Science Talk

Vinod Khosla: Searching for the Radical Solution

Clean technology investor Vinod Khosla, co-founder of Sun Microsystems, talks with Scientific American editor Mark Fischetti about the energy payoffs to be had by reinventing mainstream technologies

Clean technology investor Vinod Khosla, co-founder of Sun Microsystems, talks with Scientific American editor Mark Fischetti about the energy payoffs to be had by reinventing mainstream technologies. Web sites related to this episode include

Podcast Transcription

Steve:          Welcome to the special Web extra edition of Science Talk, the podcast of Scientific American, posted on December 23rd, 2010. I'm Steve Mirsky. On this episode…

Khosla:         …Environmentalists have been very, very good about identifying the problems we need to solve. They're horrible at picking what the answers are. So I actually believe most environmentalists, most of the time, are getting in the way of progress, real economic progress.

Steve:          That's Vinod Khosla. He's perhaps the world's best known investor in clean technologies. Khosla was interviewed recently by Scientific American editor Mark Fischetti at the Going Green conference in San Francisco, part of which became a Q&A in the January issue of the magazine. Here's an edited version of that interview.

Fischetti:          This is Vinod Khosla. I'm sure most of you know him. For the records, he is the founder of Khosla Ventures, and which does a lot of funding for the kinds of things you're all interested in; and, you know, he's also formerly general partner of Kleiner Perkins and founder of Sun Microsystems. And I think he now qualifies as a pop icon too. So the Forbes 400 is out and [Vinod's] in there.

Khosla:         Pop icon for nerds. (laughter)

Fischetti:          The question of the day here seems to be: Which clean technologies are paying off and there's lots of ways to try to answer that so I thought maybe we could use a few of mantras that we hear in clean tech and venture capital to try to, maybe, see a way to an answer and [then we'll] talk about a few other events. So, you know, one of your mantras: If it doesn't scale, it doesn't matter. So which clean technologies seem to be scaling and which seem to not be?

Khosla:         You know categories, wind of course is scaling, but there's not a lot of innovation in [wind]. There's a few brand new things, but I am surprised at how little innovation there is. What wind really needs is storage and storage hasn't yet started to scale. Almost all [the attempts are toyish],  I would call them. We need something radical, and I do believe it will come along. Solar, solar seems to be doing well. There's way too many companies trying to do the same thing and not enough trying to d[educe] anything radically different. If you want to come back to that question of what it takes to be successful in solar today—the market itself is growing, not declining in the cost fast enough and definitely not fast enough to reach unsubsidized market competitiveness, which is one of my mantras. It's mostly the fault of investors and entrepreneurs who are trying to do the next marginal thing as opposed to the next radical thing. We can come back to it. The most interesting area to me is the one that people [soured] on the most, which is biofuels. Amyris had an IPO, [it was] very, very successful. Another one of our companies, Devon, is on file. And I see a series of those because I think the economics in that sector will prove themselves out in the next few years, and there's,[if you] look inside the covers, there's enough proof that half a dozen, not just our technologies, but half a dozen technologies will make economic sense; some with subsidies, some without subsidies and those are the ones that are most interesting to me because they'll scale infinitely.

Fischetti:          Mantra number two—and I like this one—don't invest in clean tech, invest in main yech.

Khosla:         What is main tech? Solar [and wind are], sort of, this very narrow definition of clean tech, probably the least interesting segment, but there are some really interesting new things in solar we could talk about. We're doing engines—[a] really exciting area; I suspect some of the biggest companies will be built in radical new engines. We're doing air conditioners. Now that to me is main tech. Now, air conditioning that doesn't cost any more and takes 80 percent less energy—now that's what we're talking about. Lighting that costs, pays for itself in the first 12 months, not in 12 years, [that] is [what] we're talking about. We're doing glass, we're doing cement. That's the infrastructure of society. It's not this sort of fringe, clean tech stuff that depends on subsidies. That's my view. If you are going to create, as a community, say, 10 new Googles that are, in your terms, clean tech, they're going to have to main tech companies—they're going to be mainstream markets that work unsubsidized. My personal view is anything that doesn't achieve unsubsidized market competitiveness within seven years of starting to scale is not worth funding. And that rule applies to everyone of the companies we fund, because if they don't get there in that much time scaling—and there maybe three, four or five years of development before you start scaling, when you're doing R&D—but if you can't get to market competitiveness, you might work in the U.S., but you're not going to work in China or India. There's no subsidies in India, there's no subsidies for solar in Chile or Africa or most of the world. In most of the interesting markets, in energy-related markets, are high-growth, developing world markets. And if you're not competitive there unsubsidized, forget it, you are a niche company. So that's what I mean by main tech. That's why we're doing engines and lighting and air conditioners and all kinds of fun stuff that, frankly, people don't want to work on because they're not the same thing as the other next VC firm [is] doing.

