:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/62710792/energy-vortex.0.jpg)
This story was originally published in June 2016.
Several years ago, at a clean energy conference, I saw a presentation by entrepreneur and venture capitalist Bill Gross, founder of the technology incubator IdeaLab. He was discussing eSolar, a company he helped found and of which he was, at the time (around 2010, if memory serves), CEO.
In the course of his presentation, he said something that has stuck with me ever since.
Substituting computing power for raw materials
eSolar is a concentrated solar thermal company; it builds solar power plants that consist of hundreds of mirrors that focus sunlight on a central tower, where the heat boils water and drives a turbine.
The company’s innovation was that instead of using fancy, parabolic, custom-built mirrors that had to be specially manufactured and installed, it used small, plain, flat mirrors that were easy to mass-manufacture and could be installed by any skilled laborer.
What was lost in efficiency through the use of simple mirrors was regained through the development of sophisticated tracking software; the mirrors closely track the sun as it moves across the sky.
In other words, said Gross, eSolar shifted its focus away from materiel and into computing — away from stuff, into intelligence.
After all, he said, physical commodities (like silicon and steel) have a tendency to get more expensive over time, while computing power just keeps getting cheaper. To the extent that you can substitute the latter for the former — intelligence for stuff — you save money in the long run.
I’ve thought about that a lot since, the way computing power and software can help dematerialize and accelerate the transition to clean energy.
Large-scale energy transitions have historically been slow. Do they have to be?
There is a certain kind of pessimism in the energy world, represented best by the work of the University of Manitoba’s Vaclav Smil, that cites previous large-scale energy transitions to question whether the transition to clean energy can possibly proceed as fast as climate hawks want it to. (For the latest iteration of that argument, from University of Cambridge engineering professor M.J. Kelly, see this paper.)
Such transitions have typically taken many decades, up to a century, as one energy infrastructure is replaced by another.
If there is any hope in the face of this grim historical truth, it is in the nature of the current transition. There are two things about the shift to clean energy that might set it apart.
First, after centuries of greater and greater scale (peaking in the nuclear power plant frenzy of the 1970s), energy technology is starting to get smaller. Where once there were a few gigawatt-level power technologies available, now there are many options, ranging in scale from gigawatts all the way down to kilowatts: wind turbines, solar panels, large-scale energy storage (like pumped hydro or trains), microturbines, microgrids, fuel cells, home batteries, electric cars, smart thermostats and appliances, and on and on.
These "distributed energy resources," or DERs, have advantages over huge power-generation technologies like coal power plants. Because they are smaller, innovation can be spread across dozens or hundreds of parties instead of just a handful of utilities. Small technologies iterate and improve faster.
This blossoming of energy technologies means that where previous energy transitions (e.g., the substitution of coal for oil, or oil for biomass) were almost entirely infrastructure problems, with a few big players mustering huge, long-term investments, the current energy transition is taking the form, at least in large part, of a technology market. And technology markets move much faster than infrastructure.
Critics of DERs point out that they also have a flaw: They’ve become decentralized, but demand hasn’t. Electricity demand is still concentrated in large, energy-intensive cities, with their factories and hospitals, and will be even more so in coming decades. Cities need large amounts of steady, reliable power. Meeting that demand with variable renewable energy (wind and solar) is a huge challenge.
If there’s any way to tackle this problem, it’s not just to build more and more wind and solar until they can match the output of centralized power plants. (That would require a lot of land, given their generally lower energy density.) Building is only half the solution.
The other half is to link and coordinate DERs better, using sensors and software, so that maximum energy services can be wrung out of every single kWh generated.
Sensors and software don't look like traditional infrastructure. They look more like an information technology market. And we know that IT markets move very quickly indeed. Once you’ve developed good software, there’s no incremental cost to deploying it 10 or 1,000 times, and every instance of it can be continually upgraded, at minimal additional deployment cost.
To the extent that the clean energy transition is a software challenge, we can expect it to move far more rapidly than previous energy transitions, simply because software moves much faster than hardware.
Energy is becoming a software business
What made me think of all this is a story in Utility Dive about the home battery market. There have been lots of questions about whether home batteries are really ready for the consumer market, whether the payoff is worth the investment. Consequently, there’s been lots of focus on improving battery performance or finding new battery chemistries.
But it turns out what’s going to unlock the full value of all those batteries is aggregation — smart software that will link them all together and deploy them as one big super-battery, with the power and duration of grid-scale storage.
As policy director Ted Ko of the storage aggregator Stem put it to Utility Dive, "In the long run, storage is a software business."
I’d put it even more broadly: In the long run, DERs in general are going to become a software business, at least in large part. Solar cells, batteries, wireless chargers — they’re all eventually going to be commoditized. The value will come in squeezing more out of them by linking them up and making them work together more intelligently — getting scale out of coordination rather than sheer size.
In other words: substituting computing power for raw materials, intelligence for stuff.
We don’t really have a model for an energy transition that involves intelligent coordination of distributed resources rather than blunt-force deployment of centralized resources. We’ve never had the information technology to do it before. Even today, energy IT is in early development — we have only the faintest idea where it might go or what new capabilities it might unlock.
So I think Smil and other clean energy pessimists are, at the very least, wrong to be quite so confident in their pessimism. Much of what’s going on in the energy world today is genuinely novel, without historical precedent. Small, smart, nimble technologies are swarming up around big, dumb, slow ones, a process Michael Liebreich, head of Bloomberg New Energy Finance, likens to mammals overrunning dinosaurs.
It’s tough to predict exactly how it will play out. But there’s good reason to think it will happen faster than any previous energy transition.
