Marc Andreessen may be the most dedicated optimist in Silicon Valley. During a year when the public’s attention was often focused on international conflict, mass shootings, and a bitter election, the co-founder of venture capital firm Andreessen Horowitz has insistently pointed toward progress. Until it fell silent in September, Andreessen’s hyperactive Twitter feed served as a guide to the good news: people rising out of poverty, the surprising durability of the US economy, and all manner of fast-growing tech products and services.
And then, on September 24th, the account froze. “Taking a Twitter break!” he tweeted, before deleting the rest of his tweets. He has never said why — and, when we met in his offices on Sand Hill Road in Menlo Park, he declined to say much beyond the fact that his seat on Facebook’s board had presented him with a conflict.
Plenty of subjects stress Andreessen out — he believes the next election will somehow be even more contentious than the one we just survived. But generally speaking, he is as optimistic as he’s ever been, whether the subject is job creation, the future of autonomous vehicles, or the remarkable progress in the quest to build flying cars.
What’s made you feel optimistic lately?
A bunch of things. It’s honestly hard to not be an optimist in this job, because we get 2,000 founders a year who come in here, sit in that chair right there, and they just tell us everything. They unspool the future to us, and they’re optimistic or they wouldn’t be here. Then all the new ideas.
I wonder whether people outside a firm like this would be more optimistic if they could see and hear all of that stuff. I’m not sure. We’re at a very weird cultural movement. A lot of people are just fundamentally unsure and I think, to a certain extent, hearing more good news may just make people angrier.
Listening to these pitches, are there particular moments where you think, “Wow, within the next half a decade or so, life is going to change a lot more than it did in the previous five or 10 years?
The big thing happening in the economy that is not well understood is that there are two very different parts of the economy. There’s the part where there’s rapid technological change and very rapid productivity improvement. In that part of the economy, you would include things like media — podcasts versus radio is a great example of rapid technological change. Streaming versus broadcast TV and so forth. Retail is obviously going through massive productivity improvements and changes. Manufacturing — the price of televisions has dramatically collapsed in the last 10 years, and the TV [you can buy] now for 400 bucks is like science fiction compared to what you could get 10 years ago. Cars are going through rapid productivity changes now. You’ve got these sectors that have gone through these massive productivity changes. They are characterized by rapidly improving quality, but also collapsing prices and rapid productivity growth.
You’ve got this other, second part of the economy that’s the exact opposite — where quality is not improving and prices are rising. There you talk about health care, where it feels like every year you pay 10 or 15 percent more and you get some new stuff, but you don’t get a lot of new stuff for the money. You talk about education, where the rising cost of a modern college education is just staggering. Actually, it’s funny — all forms of media are collapsing in price, other than textbooks. Textbooks are rising in price exactly the same as college tuition, which is a good illustration of the difference between the two sectors. Construction, real estate — there’s been a lot of conversation recently about how we’re still fundamentally wedded to physical location way more than you would think at this point with the internet and everything else.
In the industries where there’s rapid productivity growth, everybody is freaked out, because what are people going to do after everything gets automated? In the other part of the economy, that second part, health care and education, people are freaked out about, “Oh my God, it’s going to eat the entire budget! It’s going to eat my personal budget. Health care and education is going to be every dollar I make as income, and it’s going to eat the national budget and drive the United States bankrupt!” And everybody in the economy is going to become either a nurse or teacher. It’s really funny, both sides of the economy get polar opposite emotional reactions.
I go through all of that to say that the tech industry has been able to build startups and new technologies against that first category but not that second category. Tech is super present in retail in the form of e-commerce, we’re super present in media in the form of internet, we’re super present in consumer electronics in the smartphone. We are very much not present, in what we would consider to be a healthy way, in education, health care, construction, childcare, senior care. The great twist on that is that second category — that’s most of the GDP. Most of the spending is most of the GDP, and these are the areas where we have not yet been able to crack the code.