Fischetti:          [Right. I don't think I've heard any] talk about any of those today. (laughter) Maybe cement. So that brings up a, sort of, related point. For some years now, at these kinds of conferences, sooner or later, some venture capitalist [says], "Well great idea, but you know, if it doesn't compete with oil"—for energy technologies—"at $40 a barrel, then I'm not interested." Or maybe some of them will say $50 a barrel. I'm wondering, a) why's that, since oil [has] really been well above that for sometime, and the big oil states have said that they want to make their basement at $80 a barrel. Where is the line?

Khosla:         I actually think a reasonable estimate, given that we're in a relative lull in economic growth, which means we're in the relative lull in demand growth, that the price is probably at a pretty stable point. So, one simple metric I use is, if it's market competitive today, unsubsidized [then you] work worldwide; if it's market competitive today, subsidized, [then] you're competitive just in the U.S. But let's say you're at $75, [whether it's subsidized [or] unsubsidized; chances are by your fifth plant, you'll be at $60; by your fifteenth plant, you'll be at 50. So it's a question of when. None of the biofuels technologies are at scale enough to be anywhere near their cost low point. So their first-plant economics, sort of, has to be roughly in the range and then will keep declining as you build more plants. My personal view is, there are technologies that are market competitive unsubsidized today at $75 oil. You know, if you build a reasonable size plant, 50 million gallons; not a billion-dollar plant, but say a $100 million plant. That's probably the biggest you can do on your first plant, because you don't want it to be too expensive. Those ones will keep declining in cost, and they'll decline to a point on the maturity of the technology. But you build the [fiftieth] plants for one technology, then the ecosystem around it will start developing; biomass will get lot cheaper, your feed stocks will get a lot cheaper. John Deere will be doing custom equipment for shredding and moving your biomass, and so then the ecosystem starts to help you. I think if you get to $75, we won't stop till we're at $30; and now it's real dollars, not inflated dollars. But my metric, and I['m on] the record [as] saying, by 2030, the price of oil in 2006 dollars will be $30 a barrel. I fully believe that, and I believe it won't be because we stopped using oil, it'll be because there's plenty of competition.

Fischetti:          Interesting, all right. You've [talked about] solar, wind, bio. It seems, again, if you go to these kinds of meetings, you know, [there] is a hot technology [every year]. So several years ago, right, it was wind and solar; you know, two years ago it was biofuels everywhere. Last year, it was the smart grid. It seems to be a moving target; maybe it is and maybe it isn't. If it is, [is] there a reason for that; and by the way, we [were] also talking about smart grid before. I'd like you to explain what you think smart grid means.

Khosla:         So one of the problems with hot technologies is—and maybe there are a few environmentalists here; environmentalists have been very, very good about identifying the problems we need to solve. They're horrible at picking what the answers are. So I actually believe most environmentalists, most of the time, are getting in the way of progress, real economic progress, things that make economic sense. One rule I use is economic gravity always wins. [You can't] challenge it. And that's why this mantra around unsubsidized market competitiveness. And environmentalists; you know, I was talking to somebody about the Nissan Leaf—$25,000-$26,000 car, $20,000 worth of batteries. Give me a break, right? The Chevy Volt, which is a good car, is expected by 2012, to ship 40,000 units. The Tata Nano, the first day it opened for registration, [you] could register to buy one, had 200,000 orders, right. Why does it matter what you do with the Volt? We have to make technologies like the Tata Nano—which, by the way, is the bulk of the growth in the world market, as in India and China, in automobile[s]. Economic[s]. So economic gravity is absolutely key, and this is back to main tech not clean tech. Yeah, so we can do these fringe things around electric cars, and we're very aggressively investing in batteries. But I don't think electric cars make sense economically. They make great sense for [a Tesla.] You know, a $100,000 car, [if the] battery pack [is] 20 grand, [it] doesn't matter. And it's frankly a fun, sexy car—great for anybody [to own.] But you're not a price-sensitive [buyer]. [The] bulk of the world is price sensitive; they're going to drive [the equivalent of] the Tata Nano. [You] want to solve any problem, make that low carbon. Or, and I am actually pretty hopeful that we will have batteries that cost one tenth as much. Now, almost certainly, the traditional lithium ion bulk batteries will not be around in 15 years. Now I'm saying something that nobody else believes, so I will go on the record again.

