A note about this series on energy revolutions:
Policymakers, tech plutocrats, and a cacophony of pundits serially talk about revolutions in energy and technology. “Revolution” is a powerful word but is so overused in these domains that it’s lost meaning. But real revolutions, by definition “radical and pervasive changes,” do matter. They create pivots in history. Using the proper meaning of the word, this fourth in the series (which began with Part 1) continues the exploration of real energy revolutions.
With the release of the DOE Energy Information Administration’s (EIA) latest International Energy Outlook, we are witness to a duel between two visions of the long-term energy future. One, the EIA camp, sees demand inexorably rising; another, the World Energy Council (among others), believes peak global energy demand is in sight — what they enthusiastically call a “grand transition.”
In the short-cycle realities of modern policymaking, long-term forecasts may seem, well, irrelevant. But such forecasts have power. Witness decades of energy policies based on, and hundreds of billions of dollars spent on, conquering the fear of peak energy supply.
The thread that connects the two peak theories is the need to justify policies that direct far greater use of expensive, so-called “green” energy technologies. Peak supply offered a rationale for an imperative to subsidize rapid deployment of alternatives — costs be damned — before the feared apocalypse of global energy shortages. Unfortunately for the peakists, technology has created global gluts in nearly everything, eviscerating the silly idea of peak supply.
But that lesson seems lost on those — many are the same people and organizations — now enthralled with the idea of peak demand, which offers a fresh rationale for pushing expensive alternatives. To wit: Society could more easily tolerate high-cost alternatives if overall energy demand stays flat while economies grow, since that would mean fuel spending as a share of GDP stays low.
So, what do we really know about long-term energy demand?
Since energy demand emanates from technology use, long-term energy forecasts are, at root, technology forecasts. And there are two features to tech forecasting: 1) the “known knowns,” i.e., the greater use of today’s known technologies; and 2) the “known unknowns,” the emerging technological revolutions, whose specific impacts are not yet known.
We can briefly dispense with the implications of known knowns. To believe that peak demand is in sight for the world means, effectively, that one has to believe also that the energy efficiency of known and widely used technologies, such as cars, lights, factories, or air conditioners, will improve more than the increased use of those technologies as the world’s economies grow. Simple arithmetic reveals a deep flaw in this assumption.
Ascetic Europeans, never mind profligate Americans, use 10-fold more energy per capita than do billions of the world’s poor today. Thus, when in the future half of the world’s population becomes sufficiently wealthy to afford the lifestyle taken for granted by today’s 15 percent, global energy use rises 500 percent. To keep demand flat, overall energy efficiency would need to improve by 500 percent — something that’s not possible on any imaginary horizon.
Even if you choose to believe that Silicon Valley magic could generate 500 percent efficiency gains (laws of physics be damned), to believe in peak demand you also have to accept another implicit assumption: that there aren’t going to be any significant new technologies deployed at scale that will lead to new and significant energy demands.
Of course, we know there will be new technologies and all those technologies will use energy since it’s the reality of the universe. (The physics is inescapable.) What we don’t know is the exact timing of widespread deployment, and thus, derivatively, the timing and scale of associated energy use. Nonetheless, we do know enough to predict the general energy implications of future technologies.
Consider herein some highlights of the energy implications inherent in tech’s 10 emerging revolutions — the known unknowns. All are areas where engineers and entrepreneurs are furiously inventing and itching to deploy at scale.
1. Cloud and “hyperscale” datacenters
First, the easiest prediction: The coming data tsunami will lead to more energy demand from the world’s data-centric hardware. Today’s global information infrastructure is already a huge energy consumer. Even Greenpeace has documented that the global cloud now consumes twice as much electricity as the entire nation of Japan. (This research is similar to my earlier work here.)
But that’s yesterday’s news, using yesterday’s infrastructure. As Intel notes, an avalanche of data is coming. Cisco forecasts a 10-fold increase in data traffic in just the next decade. How much more follows? No one really knows.
We are seeing the rapid emergence of a new class of energy-hungry “hyperscale” datacenters. There are 300 in the world and 100 more expected within a year, with that number doubling within five years. A 10-fold increase in this domain would be a lowball estimate, as the age of ubiquitous data takes hold.
Meanwhile, the energy efficiency gains (Moore’s Law) in computing have slowed down. The combination of these two trends — faster growth in data traffic but slower gains in data energy efficiency — portends a non-trivial rise in data-centric energy demand. And this doesn’t include the acceleration towards using artificial intelligence to extract more value from data — a class of software that is inherently far more energy-intensive than the comparatively lightweight computing done today.
2. Virtual and augmented reality
Most internet bandwidth is already consumed by video, and Cisco says it will account for 80 percent by 2020. That’s for pixels creating 2D images. What comes next? Virtual reality (VR), augmented reality, and useful 3D displays for medical, industrial, and naturally, entertainment uses. VR will unlock e-sports, already a nearly $500 million business, creating another giant sucking sound for bandwidth. Count on engineers conquering the technical challenges involved with eliminating clunky, nerdy goggles.
