States Are Opening the Door to Data Center Investment in Virtual Power Plants

In November 2025, RMI published a report making the case that off-site demand flexibility, energy efficiency, and distributed generation (together, “distributed capacity”) can help meet rapidly growing data center electricity demand while lowering costs, improving reliability, and reducing the need for new fossil generation. The report, titled How Virtual Power Plants Can Help the United States Win the AI Race, outlined how data centers and other large electricity customers could help finance and scale distributed capacity.

Since the report’s publication, new laws, commission decisions, and utility proposals have begun creating pathways for large customers to fund distributed capacity, for utilities to evaluate it as a grid resource, and for that capacity to deliver tangible benefits to participating customers. The design of these pathways will shape the types of transactions that are possible, the willingness and ability of data centers to pay for distributed capacity, and the potential scale of future investments. This article reviews several of the most significant developments since our initial report and considers what they reveal about the emerging role of distributed capacity in serving load growth.

How distributed capacity provides value across the grid

Our 2025 report outlined new commercial models that could enable the growth of distributed capacity, delivering benefits to both large electricity customers and the overall grid. Although the allocation of responsibilities varies across the models, each relies on three basic elements:

• Large customers sponsor and provide capital for expanded distributed capacity.
• Utilities or market operators establish mechanisms to compensate large load customers for the benefits that distributed capacity provides to the grid.
• Virtual power plant (VPP) aggregators, with varying degrees of utility involvement, develop and deliver the required capacity (see our VPP 101 article for more information on VPPs).

Distributed capacity can provide multiple values to the grid, each of which could be matched to corresponding compensation mechanisms for large load customers. Exhibit 1 shows four examples of how compensation mechanisms can connect value from distributed capacity to data center loads, supporting potential transactions. These mechanisms are not exhaustive; cost savings, for example, could be realized in some markets through a contract-for-differences structure rather than via large load tariff requirements, and the set of mechanisms available is jurisdiction-specific.

Much of the current discussion has focused on speed to power — reducing the waiting time for a customer to connect the grid — but recent policies and commercial arrangements are also beginning to recognize other benefits, including cost reduction and hedging, firmer service, and community benefits.

Exhibit 1: Four example transaction mechanisms between distributed capacity and large load customers

Source: RMI analysis

States are beginning to formalize these value transfer mechanisms. The examples below show how new policies and agreements are putting the models for distributed capacity expansion into real-world use.

Large load tariffs advance pass-through funding for VPPs

Model 1: Pass-Through Funding for Utility-Managed VPP

Under Model 1, a large customer provides funding for distributed capacity that a utility or aggregator operates for the benefit of the broader system. The funding could support a specific distributed capacity project or flow into an existing utility program, allowing the large customer to supply the capital while the utility retains responsibility for procuring and managing the resource.

Google and Xcel Energy are advancing a clear example of this model in Minnesota. In February 2026, the companies announced an agreement to serve Google’s proposed Pine Island data center with a portfolio that includes 1,400 MW of wind, 200 MW of solar, and 300 MW of long-duration energy storage. Google would also contribute $50 million to Xcel’s Capacity*Connect program, which authorizes Xcel to deploy up to 200 MW of distributed batteries as a utility-owned VPP. Google’s electric service agreement remains pending before the Minnesota Public Utilities Commission, and as of now there is no direct tie between the distributed capacity investment and accelerated interconnection or other nontraditional grid benefits. However, if the proposal is approved, it will demonstrate how a large customer can help capitalize distributed resources that benefit the wider system.

Other states are establishing broader procurement pathways that could eventually support similar arrangements with provisions in large load tariffs. In November 2025, the Kansas Corporation Commission approved Evergy’s Large Load Power Service tariff for customers adding at least 75 MW. The settlement includes a Clean Energy Choice Rider through which participating customers can support clean resources — including distributed flexibility — in place of or in addition to those in Evergy’s preferred resource plan. Payments for such projects must be negotiated with Evergy and approved by the Commission. The eligibility of distributed flexibility under large load tariff requirements is often not automatic and is a key design decision to monitor.

Oregon could offer a similar approach. In May 2026, the Oregon Public Utility Commission approved Portland General Electric’s large load tariff. While primarily focused on cost allocation, the order also permits PGE to negotiate special contracts that would allow large load customers to fund new clean energy resources, potentially including VPPs, in exchange for more efficient interconnection timelines.

Nevada’s Clean Transition Tariff presents a third potential pathway to scale. In May 2025, the Public Utilities Commission of Nevada approved the tariff’s first agreement, allowing Google to fund 115 MW of enhanced geothermal power for NV Energy’s system. In NV Energy’s 2024 integrated resource plan (IRP) proceeding, Google had proposed extending a similar customer-funded model to off-site demand-side resources, including residential HVAC upgrades, energy efficiency, demand response, and behind-the-meter batteries, in exchange for credit for the incremental capacity those investments provide. The Commission did not approve the proposal but encouraged Google, NV Energy, and other stakeholders to refine it for consideration in a future IRP or amendment — signaling a potential path to apply this structure to distributed capacity as well as utility-scale generation.