Fischetti:          It's a big surprise. (laughter)

Khosla:         You know we're doing solid state lithium ion batteries, maybe that'll work. We're doing magnesium batteries, maybe that'll work. We're doing radical other things, I call them quantum-nano-thingamajigs, right. We're doing all kinds of stuff, and I can't tell you what'll work. I am not saying one of our battery approaches will work. But almost certainly, the winning battery in 15 years is one that has very low probability of working in people's eyes today. I can tell you it is unlikely to be a battery that people expect to be a winner, and this comes back to a more—I'm wandering from your questions so you can ask me again…

Fischetti:          That's okay, I will.

Khosla:         I, sort of, have a thesis that the most interesting technology; and so there's clean tech companies but in main tech, to win the big battle of economic gravity and to win in the market, unsubsidized market competitiveness, all the companies that are trying things that have radical technologies—I call them, hey, if it has a 90 percent probability of failing as an attempt to develop this technology, then I like it. Why? Because they are likely to be the ones that are six-sigma technologies that have quantum jumps in performance. So I call this the black swan [thesis] of energy: Don't look for solutions in high probability areas. Those are all incremental. Look for solutions in the [tails] of the distribution of likely to succeed.

Fischetti:          I see. That brings me to a question.

Khosla:         So one of my favorite is we like to invest in stuff that has a 90 percent probability of failure. And no other investor likes that, and so we sort of have this field to ourselves which is nice. (laughter) I did tell my investors when we are starting the fund that we were non-fiduciary investors, so don't count on us to be fiduciary.

Fischetti:          Okay, [this is] a quote from you that sort of [feeds] right into that. In the last year you started two funds, [in a bad] fund-raising climate, that are quite substantial, and when you were asked about that you said, "Money is not the problem, finding innovative technical ideas is the hard part." Now a lot of people in this room would be wondering, saying, "What are you talking about? Finding money is exactly my problem. I can't find enough of it." So what do you mean by that? And I think it [feeds into] what you just talked about.

Khosla:         Well, you know look, [we have] plenty of money to invest. We're probably investing as aggressively as we ever did; we probably made more investments this year than last year, and last year we made more investments than the year before, despite how bad the climate was. I don't know if those numbers are exactly right, but roughly. You know, the trend is we have more people, so we're doing things more aggressively. The problem, at least to my point of view, is finding the people who want to do something other than the incremental or narrow non-scaleable market[s]. You know, you want to biodiesel from waste, great. It's actually a good market. Somebody will succeed, but could you build a really large company out of it? No, not likely. Could you do a one-off? Absolutely. And then there's people trying to make magic work where it hasn't worked. Algae? I looked at two dozen business plans, not one works. Now could you do an algae company for a high-value product like a nutraceutical? Absolutely. Those of you who're doing algae, I suggest shifting your [plans] to high-value products; it might work. At this point I am still open to looking. I look at every algae plan that comes in the door. But after two dozen, I haven't found one where the economics works. But it's worse than that. I can't look at their process cost and tell you what breakthrough will get them to the right point. So when I can't even identify the hypothetical breakthroughs, where's their improvement room for a 5x improvement? I can't find them. So that's how we look at [things]. You know, I think there's plenty of money. There's not enough breakthrough technologies, there's not enough great PhDs in these fields. Now I actually add up the number of PhDs in each of our companies, and I call every CEO and ask them, how many PhDs did you hire last month? I mean, this is my standard question. There's not enough technical talent for large breakthroughs and so people are doing incremental things, and our universities weren't producing any of these people till about two to three years ago when interest in clean tech or energy technologies increased. Now I think that's the fundamental problem. The good news is now all the good PhD students are going into these areas, so ten years from now, innovation will explode. That's the good news. If I might [take] another movement, there's this other mantra related to 'not enough money around' that, it takes too much money, right, to do clean tech; absolutely wrong. It's a fallacy. [Of course] journalists like to talk about it because they have something to write about. I have looked at our portfolio. The amount of money needed to cash [flow] break even or IPO or [reach a] reasonable point where you sell the company does not look any different for our clean tech portfolio, or main tech portfolio, than it does for the last 15 years I was at Kleiner looking at information technology, or telecom equipment, or pick your favorite area—enterprise software. Yeah, Internet, pretty small amount of money needed, but that's more an exception. If I look at it over the '90s, there was a lot of companies that needed $50 million to break even. Our lighting companies will need $50 million; our air-conditioning companies will need $50 million. There were a lot of companies that were doing chips or things like that, or telecom systems, that needed $100 million to break even; plenty of companies that fit into that category. Or there are a few companies that need $300-400-500 million absolutely, but we had biotech companies that needed $300-400, million and the distribution looks about the same. That's the surprise. I will shut up since we have little time.