Going from pixels to voxels — the term-of-art for 3D data — entails orders of magnitude more data because one is going from a flat plane to a volume. A single megapixel image becomes a gigapixel 3D image, and 3D VR video requires as many gigabits in a second as your cell phone now uses in a month. The sooner 3D is practical the faster data traffic soars — and likely faster than current forecasts. The energy implications are obvious.
3. 3D printing and computational manufacturing
Rapidly maturing 3D printers — machines that can print a product in three dimensions, directly from a computer design using “inks” of plastic, metal, or other materials. These promise not just to democratize but also to revolutionize many aspects of manufacturing. 3D printers offer entirely new ways to design and build things as well as enable local desktop manufacturing.
But contrary to claims made by Al Gore and others, it’s unlikely that 3D printers will save energy. 3D printer are electric-intensive: printing a plastic object uses five to 10 times more energy per pound compared to conventional industrial injection molding, and as much as 100-fold more for converting a pound of metal into a product compared to casting or machining. The point of 3D printing is not saving energy, but gaining flexibility, proximity, and enabling designs or products that are impossible to fabricate conventionally.
3D printers may one day match the pound-for-pound energy efficiency of conventional fabrication. But some fear, with good reason, that the ease of making stuff locally (literally on-site), making so many different kinds of customized products, and facilitating products that are more easily replaced will inspire “profligate” consumption. That’s precisely what happened last time there was a manufacturing revolution thanks to mass production.
4. Transient and consumable computing
The emergence of entirely new classes of electronics now promises transformations as significant as going from vacuum tubes to transistors. Researchers and entrepreneurs are bringing to market biocompatible and even consumable microchips and sensors made from polymers and other organic materials, as well as a broad class of “transient” — dissolvable or disappearing — electronics. (See especially what John Rogers of Northwestern University is up to.)
Bio- and transient electronics make possible smart sensors to wirelessly measure and connect on or within the body enabling radical advances in real-time health information, diagnostics, and therapeutics, as well as agriculture and animal monitoring where traditional electronics are either not desirable or difficult to deploy and remove. Also enabled: nano-tools for hyper-precise and potent drug delivery targeting specific organs and even cells that disappear when the mission is completed,.
Consumable and transient electronics give new meaning to “planned obsolescence.” The industrial ecosystem to manufacturing this new electronics class will rival the combined scale of today’s pharmaceutical and semiconductor industries in both economic and energy consumption terms.
5. Wireless charging
We know two things about how society is powered: electricity dominates, and power cords are annoying. Electricity’s share of U.S. non-transportation energy use has risen from 33 percent in 1970 to 50 percent today. Within that trend, there is an even more rapid growth in battery-enabled portable products of all kinds. Wire-free or wireless charging has been a hoped-for goal since the days of Tesla and Edison.
Wireless charging is a standard feature on the newest iPhone and — in particular because of the availability of high-power gallium-nitride transistors — will soon migrate to bigger more power-hungry applications. You know wider deployment is imminent when new organizations emerge to promote the technology and associated standards: the Air Fuel Alliance and Wireless Power Consortium.
But convenience comes with an energy cost. Total energy used to recharge batteries roughly doubles when done so wirelessly. (To understand why, think about filling a bucket with a sprinkler instead of a hose.) But count on convenience trumping conservation. And it’s a convenience that will confer important gains for electric cars (solutions are already in the pipeline), opening a path to far more frequent on-the-fly fueling and the option to use small cheap battery packs (instead of massive and expensive ones). Wireless refueling would roughly double energy use per electric-vehicle-mile.
6. New classes of basic and computational materials
The modern chemical industry emerged at the start of the 20th century, followed quickly by the silicon semiconductor industry a half-century later. Together, these formed the basis of an economically far-reaching — and hugely energy-consuming — materials revolution. The next basic material renaissance is now underway, similar in character and eventually in scale. New classes of materials include carbon nanotubes, graphene (enabling “magic” conductors and filters), self-assembling materials, and metamaterials that can be engineered to exhibit unnatural properties (e.g., “invisibility”).
All of these new classes of materials will, in due course, lead to the rise of heretofore nonexistent manufacturing capabilities. What are the energy implications? As a rule-of-thumb, the more complex the material, the greater the energy cost to produce it. For instance, the energy cost of materials underlying today’s digital infrastructures amounts to roughly 1,000 times more energy used per kilogram to manufacture as compared to the kinds of materials that dominated the industrial economies of the 19th and 20th centuries.
7. Self-driving cars
Hype aside, self-driving cars are eventually coming, and with two obvious energy implications. First, people who could not otherwise drive will be able to hit the road, from children and the disabled to the elderly. In 20 years, America will have a combined total of nearly 100 million citizens under 15 and over 80 years old, with the latter category expanding rapidly.