A PJM deal and new state laws advance VPP capacity transfer

Model 2: VPP Capacity Transfer

Under Model 2, a large customer contracts with a third-party aggregator to develop VPP capacity, which the utility or grid operator then recognizes as an offset to some portion of the customer’s capacity needs. Unlike Model 1, the grid operator does not procure or operate the distributed capacity. Instead, the large customer finances the resource directly, while the aggregator recruits participating customers, manages the portfolio, and demonstrates its performance.

In June 2026, Google and Voltus announced the first major commercial agreement built around this model. Under the three-year agreement, Google will fund a Voltus-operated VPP comprising up to 100 MW of accredited distributed capacity in PJM, the regional transmission organization (RTO) that covers all or part of 13 US states and Washington, D.C. This deal will support system-wide reliability while offering Google a hedge against rapidly escalating PJM capacity market costs.

States are also taking action. In 2026 legislation, Virginia directed utilities to propose voluntary demand-flexibility programs for large load customers. Those customers may meet a demand-flexibility standard by reducing their own consumption or by funding, supporting, or purchasing verified “capacity reduction credits” from other customers. The credits could be created through energy efficiency, heat-pump conversions, demand response, customer-sited storage, VPPs, and other measures that produce measurable reductions in peak demand. Utilities must file program proposals by January 15, 2027, and the State Corporation Commission must issue final orders by November 30, 2027.

New Jersey has gone further by establishing an explicit demand-reduction trading mechanism. The Data Center Fair Share law, signed in July 2026, requires the development of a new market through which data center customers may contract directly with third parties to offset their capacity obligations by investing in reductions in demand and increased flexibility elsewhere on the system. The New Jersey Board of Public Utilities must now develop the detailed standards that will govern accreditation, contracting, performance, and interaction with PJM.

State and RTO policies begin linking flexibility to faster interconnection 

Model 3: VPPs as Reliability Reinforcement

Under Model 3, a large customer signs an agreement for conditional power service. Flexibility — on-site, or off-site via a VPP — is used by the customer to “firm up” its service.  

This model could become especially relevant in regions contemplating policies that link customer resource procurement to firm service. One example comes from the Southwest Power Pool (SPP), which, in June 2026, became the first RTO to have a “connect-and-manage” approach for large loads approved by FERC. Its Conditional High Impact Large Load Service (CHILLS) framework allows qualifying large loads to receive conditional transmission service while designated resources or network upgrades are still being developed. Similar reforms are being considered across other RTOs, including MISO and ISO-NE.

In August 2026, PJM proposed another approach. Unlike SPP’s policy, which allows large loads to manage their own firm service in exchange for accelerated interconnection, PJM’s Interim Resource Adequacy Service (IRAS) would compel large loads to procure their own dedicated capacity, or else face curtailment ahead of residential and commercial customers during grid stress events. While regulations are evolving, current proposals include eligibility pathways for distributed energy resources and demand flexibility to meet new capacity requirements. Data center legislation in Virginia and New Jersey anticipates this type of requirement by allowing distributed capacity to qualify as customer-procured power.

Taken together, these policies suggest an emerging two-level system: RTOs are developing mechanisms that allow large loads to connect before the grid is fully built out, while states are beginning to define how large customers can use flexibility to qualify for and manage conditional service. The next step is for regulators to ensure that off-site flexibility provided by distributed resources is included in these policies, and to develop the standards necessary to verify distributed capacity and translate it into a reliability benefit for large customers.

From early examples to scalable frameworks

These developments move the models outlined in our 2025 report closer to implementation. Large customers are starting to fund distributed capacity directly, utilities are creating pathways to incorporate customer-funded flexibility into large load arrangements, and RTOs are developing new ways to connect large loads before all of the generation and transmission needed for firm service is in place. The challenge now is to turn these early examples into repeatable frameworks that can scale.

Doing so will require greater clarity on a few core questions. What qualifies as incremental distributed capacity, and how can regulators ensure that sponsored resources are genuinely additional? How should distributed capacity be accredited and verified for different grid needs? When can distributed capacity provide enough locational and temporal value to support faster interconnection, and what should the exchange rate look like between flexibility and speed to power or other customer benefits? And what contractual, tariff, and regulatory structures can ensure these arrangements deliver measurable reliability benefits without double counting or shifting costs to other customers?

These questions are growing increasingly urgent as the demand from data centers grows. Regulators, utilities, RTOs, large customers, and distributed capacity providers now have an opportunity to establish clear rules for how flexible resources can help bring new loads online faster and at lower system cost. In parallel with regulatory developments, market actors will need to innovate effective transaction and product structures to facilitate data center investment in distributed capacity. The states and regions beginning to test these models are providing the foundation; the next step is to collect and standardize the emerging best practices so the benefits of these deals can stretch nationwide.