← Back to TwinStack Brief
Data center

Google Scales Voltus Deal to a 1 GW Virtual Plant on the PJM Grid

Network WorldOct 24, 2026

Google has agreed with energy supplier Voltus to build what it calls a 1-gigawatt virtual power plant, according to Network World. The agreement expands a 100-megawatt deal the two companies signed earlier this year, a tenfold increase in stated scale.

How a virtual plant works

The plant generates no electricity of its own. It pools flexible grid capacity into one resource that can be called on when demand peaks. That capacity can come from batteries or from load-shedding, in which participants temporarily stop consuming power. Examples include pausing industrial processes or adjusting tens of thousands of smart thermostats at once. The effect is to move consumption to a period when demand is lower.

Voltus supplies the commercial mechanism. It offers incentives to companies that agree to cut usage for a time, and it also provides inducements to electricity resellers, which can pass them on to their own customers.

Where it applies

The arrangement will serve the PJM grid region, the largest power grid in the US, covering 67 million people across 13 states and Washington DC. Placing the program on the country's biggest grid gives the 1 GW figure a large pool of potential participants to draw from.

The pressure behind it

The report describes the deal as a public relations win for Google. Rising electricity demand from its new data centers, and from those of other hyperscalers, is pushing up prices. The added load is also straining local infrastructure in some areas, which raises utilities' capital costs and, in turn, prices for existing consumers.

Paying others to use less power addresses that tension from the demand side, without Google building generation itself. The source does not say how the 1 GW divides between batteries and load-shedding, when the capacity will be available, or what the incentives cost.

Why it matters: The deal shows a hyperscaler trying to offset grid strain by buying flexibility instead of only competing for supply. Its real effect on prices and infrastructure costs will depend on details the source does not give, including delivery timing and how reliably the capacity responds at peak.

TwinStack automated analysis, based on reporting by Network World. Check the original for full details.