The period from 2000-2016 was one of the best of times and worst of times for tech and the Valley (dotcom, financial crisis, Google IPO, Facebook founded, unprecedented growth, and so on), and John Hennessy — current chairman of Alphabet, also on the boards of Cisco and other organizations — was the president of Stanford University during that entire time. Given this vantage point, what are his views on Silicon Valley (will there ever be another one, and if so where?); the “Stanford model” (for transferring IP, and talent, into the world); and of course, on education (and especially access)?
Hennessy also co-founded startups, including one based on pioneering microprocessor architecture used in 99% of devices today (for which he and his collaborator won the prestigious Turing Award)… so what did it take to go from research/idea to industry/implementation? General partners Marc Andreessen and Martin Casado, who also founded startups while inside universities (Netscape, Nicira) and led them to successful exits (IPO, acquisition by VMWare), also join this episode of the a16z podcast with Sonal Chokshi to share their perspectives.
But beyond those instances, how has the overall relationship and “divide” between academia and industry shifted, especially as the tech industry itself has changed… and perhaps talent has, too? Finally, in his new book, Leading Matters, Hennessy shares some of the leadership principles he’s learned — and instilling through the Knight-Hennessy Scholars Program — offering nuanced takes on topics like humility (needs ambition), empathy (without contravening fairness and reason), and others. What does it take to build not just tech, but a successful organization?
- The importance of RISC across technology, and how it began with a startup [0:00]
- How the startup grew, and a discussion of changes in the startup space [12:29]
- Discussion of the Stanford Model [17:51]
- The importance of humility [24:20] and empathy [28:19] in leadership
- Interdisciplinary studies [34:37] and the real-world applicability of AI/ML [40:32]
- Academia-based research vs. corporate-based [42:42], and a discussion of talent in Silicon Valley [51:29]
Sonal: Hi, everyone. Welcome to the “a16z Podcast.” I am Sonal. I’m here today with a16z general partners Marc Andreessen and Martin Casado. And we’re interviewing John Hennessy, who is the current Chairman of Alphabet and was President of Stanford University from 2000 to 2016, which also happened to be one of the most interesting times for tech and the Valley.
So, in this episode we cover everything from the Silicon Valley and Stanford models, to if it’s possible to create other Silicon Valleys and, if so, where and how. And, of course, we also cover education, as well as the tech and economics of education, to what it takes to lead companies.
John has a new book out, “Leading Matters,” on principles for leadership, and he also recently launched the Knight-Hennessy Scholars program for graduate students focusing on both knowledge and leadership.
Finally, we discuss the evolving shift between academia and industry, including the role of universities, big company R&D, and the heyday of famous labs, and entrepreneurship then and now. Which, by the way, is why I asked Marc and Martin to join this episode, given their experiences going from university research to industry — Marc with Netscape, and Martin with Nicira, which came out of Stanford before being acquired by VMware.
But first, we begin with John’s own history as a start-up founder, based on pioneering the microprocessor architecture used in 99% of devices today.
Developing RISC and starting a company
Sonal: So welcome, guys.
Marc: I’d like to point out that Hennessy is also a Turing Award winner, which is unbelievably awesome.
Sonal: That’s like the Nobel Prize of computing.
Marc: It’s the Nobel Prize of computer science.
Sonal: So, you and Dave Patterson won that?
John: Yes, yes.
Sonal: Why did you guys win it?
John: Well, I think we won it because the work we did has reshaped the entire industry. Many times when you find a fundamental breakthrough, its importance may take a really long time to emerge, particularly in the hardware sector, it moves so much slower than software. And in this case, with the explosion of the mobile world and Internet of things, efficient processor architectures became really crucial. And that really changed the world. And that’s why our work has had such [a] great impact over time.
Sonal: Well, actually, break down RISC for us. Like that’s “reduced…”
John: “Instruction set computing.” The way to think about it is building a machine with a simpler vocabulary which can be executed more quickly. If you think about it in English language terms, imagine reading sentences that have giant $5-dollar words and are really hard to parse and understand, you’re constantly pulling out the dictionary. Now imagine a sentence that’s written in clear, precise English. Maybe it has a few more words, but you read it much faster. And we use that same key insight to try to build faster computers.
Sonal: So, you reduce the instruction set in order for the computers to process information faster, and therefore operate faster.
John: Cheaper, faster.
Sonal: And why was that so cutting-edge at the time? I mean, weren’t the dominant players, like, IBM and DEC?