The big thing happening in the valley right now is valley entrepreneurs are getting much more aggressive at starting tech companies in that second category. We are seeing a lot more startups going into, especially, health care, biotech, different applications on that side. We’re seeing a lot of startups going into education. We’re not so much seeing startups going into construction, but we’re seeing lots of startups going into the collaborative work: Slack and GitHub and telepresence and Skype and all of these things, that in theory, in the long run, will make geography less relevant and maybe solve the cost of housing and access to economic opportunity.
Financial services has had rising costs, not falling costs, which it really shouldn’t have — we’re seeing a lot of fintech startups going into that. It’s this reorientation, and the valley is still doing all of the first category things, but it feels like we’re now layering on a lot of the second categories.
That’s very exciting, in that we can have the impact in the second category like we had in the first category, which I think would be good. Of course, the other possibility is this is just pure hubris, and five years from now you’re going to play this back and you’ll be like, “Aha! Dot-com bubble 2.0. He was full of it and he didn’t realize it. The entire thing came crashing down.” How audacious or insane is it to think that you could bring tech to health care or education? It’s probably 50/50.
What’s the state of robotics, particularly robots you might use around the house?
It depends on what you mean by robot. One definition of robot is every thermostat is a robot now. Thermostats used to be literally an analog thing. Now they’re digital — they’re computers. It’s happening quite quickly.
The other thing I think people miss is, you look at a lot of these companies and you say, “This company makes self-driving cars. This company makes VR headsets. This company makes tabletop robot toys. This company makes drones.” And they all look like they’re separate categories of things, and they are; they do different things. But the grand unifying principle underneath is that they’re actually new computer companies.
It has felt, in the industry, for the last 15 or 20 years, like there was no new opportunity to make a new computer company. It’s like, “Where are all the other computer companies?” The answer is that they are all of these new categories. The drone companies are new computer companies, the self-driving car companies are new computer companies, and it’s hardware plus software. What’s interesting is there are probably more new computer companies in the valley today than there were probably since 1982 — it’s just that the products are all these different shapes, sizes, and descriptions. It’s not a laptop, it’s a flying machine that does stuff.
I will always take a flying machine.
Exactly — pretty cool! We know of three top-end flying car startups in the valley, two of which Larry Page is funding himself. We haven’t funded one yet, but we’re starting to get the pitches. I don’t know if they’ll get them to work.
I didn’t even think to ask you about flying cars. Are the pitches that you’re getting on flying cars improving?
There’s a couple of big challenges in flying cars. [For safety reasons] I don’t want human beings driving flying cars. In fact, autonomous air is easier than autonomous ground — because you have more degrees of freedom to get out of trouble.
The big constraint on all of this stuff is batteries. Assuming no battery breakthrough, what [flying car] companies are trying to solve, fundamentally, is the power problem — which is, “How do we get the thing airborne and 50 miles without literally running out of charge?”
The single biggest X factor in the next five or 10 years for all of this stuff will be if there is some fundamental breakthrough in battery technology, then basically all of these questions of what’s possible get reopened. Because if I had 10 or 100 times the amount of power that lasts 10 or 100 times longer, then we’re building Iron Man suits. Then all kinds of things start to happen.
That breakthrough could happen at any point. I don’t think that’s necessarily a Silicon Valley startup that does that. I think that’s either a research university or probably a big industrial research lab — Honeywell, GE, Toshiba, or one of these companies. There is tons of R&D going into that right now, because it’s such an obvious problem. Any time that breakthrough happens, that’s the big one — and then all of the sudden, we really do have flying cars.
You mentioned autonomous vehicles. Lyft, another one of your portfolio companies, recently said the majority of its rides would be given in autonomous vehicles within five years. What do you expect the rollout of those vehicles is going to look like? And what’s the hold up?
This is the big reason why the Google’s self-driving car isn’t out yet. It’s a really hard challenge: you’re driving through the neighborhood and you’re going 40 miles an hour, and you turn left into an area that’s going 25 miles an hour, and there’s a crosswalk, and there’s a school, and there’s kids running around. That’s hard, and you can’t screw it up, and there are serious consequences to errors.
