Google: How the Tech Giant Hit a 1GW Data Centre Milestone

As pieces of critical infrastructure, data centres are zero-downtime, always-on facilities with an insatiable appetite for energy.

They consume power at a fixed, relentless rate, and grid planners have long designed their systems around that assumption.

However, a recent study published by Duke University has found that dozens of gigawatts of new load could be brought onto the US grid if large users were amenable to even modest curtailment.

Lately, Google has been making a determined case that its facilities can be exactly that kind of flexible customer. The company has now integrated a total of 1GW of demand response capacity into its long-term energy contracts with multiple utilities across the US.

The announcement, made by Michael Terrell, Google's Head of Advanced Energy, represents the most recent step in what has become a notable shift in the tech giant's position in the energy markets.

How the system works

So, how exactly does this power-saving process work? The mechanics are rather straightforward, resting on the principles of "demand response".

In layman's terms, demand response is when a large electricity user, like a data centre, agrees to temporarily reduce or delay its power consumption when the broader grid is under strain during peak times. In practice, this helps to keep the system stable without having to build extra infrastructure.

Google has been working hard to introduce demand response capabilities to its data centres.

In effect, non-urgent computing tasks – the kind that do not require an immediate result – can be deferred or redistributed to other times and locations when the grid is under less strain.

Since announcing initial agreements with Indiana Michigan Power and Tennessee Valley Authority last year, Google has signed contracts with Entergy Arkansas, Minnesota Power and DTE Energy that incorporate demand response as a key resource for new data centres to connect more rapidly to local grids.

In a recent blog post, Michael described the logic behind the move: "Demand response can be deployed quickly to bridge the gap between short-term load growth and the longer timelines required to build new clean generation and storage solutions."

That gap is not trivial. In the US, the demand for energy for AI and data applications is projected to grow from 3.5% of total demand today to 8.6% by 2035, according to BloombergNEF, with data centre energy demand on course to outpace that of electric vehicles.

The details of the Michigan deal

Two days before the demand response announcement, Google revealed a separate but related agreement in Michigan that illustrates how these ambitions translate into practice.

The organisation announced plans to develop a new data centre in DTE Energy's service territory in Michigan, with a commitment to enable 2.7GW of new clean resources to support the local grid.

The breakdown of that figure tells its own story. 

The package includes 1.6GW of solar power, 400MW of four-hour energy storage, 50MW of long-duration energy storage and 300MW of what Google terms "additional clean resources", which could encompass wind, hydro, nuclear or geothermal. The remaining 350 megawatts will be covered by demand response.

Alongside the energy infrastructure, Google is introducing a US$10m Energy Impact Fund to scale and accelerate energy affordability initiatives designed to drive down monthly bills for communities in Michigan, including home weatherisation, efficiency technology innovations for households and energy workforce development projects.

Whether US$10m is proportionate to the scale of the infrastructure being built – or indeed the pressure that energy-hungry facilities exert on local grids – is a question the announcement does not directly address.

The ratepayer argument

Google's central argument is that flexible demand benefits everyone on the grid, not just the company itself.

Demand-side flexibility reduces the need for new infrastructure designed solely to meet short-term peak system use, which is a primary driver of electricity prices.

Michael frames it in terms of system-wide savings: "By allowing utilities to cover peak demand periods with existing grid resources, demand response can help optimise the build-out of new transmission and power plants."

Research cited by Google supports the broader principle – that even modest flexibility in large electrical loads can reduce costs across an entire power system.

Still, the company is candid about limits.

"There are limits to how flexible a given data centre can be," Michael reflects, "and this capability will only be available at certain locations."