John: IBM and DEC. And this is a time in the ’80s when, if you wanted to go talk to leaders in the computer industry and you were in Silicon Valley, the first thing you did was get on a plane and fly back east. It was a very different environment. They were building machines which were getting increasingly complicated rather than simpler. And they missed the whole importance of the microprocessor and VLSI and how we completely changed the industry.
Marc: So, RISC was invented roughly when?
John: Early 1980s.
Marc: Early 1980s. And when do you think it really tipped to become as mainstream? It’s so mainstream today, RISC processors run almost everything.
John: Yes, almost everything, except the desktop.
Marc: Except the desktop and the server.
John: And some of the servers.
Marc: Some of the servers. But, like, every smartphone, every IoT device.
John: Every smartphone, every IoT device.
Marc: Every camera.
John: You know, you probably own 100 of them that you don’t even know about.
Marc: Right, right. So it’s, like, by far the dominant architecture today.
Marc: So how long did it take from inception to kind of when the market tipped to when we knew it was going to be absolutely dominant?
John: You know, there was an early run at the mainstream market in the late 1980s and it almost flipped then. But what happened is, rather than the industry converging on one RISC architecture, they converged on three or four.
Sonal: Oh, interesting.
John: That gave Intel a real lead-up, because they didn’t have to beat any one, they had to kind of beat these little three or four.
Marc: IBM, DEC, Silicon Graphics, Sun.
John: Yeah, and they were all kind of beating up on each other, right? And so rather than getting behind one architecture, which would have made it much easier to build a software stack for it, that didn’t happen. So, there was a period where they were a lot faster, and then Intel really came back. And it probably took until the emergence of the cell phone.
Sonal: What, that long?
John: Yeah. Probably until mid ’90s. So, there was a period where it really wasn’t — it was working in the scientific computing space, but the scientific market is relatively small compared to the general-purpose market.
Marc: But even so, right? The cell phone previous to the smartphone was not that — yeah, it was a phone, it was great, it was a phone with a RISC chip, but it wasn’t a computer in the sense that we understand, right? So really the iPhone probably is the…
John: Yeah, the iPhone was really the taking-off point. Some of the earlier Nokia phones began to use the technology. But then when the iPhone came along, boom.
Marc: So, 1980-something, 1980, early ’80s, through to 2007 to really have…
John: Yeah, to really have that big effect.
Marc: Yeah. So, I just think it’s a great example of, like, these things are generational. Like, the really, really big things do take a very long time. But then when they tip, right? How many RISC chips do you think are globally today?
John: Well, 99% of the market space. So, you know, it’s much larger than the number of — now, that’s counting processor chips, right?
Marc: Right. But including embedded systems, 10 billion chips worldwide?
John: Oh, more than that. Probably 50 billion.
Martin: So, I worked for years under Pat Gelsinger at VMware, who was the GM of the 486 at Intel, and a longtime proponent of CISC. And he still maintains that CISC is the right architecture and, you know, dollar value, it’s still the predominant market or whatever. Are these different problem statements, or do you think it’s still just dying a slow death and we just haven’t got there yet?
Marc: We should know, CISC stands for “complex instruction set,” so it’s the opposite of RISC and the classic Intel model.
Martin: Right, exactly.
John: Yeah, I think you have to separate out the technical argument from, “Does it have a large, established base and, hence, a large software stack?” I think on the latter point, Pat is exactly right, it has a large software base with large, established software. But in terms of things like energy efficiency, which now it becomes the primary concern. And as we get to the end of Moore’s law, and energy efficiency becomes more important, which you carry around a lot of devices in your pocket, they’re battery-powered. The fascinating thing people don’t realize is that after the cost of the physical servers themselves, the second biggest cost in a large data center is power.
So, you care about energy efficiency even in these large data centers. And when it comes to that measure, the CISC architectures are far behind.
Sonal: One of the things that surprised me is that the chips were used in early gaming systems, the PS4 and all these.
John: Yes, that was one of the earliest breakthroughs for the RISC people in the embedded space, were games, high-end network switches, places where there was really — high-end color printers, where there was really a fair amount of performance demand, but also considerable sensitivity to price.
Sonal: So, why [were] games, like, the breakthrough then? The reason I think about this is because I think about what happened with GPUs and Nvidia, and how it then became the enabling for, like, artificial intelligence.
Sonal: More parallelized computing. So, I was just trying to figure out what the parallel was with the RISC story.