There are lots and lots of edge cases, because roads move. What happens when there’s construction? You’ve got the guys with the stop and the slow sign and the flags. What’s the computer supposed to do? Think of how advanced the computer has to be to understand that as compared to understanding just a stop sign. And they have to be able to deal with that because those things pop up in random places all the time.
What that basically says is it’s unlikely that there will be a mass rollout of self-driving cars all at once for all traffic scenarios. Therefore, [it’s] unlikely that there will be an instant, freeze-dried overnight Uber or Lyft competitor that is just self-driving cars.
Think about a city with suburbs. Some percentage of the trip requests you could satisfy with a self-driving car. But to start with, that will be a small percentage of the total trips. A much larger percentage of the trips will still require a human driver. Maybe a human driver for the next two years, and then maybe a human monitor in the car, ready to take over when the computer punks out.
That’s the other thing — you get the so-called Level 3 autonomy, which is basically like the way airplanes work today. Yeah, you’ve got the autopilot. Yeah, the thing can fly itself, it can even take off and land itself, but you’ve still got the pilot sitting there in the seat, awake and paying attention, and it’s a big problem if you don’t because, if you saw Sully with the geese in the engine, you still want somebody with judgment sitting there making the call.
I think most of those people in this agree with this: the deployment model is not going to be an either / or deployment model, it’s going to be a matrix or a hybrid thing. This is to say, you can see where it is today, you can see where it ultimately ends up, and everything in the middle is complex and will take time. That’s why we think Lyft and Uber and others are going to be central to this.
I think most people think self-driving cars are more likely to get deployed as a service than as a product. It’s easier to deploy them as a service, because you can just put them in the street and let them start giving people rides, as opposed to trying to convince somebody to buy a self-driving car. Especially if they’re more expensive to start because of the sensor packages.
How do you expect these kinds of services to affect cities?
There are mayors that would, for example, like to just declare their city core to [ban] human-driven cars. They want a grid of autonomous cars, golf carts, buses, trams, whatever, and it’s just a service, all electric, all autonomous.
Think about what they could do if they had that. They could take out all of the street parking. They could take out all of the parking lots. They could turn the entire downtown area into a park with these very lightweight electric vehicles. No pollution, no noise, no nothing. It would be almost like going to an airport, where you could drive and then you drop your car off and then a self-driving golf cart would take you into town. There are cities that want to do that, and not just in the US — there are cities internationally that want to do that, including in some countries where the government can order that to happen. College campuses, retirement communities, amusement parks, industrial campuses, and large office complexes in some cases are places where this stuff can get rolled out in a top-down way. I think you’ll see a hopscotching kind of thing, as opposed to sudden mass adoption.
Let’s say we do get to this fully autonomous vision of the world, and all the people who had long commutes suddenly have all of this free time. What sort of companies do you see emerging from a world where autonomous vehicles are the norm?
This is the thing that I believe is completely missed in all of the fear and uncertainty that people have around this. Everybody is a professional pessimist in business these days, including a lot of the VCs. It’s like, “Oh my God, the self-driving vehicles are going to displace 5.5 million professional driving jobs in the US over the next five years! Oh my God, the economic devastation! It’s going to be terrible for the economy.” But the other side of it is all the productivity improvements that come out the other side from people not having to literally sit there paying attention in the car.
The first order of productivity thing that happens is exactly what you said, which is that the entire amount of time spent in the car sitting there paying attention — or playing Pokémon Go when you’re supposed to be paying attention — all of that time frees up. All of the sudden commutes become viable and, possibly, attractive.
At that point, the form factor of the vehicle changes, because at that point, the driver doesn’t even need to face forward. There is no driver! You can have cars that are just rolling living rooms or bedrooms. You could literally just nap on your way to work. Or you could have an office. Or you could have entertainment pods — you could have VR. Or you can have a bus that’s a giant classroom and the kids can go on a field trip and the teacher could be up there teaching a class for two hours. All of that time, the productivity gained through that to the economy, and to all of us as individuals, is just gigantic.
