Google Enters Into Renewable Markets

Google and its parent company, Alphabet Inc., have long put themselves forward as leaders in renewable energy use. Last week, they took that a step further by announcing that they are now “working toward 24×7 clean energy everywhere we have data centers.”

Because renewable energy supply fluctuates based on the availability of things like sun and wind, that means that in practice, Google will have to shift computing demand in its data centers to renewable energy at times when the supply on the power grid is at its peak. This is a form of demand response in which energy consumers change their power consumption to better match the grid. But this version has a twist—a twist so big that it’s really an inversion of the concept.

Demand response has existed for years. There are entire companies built for the purpose of managing large energy consumers’ demand at times when  supply is tight and prices are high. In this supply-constrained paradigm, demand response is about reducing strain on the grid and lowering prices. It was a substitute, in other words, for increasing energy supply, which would require building more power generation capacity and transmission, consuming more fuel, and emitting more carbon dioxide.

Google’s plan is pretty much the opposite of that. It operates in a supply-abundant paradigm where demand moves to soak up power—in this case, renewable power—at times when power is plentiful. Rather than responding to high demand by shutting things off, Google will respond to high supply by turning things on.

Prior to the wide deployment of renewable energy, peak demand hit at midday in summertime. That’s when temperatures would tend to be hottest and demand for cooling was highest. Demand response sought to turn things off at those hours.

Maximizing renewable energy consumption, though, means optimizing for a supply curve that looks different. In grids with a significant amount of solar generation, there will be ample zero-carbon generation at midday, meaning that companies will want to shift consumption to those hours. Wind generation peaks in the mornings and evenings, meaning that any large electricity consumer will want to push its peak consumption into those same hours.

There are a few other things about this plan that highlight how much our grid has changed, and how responsive to the grid a big power user like Google can now be. Most traditional demand response programs target things like cooling, refrigeration, and lighting, but Google’s demand response is specifically about computing. Information technology accounts for more than 80% of today’s total power consumption, and the “hyper-scale” data centers Google and others operate have wrung out basically every other possible energy efficiency. There’s not much else left to move around.

Another thing to consider is what Google hints at: demand can’t be shifted only through time, it can also be shifted through space. “The first version of this carbon-intelligent computing platform focuses on shifting tasks to different times of the day, within the same data center,” the company says. “But, it’s also possible to move flexible compute tasks between different data centers, so that more work is completed when and where doing so is more environmentally friendly.”

Google calls the computing load that it can shift its “carbon-aware load.” That fungibility of processes is something special—I can hardly change my home air conditioning demand by cooling someone else’s house—and not every company will be able to shift its energy demand around the globe in quite the same way.

Google didn’t give any financials for its plan, but it’s entirely possible that it will actually save money as grid power prices go to zero (or below) when renewable generation peaks. Newly built renewable energy, after all, is not only the cheapest source of new power on most grids, it’s also cheaper than a lot of already-operational power sources. It might even act as a hedge against some of Google’s own renewable energy purchase agreements, signed years earlier at fixed prices that are higher than current grid power costs at certain times.