John: So, one of the reasons was the RISC architectures — MIPS was the first architecture, along with the Alpha architecture, DEC, to get to a 64-bit implementation. And in the games, as in graphics, how quickly you can move data around makes a really big difference. And so, 64-bit architectures were much better at doing that. And that accelerated their — and first with, you know, Sony PlayStation being the first big breakthrough in terms of creating a much more realistic graphics framework for games.
Sonal: By the way, is that why Nintendo is called Nintendo 64, because of the 64-bit?
John: Yeah, Nintendo 64 is called from that.
Sonal: Never connected those two dots. Back to Marc’s question though, what do you think made RISC tip? Yes, it took a long time, but how do you think — especially because in that time you founded a start-up, MIPS Technologies, to bring it to market. You could have just left it as a paper and expected the industry to adopt it.
John: I was a bit of the reluctant entrepreneur. I mean, when we wrote our papers, we thought the evidence was so convincing that industry would just pick it up.
Sonal: Yeah. I mean you said that about Nicira, I remember that.
John: That’s what we thought. And, in fact, Digital Equipment Corporation actually had a research lab out here that took some of our ideas, some of the people who worked with us, and worked on the technology, but they couldn’t sell it back east, and that’s where the headquarters of the company was. You know, IBM canceled their project several times.
So, eventually what happened was a famous early computer entrepreneur, Gordon Bell, who’d been one of the people that built Digital Equipment Corporation, came to me and said, “You know what? If you want to get this technology out, you’re going to have to go start a company.” And eventually he convinced me, although I have to say I was the technical entrepreneur that didn’t know the first thing about running a business, not the first thing.
Sonal: We have so many founders who do that. What was, like, the biggest thing when you went to start a company that was like, “Holy crap, I don’t know what I’m doing”?
John: I thought engineering should get roughly half the revenue. I didn’t realize how important salespeople really were. I thought, if you have a great product, people just buy it. So, there were a lot of things like that I didn’t realize.
Marc: So not only did people not go ahead and build the products until you did, you had to start the company and build the products. Once you had built the products, they didn’t even just buy them?
John: What we needed to do was find people who were — you know, companies are always a little reluctant to take a risk on a start-up, particularly with something like a new architecture, which really is a long commitment. So, what you had to do was find companies who felt like they needed a leg up over the other players in order to advance themselves. And that helped, we found a few players like that early on.
Sonal: It’s kind of shocking that you founded your company in 1981, and we’re talking to founders in 2018, and it’s the exact same conversation.
Marc: Yeah, you just described the exact same dynamic we see.
John: But, you know, somebody said to me once, I mean, “What’s the difference between you and somebody else who’s read about technology?” I said, “Well, the people who’ve worked on it, they see the glass as half full, not half empty.” People said to us, “Well, that’s a nice academic experiment, but you’ll never be able to make a real product out of it, it will lose all its advantages when you try to engineer the rest.” Because we’d built a university prototype, it wasn’t a commercial product.
Marc: There’s an old line, I forget who said it, but there’s an old line in the industry which is, “Everybody worries about protecting their idea. But if your idea is actually any good, you’re going to have to bludgeon people to adopt it.” Right?
Marc: This was a great example of that.
Sonal: I think it’s interesting that you said that it was, like, a prototype, like research. Do you think that’s changed today where, because of all the systems that we have available to us — you know, AWS, all these different things where you can essentially prototype in the cloud — do you think that people now have more — when they are in a university of research lab, is their stuff more immediately and more easily transferable, because it’s more pre-industry scale or production-ready?
John: Well, I think it’s probably a whole lot easier to transfer a software product than it is to transfer a hardware product. Software now, the students are incredible programmers, I mean graduate students, and you can really build something that’s pretty good shape. I mean when both Yahoo! and Google left the Stanford labs, they were pretty good pieces of software. They weren’t yet scaled up to deal with millions of users at once, but they were pretty impressive.
Sonal: Yeah. Was that true for you guys, actually? I mean, when I think about Netscape, did you have to do a lot more work based on what you…
Marc: Well, there were two things that happened. One is, when we were at Illinois, we started actually getting, like, people actually using our software, and then we ended up getting lots of customer support calls. And so we applied for an NSF grant to staff a customer support operation.
Sonal: Oh, that’s hilarious.
Marc: And the very nice people at the National Science Foundation explained to us that that was not actually the purpose of taxpayer-funded research. Which was a gift, in retrospect, and that catalyzed us in part to start a company. But then the other thing was we actually rewrote…
John: You rewrote everything.