The long-run thing that’s going to happen here, I think, is very enticing. The really big economic impact of cars was not the car industry — the really big economic impact was suburbs and retail and package delivery and movie theaters and motel chains and theme parks and the interstate and truck stops and all the other implications. Basically, the entire way we live today is a consequence of the invention of the automobile. Because, before that, people just never went anywhere. Therefore, everything that you travel to is a consequence of the automobile.
We talked about the role of cities earlier. One of the big issues with inequality, which is very vivid right now in the valley and in San Francisco, is can you literally afford to live someplace? It’s like this stupid Prop. 13 thing we have in California. If you’re an old homeowner you pay no property taxes, if you’re a new homeowner you get completely drilled. It presumably should be the other way around, but politically, it got corrupted and it got wired up the wrong way. Rents in San Francisco [have] doubled in the last five years, which is just complete lunacy. In Detroit, you can buy a house for $100. In the valley, it’s hard to buy a house for less than $2 million. It’s just complete lunacy. Half of the people in Detroit should move to the valley and can’t. It’s just nuts.
Then it’s like, “Okay, smart guy: internet, the phone, video conferencing, telepresence, VR, AR, collaboration, Slack, GitHub, Asana, all of these things.” Seriously? The joke in the valley is, “Help wanted. Programmer / designer wanted for state-of-the-art Silicon Valley telepresence software company making collaborative work easy across geography and time zone. Must be willing to relocate to San Francisco.” It’s just nuts.
Stage one, I think, would get transportation to be something where if an hour in the car is actually pleasurable, it’s one of the best parts of the day as opposed to one of the worst parts of your day. That would help a lot. Because then you could live south of San Jose and not kill yourself trying to commute to Redwood City. That’s one. But the other would be go after telepresence. I think telepresence, ultimately, is the answer. Because you just have computer-mediated communication, which is just exactly like you and I are in the room together. The way I think about it is, that’s the one-two punch to solve all of this real estate-related nonsense. We can’t technologically fix real estate, but we can make it less necessary.
You mentioned having the answer to this question, which you get a lot of criticism for, particularly on Twitter, about what happens to the 5.5 million jobs? What is your answer to that for how the economy gets remade?
[It’s] a two-part answer. These are things that are true and nobody knows. The American economy this year will create 24.5 million gross new jobs, and it will destroy 21 million gross jobs. This is where the headlines are misleading. They’re unintentionally misleading, but they’re misleading. When you read the job headlines it’s like, “The US economy created 300,000 jobs last month,” and you’re like, “That seems like a really small number.” It’s always reported as the net. The gross is much larger than the net.
The first-order answer to the question is the US economy, just this quarter, will destroy over 5 million jobs normally. The US economy is like the duck: it looks calm on the top and is paddling furiously underneath. We will destroy and re-create over 5 million jobs just this quarter, and we will do it again in Q1 of ‘17, and we’ll do it again in Q2 of ‘17, and we’ll do it again in Q3 of ‘17. Five and a half million jobs get reallocated over the next five years — it sounds like a lot, but it’s a drop in the bucket compared to the overall level of change happening in the economy.
The other part of the answer is dynamism. The rate of gross creation and destruction of jobs, economists call that dynamism. Would you believe that the rate of dynamism, which is to say the absolute size of growth creation and destruction of jobs in the US economy, has been rising continuously? Has the rate of change been rising continuously for the last 40 years or falling continuously?
I would guess rising.
Everybody thinks it’s rising, and the answer is it’s been falling. What the numbers show is the US economy is becoming less dynamic over time, not more dynamic. The same thing on startups. The rate of startup formation in the US has been declining for 40 years. Everybody thinks it’s been rising; it’s been declining for 40 years.
The economy tends to get stereotyped, especially in political seasons — everybody either works for a manufacturing company or flips burgers, and you get this wild, oversimplification of the economy. In reality, the economy as it exists today, is this incredibly complex, multidimensional, multifaceted thing with just huge oceans of people doing jobs that we probably have never even heard of, huge numbers of people in all of these different sectors of health care, and in-home care, and childcare, huge numbers of people in education, and infrastructure, and government jobs, and the military, and all of these different fields and professions.