Second, combine self-driving with ‘Uberized’ sharing and the net cost of car ownership plummets. If history is any guide, that will lead to more road-miles. When the technological tipping point is reached — lower cost of automation combined with higher incomes — we may see the end of mass transit. Uberized — and fuel hungry — self-driving cars will displace tens of billions of passenger miles now logged on (fuel-efficient) buses and trains.
8. Drones and personal air taxis
The combination of advanced lightweight materials and the class of sensor-computer technologies used for self-driving cars will also make drone use much more practical. Today’s small consumer drones can barely carry marginally useful payloads of a pound or two, and can only do so for mere minutes. What comes next are drones that carry serious cargo, including people. The energy impacts that follow will be dictated by the physics of gravity and air resistance.
Using an aircraft instead of a truck increases energy use about 10-fold per pound-mile of stuff carried. Thus when drones carry, say, 10 percent of the three trillion ton-miles of truck freight in the U.S., overall energy use for freight transport will just about double.
And if Uber and Airbus have their way, the age of auto-piloted air taxis is also coming. Who wouldn’t rather soar over traffic in an air taxi? And a new multi-billion-dollar drone-building industry will consume more energy than conventional factories thanks to the energy-intensive fabrication of ultra-lightweight materials.
Anthropomorphic robots constitute another class of energy-consuming automatons — albeit further in the future than self-driving cars and drone taxis. Though not in commercial use today, it’s no longer science fiction (as a DARPA contest dramatized a few years ago) to forecast practical robots on the horizon deployed in industry, health care and even domestic service. Like cars and computers, millions will be built in due course. Robots are extremely complex, also likes cars and computers, and energy-intensive machines to fabricate — and they need fuel to operate.
Humans are very efficient bio-machines requiring (and producing) roughly an average of 200 watts. The kinds of ambulatory robots that can be envisioned today require at least 10 times more power than humans. This means that when robots achieve the same market penetration as cars did circa 1920 — one per 10 people — energy used by robots will rival that used to feed all humans.
10. Hypersonic travel
The perennial pursuit of fast travel won’t stop with supersonic speed, which now looks feasible without the boom. Next is hypersonic flight at more than twice supersonic. What’s not to love about New York to Shanghai in two hours?
Consider, though, that at four to five times the speed of sound, atmospheric friction heats an aircraft surface up to a metal-melting 5000oF. That’s a non-trivial challenge, but odds are good that new classes of materials will conquer it. (Some promising prospects already exist.) The energy implications? Supersonic flight runs at least triple the energy use per mile — hypersonic yet more. Even if only a small minority of travelers fly hypersonic, overall aviation energy use will still rise significantly.
Of course, there are yet more known unknowns on the horizon aside from those listed above. One particularly worth mentioning is space tourism.
Here, it looks like the science is finally conquering the fiction. Richard Bransom’s Virgin Galactic has received FAA approval to fly tourists in to space. Jeff Bezos has said his Blue Origin vision is to see “millions of people living and working in space.” Elon Musk’s Space X business is anchored in a booming future for satellites with, he claims, a goal of putting lots of people on Mars.
When it comes to power there’s nothing quite like a rocket ship. Putting a half dozen people into orbit for a few minutes using a Space-Shuttle-sized craft requires engines producing 45 gigawatts per launch. One gigawatt is the power output of a nuclear power plant.
But long-term energy forecasters ignore pretty much anything to do with guessing technology’s future. Maybe that’s because economists dominate the forecasting game. They do so because of the obvious link between economic growth and consumption.
Of course, it is undeniable that the money factor is vital in estimating how quickly and broadly people will be able to buy and use technologies. Also crucial, however, are the episodic technological “disruptions” wrought by engineers. And economists have a long and miserable track record when it comes to anticipating such disruptions. Few examples are more relevant than the Club of Rome’s famous 1972 book Limits to Growth.
It’s fitting that dreams of peak energy demand are in vogue today, on the 45th anniversary of Limits, the godfather of peak supply theory. The book sold over 10 million copies back in the day. Perhaps not since Nostradamus had there been such success in popularizing the idea of end-days, inspiring decades of prognosticating and heavy breathing over the end of food, minerals, oil, and natural gas. We know what technology did to those forecasts. The fiction of peak demand will suffer the same fate.
Returning to economics, the best estimates today see global GDP expanding by twice as much in the next two decades as it did over the past 20 years. This means we are about to witness the biggest rise in buying power in human history. That will catalyze the emergence and use of all kind of technologies — as well as the energy needed to power them.
Take this to the bank: There will be a lot more energy used in the future.
Mark Mills is a Senior Fellow with the Manhattan Institute, a Faculty Fellow at Northwestern University's McCormick School of Engineering, and worked on science policy in the Reagan White House.