Marc: We rewrote everything. And I actually think at Nicira, you guys did something very similar.
Martin: Yeah, yeah, yeah.
Marc: And so you do end up…
John: You end up re-engineering.
Marc: When you have paying customers, you do end up having to do a set of things that are not…
Martin: Well, I always thought that was really interesting. And so my experience was very similar to yours, which I had these academic papers, the academic community liked it, industry hated it. And I found out it was actually much easier to sell somebody something than to give it away. And I don’t know what the psychology is about it.
Sonal: That’s fascinating.
Martin: This actually happened to me twice, where I’m like, “Oh, like, the paper is done, the research is done, I’m going to do the next thing.” Now I want someone to adopt it and I have the conversation, and then they won’t put the effort in or whatever. And in both cases I ended up just selling it to them, and in the cases of companies.
And I think it does two things. One of the things it does is actually just qualifies. Because if you ask somebody for money, like, if they’re actually not interested, they’ll say “no.” And the second one, if you get a transaction to happen, you actually have some skin in the game, you actually have something behind it.
And so, I actually tell this to a lot of academics coming out of industry now, I’m like, “Listen, like, it’s hard to give something away, it’s much, much easier to sell it, especially if you want to have impact afterwards.”
Marc: What I propose, I propose the third rule from that, which is the more you charge, the more successful the implementation.
Martin: 100%. And it will set the value.
Marc: Right. Because the more painful it’s going to be for them to write it off.
Marc: And so they have to commit.
Sonal: Right. That’s your two-word mantra, is, like, “Raise prices.”
Marc: “Raise prices.” It’s another great example.
Sonal: Well, you know, Nicira — and before you guys were acquired by VMware, I remember you wrote about how you guys actually had some early adopters, but then you had, like, sort of a hump. And you talked about, too, how you had an initial fast — and then you kind of stall. And so one question I have is, like, when you get to that moment, coming out of academia and then into industry, what sort of tipped you over to sticking it out, and then figuring out how to get over that hump?
John: Well, we had a situation where we had probably expanded a little bit fast and the first CEO — remember, this is a bunch of three technical founders who didn’t know anything about really running a company. He had expanded too fast on the evidence of the first customer and, you know, we had too many people. And we were about to run out of cash. So, we had to kind of do a reset on that. We had to go through a layoff, which was a really tough situation. 120 people, you got to lay off 40 of them, you know everybody. And then the CEO asked me to get up at the Friday TGIF and give the rally call for the company — how we were still going to be a great company and this was a small hiccup on it. But I had to learn from that process and re-energize the company.
Sonal: I mean, your whole book is about leadership lessons. What was, like, the biggest leadership lesson in that moment?
John: Well, for me it was if you have a crisis and you’ve got to take a tough step, do it quickly, get it over with, and move through. Reset the clock so you can then charge ahead. And that turned out, when the financial crisis hit, you know, Stanford lost billions of dollars of its endowment, about 28% of the endowment vaporized in a six-month period. So, there was no way we could continue to spend money the way we were, we were going to have to go through that process again. I realized, you know, that’s going to lead to 5 or 10 years’ worth of small budget cuts that are going to not be very efficient, and we’re going to not be able to do anything new. So, we sat down and said, “We need to do this quick.”
Sonal: So instead of death by 1,000 cuts, you’re going to do, like, one hard stab.
John: Yeah. We did it quick. “We’ll be generous, we’ll be humane, we’ll give nice severance packages, and then we’ll restart and begin to rebuild the financial core of the university.” We had one year that was sort of a down year, and then we’re back.
Sonal: Yeah, that’s great. You started a company in the ’80s. And you started a couple of companies, in fact. And you IPO’ed only five years, I think, after starting your company. And today a lot of companies don’t IPO so quickly, so that’s one big trend shift. What are some other shifts that you’ve seen, especially since you counsel and meet a lot of entrepreneurs, between then and now?
John: I think probably one of the biggest shifts — the space of start-ups has changed dramatically. You know, when we were starting, our goal was to build a product that was more efficient, that solved some particular problem. Now, with so many software companies, the whole big question is, you know, “Will the dogs eat the dog food?” I mean, is it really going to get traction, is it going to go viral? I think that’s a very hard thing to predict ahead of time. I mean, look, I was sitting at Google when Facebook came along. Nobody foresaw how big social media — I mean, some did. Mark did, clearly. A few other people. But most of us didn’t see how big it was going to be. And that happens all the time.