It turns out the rate of dynamism in Western Europe is a fourth of what it is in the US. They’ve designed their economic system that way. They want change to happen slowly, it’s why they have all of the laws and rules and work restrictions and unions and all of this stuff. It directly translates to their economy [being unable to] adapt, therefore high unemployment. If you’re a capitalist, if you’re like me, it’s obvious that they’re going to have high youth unemployment coming out of that because they’ve wired their economy so that it can’t adapt.
Conversely, if you want to have opportunity for people to be very easily able to get work, and be able to very easily change work if they decide to go to a new field or they decide to go get a new level of skill or a new kind of college degree or if they decide they want to move to a new place, you want more change — because that creates more opportunity. To your point, the prevailing view is that must be bad, and it’s just part of the national pessimism.
It is interesting. Notwithstanding the fact that we do all of these things to screw up our own country, we are actually still good at this.
This year we saw a ton of VR and AR headsets come to the market. Does it seem inevitable to you that these will be fully mainstream technologies, or are we still waiting for the right combo of hardware and software to get there?
Yes and yes.
Any guesses on what will be the breakthrough moment for VR?
In general, in venture capital and startups, the hardest thing to call is timing. The way it gets written after the fact is, “It was obvious it was going to happen at point X, and the people who thought it would happen earlier were stupid.” When you’re in the middle of it, it’s actually not that obvious. I always like to point out Apple came out with the Newton in 1989, and boy, it seemed like it was the time. And it turned out it took another 20 years to get to the iPad.
Maybe it happens now, maybe it doesn’t. If it doesn’t happen now, what we’re trying to do is set up, at our firm, basically a willingness to make the same bet five years from now, and then if it still doesn’t work, 10 years from now, and if it doesn’t work, 15 years from now. Basically, just keep coming at it.
Is there anything about your everyday life that you think actually is going to be radically different five years from now?
One of the ideas that we play [around with] here at the firm, we call Harry Potter world. One of the things about Harry Potter is everything is moving around and alive. The photos are all going, “Hey,” and the plate goes and refills itself and brings you fresh food, and your beer mug tells you you’re drinking too much. Everything is just smart. This is my view of the Internet of Things: you’re able to infuse intelligence into everything, you’re able to put a chip in everything, you’re able to put software in everything, you’re able to connect everything online, and just everything is a lot smarter. The doorknob is a lot smarter, and the lightbulb is a lot smarter, and your wristwatch is a lot smarter. Everything starts to get really, really smart.
The other thing is this geography topic. If anything, the geography problem becomes more intense five years from now. The price of cities relative to the price of not being in cities continues to grow. I think that’s going to be a major political stress. I actually think that explains a lot of what’s happening in this election. I think 2020 is probably going to be a worse version of this election.
With even worse candidates. I think it gets worse. It’s not even so much have or have-not in terms of wealth or income, it’s have or have-not in terms of opportunity. And if I’m in rural Wisconsin or rural Ohio or something, and I’m in a coal mining town or a steel town where there’s just not going to be another steel mill, that is a very stressful way to be. “Boy, I would love to move to San Francisco, but I physically can’t because I can’t afford it.” That’s tough. That, to me, explains a lot of what’s happening in politics right now and in economics. I think that gets more intense.
However, it is also true that your kid in the village in Indonesia, or the kid of the steelworker in Cleveland, they are going to have access to more and more opportunity. Whether it’s being able to do remote programming work, or remote design work, or different kinds of educational alternatives.
I think there’s going to be more economic opportunity opening up in five years — more ways for people to be able to provide services online, or be able to sell products online, or be able to find jobs online, or be able to get skills online. I think we really start to move that out in five years. It’ll be the steady drumbeat of empowerment and opportunity coming underneath what looks like a very stressed, very angry time.
This interview has been edited and condensed for clarity.