February 4, 2026
After roughly two decades of flat U.S. electricity demand, projected load growth has suddenly rocketed upward. Utilities have revised their five-year forecast for summer peak growth by 2030 from 24 GW in 2022 to 166 GW in 2025, a more than six-fold jump according to a recent Grid Strategies report. Datacenters, new industrial loads, and ongoing electrification are driving demand growth on a system where major upgrades take years to permit, finance, and build.
Over that same roughly two-decade period of flat demand, residential demand response has been mostly limited to mechanisms like thermostat adjustments, behavioral nudges and time-of-use rates. These pilot programs have importantly built the foundations for virtual power plants (VPPs) as a resource that can increase the load served by the grid without adding expensive power generation, transmission and distribution upgrades. But to serve the grid of the future it’s clear we need an ‘all of the above’ approach that includes scaling existing VPPs for more gigawatts of impact and to expand their concepts to new large loads like datacenters as they come online.
VPPs require coordination and flexible loads
A Virtual power plant is a demand-side management approach that coordinates flexible energy resources so they can act as a dispatchable grid resource. The coordination between grid stakeholders is essential to the viability of a VPP. As loads become more distributed, more orchestration between loads and customers is required to aggregate a meaningful response. As load becomes more concentrated, you can achieve scale with less coordination, but flexing or curtailing the load is often more expensive or onerous to the customer. Grid demand growth is coming from all sectors so developing both coordination and flexibility potential into grid loads should both be pursued for maximum impact.
Flexibility increases grid infrastructure utilization
Most VPPs have previously focused on reducing or shifting usage during the highest-demand hours to relieve grid stress. This is still a major value proposition for VPPs but it is not the only one. When T&D infrastructure cannot be expanded quickly enough to meet projected grid demand, the next best option is to increase the usable capacity of existing infrastructure. VPPs can enable that by reshaping load during any hours or locations where the system is constrained. It is a way to create deliverable capacity without waiting for distribution system upgrades or a new transmission line.
VPP’s also provide more potential benefit as grid spending is increasingly focused on infrastructure. EIA reports that major-utility spending to produce and deliver power has risen over the last two decades, driven primarily by capital investment in electric infrastructure, which more than doubled over that period. In the coming years, total electricity use is forecast to increase even faster than peak demand, driven by new customers with high load factors such as datacenters. When the system runs tight for more hours of the year, load flexibility becomes valuable beyond just the top few peaks. VPPs can maximize infrastructure through:
- Congestion Relief: shifting or reducing load when transmission is constrained
- T&D Asset Management: reducing feeder and transformer overload risk
- Interconnection Acceleration: making new loads feasible before grid upgrades are built.
In other words, the grid now needs flexibility not only for a few afternoons, but for the broader set of ‘hours that matter’ across more months and locations. As new large loads are connected those months and locations increase, and so should VPP utilization.
VPPs look different across sectors
Residential: lots of devices and sites, lots of optionality
Residential flexibility scales when it is automated and coordinated enough to aggregate many entities and often takes the form of:
- Behind-the-meter batteries that can export to the grid or simply power homes to remove their load from the grid
- Managed EV charging can shift large, flexible loads away from constrained windows
- Smart HVAC control can flatten demand when coordinated amongst many locations
Commercial: fewer sites, big controllable building systems
Commercial buildings often have centralized controls and thermal mass that lend themselves to repeatable dispatch such as:
- Pre-heating or cooling and other HVAC system optimization
- Lighting controls
- Building management system schedules tied to price and grid signals
Industrial: targeted programs, strong economics when aligned
Industrial flexibility tends to be site-specific because they operate in a complex business environment and must manage many parameters that affect their demand flexibility. The best industrial programs focus on a limited number of high-value sites where the operational and economic case is clear with mechanisms like:
- Industrial process scheduling
- Onsite generation and battery or thermal storage
- Interruptible load contracts
Energy efficiency is also a critical force multiplier for VPPs because it lowers the load served at any site and expands the load that can be shifted or curtailed without affecting comfort or operations. In homes, tighter building envelopes and efficient HVAC systems reduce runtime and cycling, making dispatch more predictable and curtailments easier to sustain. In commercial buildings, LEDs, VFDs, and tuned controls create dispatchable demand response during peaks and improve forecasting. In industrial settings, energy efficiency makes site-specific strategies more reliable and economical.
Datacenters: developing load flexibility potential
Datacenters are the defining new load on the grid because of their massive demand projections and unique load profile. Datacenters tend to be large, built quickly, and have a high load factor. That combination stresses interconnection and planning, but if power demand flexibility is designed into the operations of datacenters they can have a softer impact on grid infrastructure and grid demand peaks.
A recent study by Camus Energy, encoord, and Princeton ZERO Lab argues that a small amount of onsite flexibility combined with a ‘connect and manage’ approach to interconnection can enable datacenters to connect to the grid 3–5 years faster than traditional approaches. They argue that instead of requiring fully firm service for maximum demand all 8,760 hours per year, the load agrees to be capped or interrupted during the few hours when the system is constrained. Grid-wide VPPs can also help make those interruptible contracts workable by freeing up grid power from aggregated flexible resources elsewhere on the system to limit datacenter interruptions.
That idea is strengthened by early demonstrations of AI cluster flexibility. Emerald AI’s Phoenix field work and corresponding paper estimates the potential to reduce AI cluster power demand by ‘roughly 25% for up to 200 hours a year.’ The study showed how a datacenter could quickly reduce site demand in the case of a simulated grid emergency which resembled an actual CAISO load shed event in August 2020.
Reenacted CAISO power event and how Emerald Conductor responded to the curtailment requirement
Flexibility on this scale with quick-reacting, software driven load flexibility can dramatically change datacenter interconnection timelines by reducing the grid infrastructure upgrades required before the datacenter comes online.
Build more grid, but also build more load flexibility
The U.S. power grid has entered a period where demand projections are growing faster than major infrastructure can be planned and built. VPPs are a key part of the ‘all of the above’ strategy needed to meet this demand. The same coordination concepts proven in small scale pilots now need to be standardized and expanded in scope across all major end users, including datacenters and industrial loads. VPPs can arrive fast enough to accelerate load growth in the years it takes for infrastructure permitting and construction while also enhancing all grid infrastructure’s ability to serve demand economically and reliably.