Martin: Yeah, it’s interesting. Even enterprise companies now are having this type of characteristic. So, it used to be the case, you’re like, oh, a consumer company is kind of a popularity contest. You’ll have three companies that all look the same. One will get adopted, two won’t. But the enterprise was kind of core tech, and then you could actually talk to the buyer, and then you could predict somewhat whether it’s going to do well or not. Or at least whether a category is going to do well or not. But what’s happening now is, especially because developers are so influential in the enterprise, and developers are also kind of fickle and, you know, they have their own philosophies and so forth, whether or not a company is going to do well is somewhat independent of technology often, and somewhat independent of the approach they take. And it’s more like, you know, “Do they become the popular one that they use?” So, I think this is something we see across the industry.
Sonal: Yeah. People in the enterprise, it’s not just developers. You guys talk a lot about, like, departmental-level buying even across…
Martin: Yeah, yeah. Vertical SaaS, that’s right. Yeah, yeah.
Sonal: Yeah, exactly, it’s coming from the bottom up.
John: But I think even in complex organizations, universities like to have a very slow, deliberative process. But in a complex organization, all decisions are gray when they get to the top. And so you’ve got to get comfortable making decisions, making calls in that situation. And I learned that in the start-up environment. And I wouldn’t have learned — it would have taken a long time to learn in university.
Sonal: Right. Well, what do you think about — we have this view that professors that are part-time co-founders, I mean — we don’t believe that when a professor is listed as a cofounder in a company, that if they’re a part-time — that they’re actually fully committed. We need to see more skin in the game.
Martin: Having lived through this.
Sonal: Oh, did they tell you the same thing, were you trying to do this part-time thing?
Martin: No, no. I had two part-time professors and I was full-time.
Marc: Yeah, you were full-time.
Martin: Yeah, yeah.
John: Yeah, you had two part-time professors, right?
Martin: Yeah, I had two part-time professors. I mean here’s the reality — start-ups require a tremendous amount of work and effort and time, and you make real commitments to customers and teams and investors. And early on, while you may have a great idea, the investment is in you. And so there’s really a mismatch in expectations between someone giving you money, a team coming to join you, if you’re not going to be there long-term.
And so, we like to know, if we’re investing in someone — whether they come from academia or not — that they’re going to stay with the company for the duration of, kind of, the team and the investment. Now that doesn’t mean that a part-time professor doesn’t come in and help out, right? I had two, and they helped out a tremendous amount. But what we like to see is someone that is fully committed.
The Stanford Model
Sonal: What advice would you give to universities who are trying to do something like the “Stanford model”? Which, I don’t even know if we defined what the “Stanford model” is, but it’s pretty cutting-edge — and we take it for granted in the Valley that Stanford and Berkeley, for that matter, will give away more IP than they hold onto. And I used to see, when we were at Xerox PARC, a lot of university tech transfer offices. And it’s so extractive.
Sonal: And kind of nightmarish, in fact.
John: Right. Marc has the great experience at doing this, but my view of — people think of their technology licensing office as extracting blood, as opposed to being partners with their entrepreneurs. And the purpose of technology licensing, from a federal government’s viewpoint, is the university should get their technology out there. If they focused more on that, that would be great.
And be more flexible with respect to faculty. My experience is, the faculty members I know at Stanford that have gone out and started companies are better researchers, they’re better teachers. They’re all around better, because they have a wider range of experience. And most of the students we educate, they’re not going to become future academics, they’re going to go work in industry. So, a faculty member that has experience from that is actually a better teacher.
Marc: So, let me play devil’s advocate. Which is, okay, that’s all fine and good for you to say, but we only have so many professors. If they go leave and start companies, like, they may or may not come back, they’re distracted, they’re not teaching, they’re not doing research. Then aren’t we depleting the core mission of the university of doing research and education by enabling that?
John: It’s a good question. I think we’re in a tricky position right now, especially around the machine learning/AI area, where there are lots of faculty who are leaving. And that will hurt the industry in the long-term, because that means we’re eating the seed corn. I’m a great fan of faculty members who go out, commit themselves to a company for some period of time, but say clearly that their long-term goal is to go back to the university. That works well. I think if all the faculty leave, then we will have a problem long-term.
Marc: But there’s also some, presumably, benefit to being the place where people feel like they have a lot of flexibility, the place that encourages creativity, the place that encourages ventures, that presumably will play a role in attracting.
