The Fuse

Arun Majumdar, former head of ARPA-E, on Energy Innovation and the Valley of Death

by Leslie Hayward | April 03, 2015

Arun Majumdar has served as Vice President for Energy at, where he created several technology initiatives and advised the company on its broader energy strategy. Previously, he served the Obama Administration as the Founding Director of the Advanced Research Projects Agency – Energy (ARPA-E). He is currently the Jay Precourt Provostial Chair Professor at Stanford University, where he serves on the faculty of the Department of Mechanical Engineering.

HAYWARD: Arun, there has been a lot of discussion in recent years about the role of government in spurring innovation, but you are unusually well qualified to give insight on this topic. What can government accomplish that industry cannot?

MAJUMDAR: The role of the government is in the area of research. If we are to change course in our energy ecosystem—whether it is transportation fuel or electricity generation, and whether it’s for energy security, the economy, or the environment—that shift has to rely on innovation.

Innovation comes from long-term research in science and engineering, which has to come from the government. These days, we can’t count on industry to support risky research ventures that might only produce revenue in the 15-20 year time frame.

Of course, this wasn’t always the case. If we look to history, Bell Labs was the source of most of the information technology that we see today. Bell was part of AT&T, and AT&T was a monopoly. It needed to develop technology for the public good, and it was the origin of breakthroughs like transistor-integrated circuits, lasers, and fiber optics.

Institutions like Bell Labs are no longer supported by large companies for a variety of reasons. So research that will have a commercial impact in 15-20 years is done in universities, national labs, and startups, with the support of the U.S. government.

Government support of energy technologies has come under fire in recent years. Can you tell us why?

It’s very important to draw a distinction between government support for research and government support for manufacturing. A few years ago, there was a great deal of politicization of government support for Solyndra, which was unfortunate. Tesla was given a Department of Energy (DOE) loan guarantee to create a manufacturing plant, and they have returned the loan with interest, ahead of schedule, while introducing significant innovations in the auto industry. But the loan guarantee program that DOE used to support these companies were supporting manufacturing, not fundamental research. I don’t think there’s any political argument that there isn’t an important role for government in supporting basic scientific research, and not only in the energy sector.

I don’t think there’s any political argument that there isn’t an important role for government in supporting basic scientific research.

I will also add that when those energy companies that went out of business a few years ago, it was a normal phase that new industries generally go through. In any emerging industry, you will see consolidations and bankruptcies—it’s part of the process. We always talk about the Big Three auto manufacturers in the United States, but not of the hundreds of other companies that failed or consolidated in the early days of the personal automobile.

Can you tell us about the “valley of death” that emerging technologies face as they move from the lab to commercial production, and what can be done by both industry and government to help promising energy technologies bridge it?

There are multiple valleys of death. I’ll discuss a few of them.

The first valley of death is demonstrating proof of concept. If someone has an idea, and if they try it out in the lab and they can get it to work—that’s proof of concept. Proof of concept is necessary but not sufficient. ARPA-E funded proof of concept research.

Then comes proof of integrated systems. That’s when you take your technology, which has demonstrated proof of concept, and put it into prototypes that enable people to see how it can serve a useful purpose. It’s the next step beyond the “idea” stage, where you justify funding further research.

After the proof of system has happened, then the industry needs to determine that they can develop a product or business around the technology. Then, you face the challenge of access to capital, and the various valleys of death that relate to achieving scale in manufacturing the product.

In fact, there are other valleys of death as well. But in the early stages, that need for capital to build manufacturing capacity and scale is the biggest concern. And frankly, the venture capital market has withdrawn from energy in terms of new investments. It creates a huge gap.

Can you tell us why venture capital isn’t putting money into energy?

The current venture capital system is designed to return five times the investment in five years. That’s not how energy systems work.

There’s a few reasons for that. But the biggest concern is that the current venture capital system is designed such that they invest in a company and it produces, say, five times the investment in five years.  There are some exceptions to this rule, but five times the investment in five years is the general expectation. It might be fine for information technology and computing, but that’s simply not how energy systems work, when you face a longer timeline between initial investment and producing revenue.

There are a very few venture capital groups looking long-term and focusing on energy, but it’s not the norm. Young companies and technologies can also struggle to access public capital once they’re out of the venture capital phase. These are some of the systemic issues facing young energy technology.

The other challenge is unpredictable market signals—we don’t have a long-term policy on carbon, for example, and that creates uncertainty for investors.

So venture capital is an area where we see division between information technology (IT) and energy technology. But what about the growing links between energy, specifically in transportation, and IT. What are some exciting things that you see emerging?

There are so many new business models emerging from this growing connection: car sharing, Uber, Sidecar, etc. And then there’s the possibility of self-driving cars, which can potentially solve so many of our transportation problems.

I think that the electrification of transportation, and the ways we can harness that to reduce our fuel consumption, is a very big deal. There is so much innovation going on, here at Stanford and elsewhere, to improve the performance and lower the cost of lithium batteries. Lithium-ion is just one type of battery, and there are all kinds of new battery technologies emerging—I believe the cost will probably come down by a factor of 2-3 by end of this decade, which will be pretty amazing in terms of what it does to affordability and the range of electric vehicles.

There’s also a big role for IT to play in how we manage capacity utilization. Our system now is very inefficient; most people drive individual cars, but they usually parked. Owning a car and keeping it parked 95 percent of the time is just not good use of time and money, but IT and autonomous vehicles can help solve that. People will be open to vehicle sharing if it helps them both save money and move faster from point A to point B, and there are already plenty of cities where people prefer not to own cars, because it’s too much of a liability.

This might be less important in the United States than around the world. Internationally, we are seeing many countries globalizing and experiencing rapid economic development and increasing transportation demand—but it’s nearly impossible to follow the U.S. model of personal transportation in Singapore or Hong Kong or Delhi. It’s not sustainable for every person to have a car.

So, the imperative is to move towards public transportation, and maximize capacity utilization, such that the number of people being transported goes up, and the per-capita energy used goes down. It’s a problem that I don’t know how to solve, and I don’t know anyone who does at the scale and cost that is needed. But given population density, it’s a huge imperative. I also think we are going to see this area develop in ways nobody could have imagined as recently as five years ago—but we still have a long way to go.