). Throughout the program, PUE improved significantly: the national average PUE dropped from over 2.2 in 2015 to 1.64 and 1.63 in hyperscale and large data centers, respectively, by the end of 2017.
The policy yielded the greatest impact in cities characterized by high carbon emissions, limited civic development, robust network infrastructure, and strong innovation ecosystems — factors that either heightened the urgency for reform or enhanced implementation capacity. While the economic and political context in China differs substantially from that of Virginia, the policy's core success factors — targeted subsidies, infrastructure alignment, and performance-based goals — offer useful lessons. These elements may help guide the design of subsidy-based policies tailored to Virginia's unique regulatory and market landscape.
Limitations
As data centers have rapidly expanded in recent years, most policies regulating their energy efficiency remain relatively new, fragmented, or inconsistently applied, making it difficult to assess their long-term effectiveness. This represents one of the most critical gaps identified in the literature. Appendix C illustrates the policy timelines across six major data center hubs, highlighting not only the recent nature of these interventions but also frequent shifts in government approaches. Without robust, long-term evaluation, it is difficult to determine whether these regulations are achieving their intended goals, such as reducing energy demand, increasing renewable integration, or mitigating emissions, or if further adjustments are necessary. Moreover, much of the available data on policy outcomes is based on short-term projections, simulations, or case-specific modeling, rather than retrospective analysis. This lack of standardized performance tracking significantly limits the field's ability to draw generalizable conclusions, especially when comparing across jurisdictions with different regulatory capacities and energy profiles.
Conclusions from the Literature
The literature consistently affirms that while technological solutions for reducing data center energy use are well-developed and effective, the effectiveness of the policy instruments designed to drive their adoption remains unclear. Clean energy integration, cooling and layout efficiencies, and workload optimization strategies have all demonstrated measurable reductions in energy demand under controlled or pilot conditions. However, without consistent regulatory enforcement, coordinated infrastructure planning, and long-term impact evaluations, these innovations may fall short of their potential. As such, the success of the global effort to decarbonize digital infrastructure will depend not just on technological progress but on the quality, accountability, and adaptability of the policy frameworks that support it.
Analysis of Alternatives
Given the scale of Virginia's data center growth, the accelerating energy demand, and the urgent need to meet VCEA clean energy mandates, it is clear that relying on voluntary adoption and fragmented regulation is no longer a viable option. Instead, in this section, we evaluate four policy alternatives selected for their potential to reduce energy intensity, support renewable integration, and align future data center expansion with the state's climate goals.
Note: As the Virginia Department of Energy lacks direct regulatory authority over data center taxation or building standards, we recommend that the agency serve as an energy sustainability advocate and technical consultant in the exploration and potential implementation of the following policy options. Each would require action by other state agencies or governing bodies, with the Department providing guidance and expertise to inform decision-making and execution.
Evaluative Criteria
Effectiveness measures each policy's projected impact on lowering electricity consumption from data centers between 2030 and 2050, allowing a five-year lead time for infrastructure development and implementation. Ratings of low, medium, or high reflect the estimated scale of grid demand reduction in GW. This criterion is essential to ensure compliance with the VCEA and to avoid further grid strain.
Political Feasibility assesses the likelihood of policy adoption based on anticipated support or resistance from key stakeholders, including lawmakers, utilities, the data center industry, and advocacy groups. A high score indicates strong political alignment and minimal opposition. This criterion ensures proposed solutions are not just ideal, but realistically implementable.
Cost Favorability assesses the overall financial burden of each policy, including administrative costs and potential losses in state tax revenue. Policies are rated low, medium, or high based on their overall cost burden to the state (lower cost = high cost favorability). This criterion is critical to identifying solutions that advance energy goals without placing undue strain on the state's fiscal resources.
Alternative 1: Moratorium
This policy would establish a one-year moratorium on the development of new data centers and the expansion of existing facilities in Virginia, allowing time for essential technological advancements to support the state's electrical grid and the integration of renewable energy sources. We recommend that the VDE advocate to the State Corporation Commission (SCC) and relevant local governments to implement this moratorium. The goal is to halt growth-driven energy demand while regulatory and infrastructure frameworks catch up to the industry. A moratorium in Virginia would be most effectively implemented as a temporary, targeted pause on new data center permits, focused specifically on the overburdened counties of Northern Virginia, Loudoun and Prince William counties.
Effectiveness Score: Low (see Appendix D for calculations)
Data center energy demand in Northern Virginia is estimated to increase from 150,000 gigawatt-hours (GWh) in 2025 to 168,000 GWh in 2026. A one-year moratorium on new data center development could prevent this increase, resulting in the avoidance of 18,000 GWh of electricity consumption. However, despite substantial short-term savings, this one-year scope limits long-term grid impact.
Political Feasibility Score: Low
This policy is likely to face political pushback, especially in pro-business or tech-friendly areas of the state. Based on our conversation with Elizabeth Andrews, an environmental attorney and natural resources policy expert, we do not anticipate a moratorium being a feasible solution on the state level, however, counties may choose to enact county-level moratoriums (E. Andrews, personal communication, April 15, 2025). Because the VDE does not have the authority to impose a moratorium, the agency would need to work in close coordination with the SCC and encourage county-level implementation.