Fischetti:          Yeah, let's open it to questions. Anybody, please use the [mics], yeah, so I can repeat it.

Fischetti:          What's the most exciting innovation [Vinod sees] in the next five years? And I would, can I amplify that with another quote: "I only work on things that are intriguing to me, and so I always switch every few years. I get [bored] of an area, I understand an area, I know it well, and it becomes time to learn something new.". So [where are] you learning [next]?

Khosla:         Look, I think every area I look at is exciting. I mean, I'm not being facetious. You know we looked at air conditioning, we expected we'd find better compressors. Well that's marginal. We discovered a brand new thermodynamic cycle. Nobody has really commercialized the new thermodynamic cycle in 50 or maybe even 100 years. I wouldn't have expected that; At least scientists I knew and experts I knew had done nothing new. I didn't think we'd been investing in glass, I didn't think we would be inventing a brand new engine type that's 50 percent more efficient at less cost than today's engines. When a hybrid is only 25 percent more efficient at best, at best; an internal combustion engine that gives you twice the efficiency boost that a hybrid does? So frankly, what's clear to me is as we look to old areas, what's wrong with it is people who've been in that field for 30 years are not taking a fresh look at what they've done when the world around them has changed. Power electronics—we're starting to talk about smart grid, but very little has changed. It should be completely revamped. Even 15 years ago there was a few hundred sources on the grid and millions of consumers. Today there's tens of thousands of sources. The grid needs to be redesigned. It has to be redesigned incrementally, one little piece at a time. Smart grid is about that; it's not about smart meters. Smart meters—there are only 100 million electric connections in this country or something, give or take 20 million. That's a small number; you get a buck or two a month out of a meter in savings; your share of the revenue, it's a pretty tiny market compared to $200 billion for reinventing engines. So I think every area I look under, no matter how archaic, there's innovation to be had. That's a surprise to me. I wouldn't have expected that.

Fischetti:          Let's try to give one more question and then we may have to wrap it up.

Wiedmann: My name is Jerome Wiedmann and I [want to talk to you about a] LaunchPoint technology project [that sounds like a new engine] It's an electronically controlled valve system, and studies have shown, from Bosch and others, that an electronically controlled valve system can increase the engine's efficiency by 30 percent without any other changes. Is this the type of technology [you're talking about?]

Khosla:         So, you bring up an interesting point. I don't know LaunchPoint Technologies.

Wiedmann: Yeah, it's a new, it's a small company, [a bunch of] PhDs.

Khosla:         But there's many, many areas of traditional mechanical engineering, where we are still using old systems like cams for timing things, like that. And when I mentioned power electronics, what you need to do to have precise timing is much better power electronics devices to rapidly switch on and off a valve, I don't know what…

Wiedmann: Yes.

Khosla:         Right, so it is absolutely an area of innovation. Engines shouldn't have timing based on the cam. That's a fixed thing. Driving conditions change, you should change timing. So it's just one example, whether that's an example or not there's a hundred things like that I've seen in power electronics, that should change and would change the application dramatically, that's one of them. I've always wondered why all valves aren't electronically controlled in engines. Why wind turbines have to run at a particular speed with a gear box instead of converting all the energy into power and then converting the power into the frequency and space and everything else unique? Entirely possible, very likely; nobody's paying attention to it, because smart meters are easy to do.

Wiedmann: Well [they have] a very nice functional prototype engine.

Khosla:         Thank you.

Fischetti:          Thank you Vinod for being here.

Khosla:         Thank you everybody. (applause)

Steve:           Check out the edited print version of this interview in the January issue of Scientific American. For Science Talk, the podcast of Scientific American, I am Steve Mirsky. Thanks for clicking on us.

Share this Article:


You must sign in or register as a member to submit a comment.

Starting Thanksgiving

Enter code: HOLIDAY 2015
at checkout

Get 20% off now! >


Email this Article