Cost Favorability Score: High (see Appendix E for calculations)
While administrative costs are relatively low, the opportunity cost of halting new growth of data centers can be substantial, especially if the policy deters long-term investment, tax revenue growth, and job creation for Virginia. Virginia's total revenue from data centers in FY 2021 was $1.64 billion. Assuming that the revenue grows proportionally to the annual growth rate between 2022 and 2023 of 16%, we estimate this policy would create $262.4 million in foregone tax revenue.
Alternative 2: Tax Incentives for Onsite Renewable Energy Investment
This policy alternative proposes a 30% tax credit for data centers in Virginia that invest in onsite renewable energy systems, such as solar or wind installations. We recommend that the Virginia Department of Energy work in partnership with the Virginia Department of Taxation to develop and advocate for the implementation of this incentive. This incentive is modeled after the federal Investment Tax Credit (ITC) and would be available from late 2025 through 2030 to encourage clean energy adoption at the facility level. By incentivizing these investments, Virginia can reduce the strain on its electrical grid while supporting sustainable growth in its data center industry.
Effectiveness Score: Low (see Appendix F for calculations)
No current information on take-up rates, so we have devised three scenarios of effectiveness based on varying take-up rates: 25% maximum take-up, 50% maximum take-up, and 75% maximum take-up. Under a medium take-up scenario, we estimate that 50% of Virginia's 150 data centers would take advantage of the tax incentive and install a 1 MW renewable energy system, generating 1.5 GWh annually. This would result in 75 MW of installed capacity, offsetting approximately 112.7 GWh of energy per year, and totaling 2,250 GWh over the 20-year clean energy transition from 2030 to 2050.
Political Feasibility Score: High
The political feasibility of tax incentives for onsite renewable energy adoption would likely be high. Virginia already has existing tax incentives for data centers, demonstrating the state's willingness and ability to successfully enact and promote these incentives. Creating and tailoring tax incentives to encourage onsite renewable energy investment would allow Virginia to balance the continued growth and presence of data centers while working towards compliance with the VCEA. The elective nature of tax incentives also bolsters their feasibility and bipartisan support. Politically favorable, optional nature is attractive to both lawmakers and data center developers.
Cost Favorability Score: Medium (see Appendix G for calculations)
To estimate the fiscal impact of offering a 30% tax credit to incentivize onsite renewable energy investments at data centers, a scenario was modeled in which 50% of Virginia's 150 data centers each install a 1 MW solar system between 2025 and 2030. Based on an average installation cost of $1.3 million per system, the total capital investment would amount to $97.5 million. Applying a one-time 30% tax credit yields a total estimated state cost of $29.25 million over five years (2025–2030). Importantly, the program could be administered through the existing infrastructure of the Virginia Department of Taxation, minimizing additional administrative costs or staffing needs. While less expensive than other alternatives, cost favorability is rated Medium due to limited energy impact per dollar spent.
Alternative 3: PUE-Based Tax Credits for Data Center Energy Efficiency
This policy proposes offering one-time 30% tax credits to data centers that invest in energy efficiency measures to reduce their Power Usage Effectiveness (PUE) to 1.2 or lower. The proposed threshold is designed to mirror performance targets in comparable international policy. We recommend that the Virginia Department of Energy work in partnership with the Virginia Department of Taxation to develop and advocate for the implementation of this incentive. The tax credit would be a one-time tax credit for a certain percentage of the energy efficiency improvement cost and would scale with the level of efficiency achieved, incentivizing facilities to reduce operational energy consumption while allowing continued growth in digital infrastructure.
Effectiveness Score: High (see Appendix H for calculations)
According to Ruszkowski (2024), if the average data center PUE is reduced from 1.5 to 1.2, this represents a potential energy savings of 14.28% to 25%. No current information on take-up rates is available, so we have devised three scenarios of effectiveness based on varying participation levels: 25%, 50%, and 75% maximum take-up. Under the mid-range scenario of 50% take-up, implemented starting in 2026, projected energy savings from 2030 to 2050 range from 350,339.8 GWh to 613,340 GWh. This corresponds to an average power demand reduction of approximately 40 to 70 gigawatts (GW), representing a substantial contribution toward closing Virginia's projected energy gap.
Political Feasibility Score: High
Tax credits for PUE improvements are likely to be politically feasible due to their alignment with both environmental goals and economic growth. They promote energy efficiency without imposing mandates, making them market-friendly and pro-business, which could garner bipartisan support. Given the increasing focus on sustainability and energy savings, especially among tech companies, this policy would likely resonate across party lines. However, some resistance may come from smaller data centers due to potential upfront costs, and there could be concerns over short-term fiscal impacts from reduced tax revenue. Overall, the long-term benefits make the policy an attractive, politically viable option.
Cost Favorability Score: Low (see Appendix I for calculations)
This analysis also considers the fiscal impact of a one-time 30% tax credit aimed at encouraging data centers to invest in energy efficiency upgrades that reduce Power Usage Effectiveness (PUE) to 1.2. Modeled after the federal Investment Tax Credit, the proposed policy assumes a 50% participation rate among Virginia's data centers. Given a total estimated statewide upgrade cost of $25 billion, the credit would apply to $12.5 billion in investments, resulting in a projected state cost of $3.75 billion. As with the renewable energy tax credit, administration of the program could be streamlined through the Virginia Department of Taxation's existing infrastructure, ensuring efficient implementation without the need for additional staffing.
Alternative 4: Require Energy Efficiency Technology Adoption in Newly Built Data Centers
This policy mandates that newly constructed data centers in Virginia adopt high-efficiency technologies, aiming for a Power Usage Effectiveness (PUE) of 1.2 or lower. The policy applies to the 32 large data centers expected to be operational by 2030, as well as any additional data centers built thereafter. The goal is to significantly reduce energy consumption in these facilities, easing the strain on Virginia's electrical grid and contributing to the state's clean energy targets.
Effectiveness Score: Medium (see Appendix J for calculations)
Requiring newly constructed data centers in Virginia to adopt high-efficiency technologies with a Power Usage Effectiveness (PUE) of 1.2 can significantly reduce strain on the electrical grid. With 32 large data centers expected online by 2030, each operating at 100 MW continuously, the policy could increase energy efficiency by between 14.28% to 25% compared to a baseline PUE of 1.5. This translates to energy demand reductions of 80,059 to 140,160 GWh over a 20-year period.
Political Feasibility Score: Medium
The political feasibility of requiring energy efficiency technology adoption in newly built data centers would be medium. Various factors and constraints can both increase and decrease feasibility. These new regulations could be framed as working towards the state's clean energy goals. Additionally, major developers of data centers such as AWS and Google have committed to energy efficiency standards; these mandated requirements could be posited as in alignment with their goals (Amazon, n.d.; Google, n.d.). However, some opposing factors, such as potential pushback from data center developers and lawmakers, citing regulatory overreach and potential slowed development, which could affect revenue and growth (E. Andrews, personal communication, April 15, 2025).
Cost Favorability Score: Low (see Appendix K for calculations)
The implementation of this alternative would require Virginia to cover oversight costs associated with these new regulations. Costs include the addition of compliance staff to monitor the new requirements to ensure regulatory compliance of energy efficiency technologies for newly built data centers. According to Ruszkowski (2024), the Virginia Economic Development Partnership estimates needing two additional full-time employees to monitor compliance with the new requirements (Virginia Department of Taxation, 2024). Two full-time employees would cost $230,000 per year (Virginia Department of Taxation, 2024). Over the 25-year period, this would cost the Virginia state government $5.75 million.
Recommendation
Based on our evaluation of the criteria, we recommend that the Virginia Department of Energy advocate for Alternative 3 — the PUE-based tax credit — as a policy solution. This voluntary, market-driven approach offers a one-time 30% tax credit to data centers that reduce their PUE to 1.2 or lower, promoting measurable and credible energy efficiency gains over the current industry average of 1.5. If adopted by just half of Virginia's data centers, the policy could deliver statewide energy savings equivalent to 40–70 gigawatts (GW) of average power demand through 2050. Its opt-in design avoids regional inequities and political friction while encouraging meaningful action. The projected $3.75 billion public cost is modest relative to long-term benefits and can be efficiently administered using existing infrastructure.
Crucially, Virginia faces a growing energy gap — estimated at 20 gigawatts by 2035 and expected to increase beyond that — and Alternative 3 is the only intervention with the potential to close it. While the proposed 30% credit mirrors solar tax incentive precedents, there is no established benchmark for PUE-based policies. Given this lack of precedent, the credit rate is flexible. A lower rate, such as 10%, could still achieve meaningful energy efficiency improvements while reducing program costs, should cost concerns arise during the agency's review or the legislative process.
Implementation Strategy
We recommend that the Virginia Department of Energy take on an advocacy role before the Virginia Senate Committee on Energy and Natural Resources and support legislative efforts to authorize the PUE-based tax credit policy. We also suggest that the Department engage the Clean Energy Advisory Board to build additional support and momentum. Once enacted, the credit would be administered by the Virginia Department of Taxation, which would manage application intake and disbursement. Data centers applying for the credit would be required to submit documentation verifying compliance with the 1.2 PUE threshold. Upon verification, qualifying facilities would receive a one-time tax credit. Ongoing monitoring and evaluation would be necessary to assess the policy's effectiveness, measure energy savings, and inform future policy refinements.
Conclusion
Virginia's ability to achieve its clean energy commitments hinges on whether it can meaningfully address the accelerating energy demands of its data center sector. The PUE-based tax credit proposed in this report offers an ambitious yet pragmatic, cost-effective pathway to reduce data center energy consumption without stifling economic development. Without swift and coordinated action, the window to course-correct is rapidly closing. The Virginia Department of Energy must step forward and act as a driving force, using the full extent of its influence and authority to offer strategic advocacy, technical guidance, and interagency collaboration to push this policy forward. With decisive action, Virginia can preserve the integrity of its clean energy goals and also set a precedent for balancing technological growth with climate responsibility.