John: Right. So, you’re a young person, you’ve got multiple faculty offers. You might be interested someday in taking your technology out. Where’s the place to come? Well, it’s pretty obvious where the place to come is, and that’s a big benefit to the university in terms of recruiting people.
Marc: And so, we all the time get the delegations from, you know, various countries, various cities in the U.S., various countries outside the U.S., and sort of the question is, you know, “How do we create Silicon Valley of X?” It could be “Silicon Valley of Chicago” or it could be “Silicon Valley of France.”
Marc: Or Kazakhstan or, right, anywhere, anywhere. And I’m sure they come and see you, as well. And so what is your answer to that question?
John: First of all, build some great universities, because they are a center of innovation, and many of the ideas which build not just a single niche company, but help transform an entire industry and create an entire industry coming out of universities. Build the rest of the ecosystem out. I mean, the fact that venture was out here and people were comfortable with it, the fact that you had legal firms who knew how to work with start-ups and make that work. But risk tolerance is a big part of it. You can fail in the Valley, provided you had a reasonable strategy and a reasonable set of goals, and reboot — and it works okay. That’s not true in many parts of the world.
Marc: So maybe let me polarize the question a step further. So, the cynical view would be you can’t. You can’t create Silicon Valley anywhere else because there’s only a couple areas of technology where it’s even feasible to create a Silicon Valley, and Silicon Valley already has information technology. And then further, the things that you just described, like, they’re just too difficult to do. It’s very hard to create a new research university from scratch, it’s very hard to change the culture of the country that you’re in. That’s why there’s only going to be a handful of these places.
The optimistic view would be, “No, no, no. All these ideas are now spreading, the world is globalizing, technology is globalizing, the knowledge of how to do all these things is globalizing.” And then there’s many new areas of technology that are becoming, kind of, more amenable to this kind of flexible innovation, and many countries that, you know, want lots of entrepreneurship, and many kids worldwide who are growing up watching YouTube videos of, you know, Stanford classes on how to build a start-up, and then, you know, getting out their compiler and getting to work on writing code and starting their companies. And so, in that positive vision of the world, there’s, you know, 80 or 100 Silicon Valleys in 10 or 20 years. Where do you come out on that?
John: I don’t know that there are 80 or 100. So, it is going to happen in China, I have no doubt about it. The government is pouring enormous amounts of money into building their top half dozen research universities. The people are very entrepreneurial, there’s a lot of risk capital available. There may be some issues around liquidity and exits that are a little difficult, but they’ll work that out over time.
It surprised me that nobody in the U.S. has built a real competitor. In fact, just the opposite has happened over time. If you were to ask me 15, 20 years ago, “Will there be another Silicon Valley in the U.S.?,” I would have said, “Yes, for sure.” In fact, just the opposite has happened — the Valley’s lead has gotten bigger.
Now, we may be the victims of our own success, given land and traffic and cost of housing. We may be laying the foundation for some other Silicon Valley area, but it’s got to be a place where people want to live. And that helped bootstrap it. And so, we should be looking and thinking, “Where is that going to happen next, where is that a kind of opportunity?”
Marc: Do you think we’re at risk of strangling our own success by all of the fundamental issues around housing, transportation?
John: I think we are.
John: I think we are.
Marc: A state government that seems to hate us. A city government in San Francisco that seems to hate us…
John: Yeah, I think we are. Or hates us and loves us at the same time, right? You know, our cities and the state have such dramatic issues. And yet, you pull out the high-tech sector, I mean the state and the city of San Francisco will collapse.
So, we’ve got to think about it. And it really — you know, the younger generation moves to this area, but without that kind of suburban dream of, “Oh, I need the large house with the lawn.” I mean, they’d rather have something maybe a little smaller, not have the big yard, to have some nice parks, have some open space, and, by the way, be able to walk to three restaurants and a movie theater. And that’s a different view than the Valley grew up doing. Then you’ve got to figure out how to make the transportation network. It may be that rather than rely on government, we’ve got to get the companies to play a much bigger, forceful lead in pushing governments to do the right thing.
Sonal: I mean, one could argue that’s what’s already happened with the shuttle system.
John: Yeah, the shuttle system is that.
Sonal: As, sort of, this private tunnel.
John: It’s a patch.
Sonal: Right, it’s like a patch, exactly, into, you know, this public infrastructure. The newest trend that I’ve seen, because I am friends with a lot of 20-year-olds — they are doing a lot of cohousing arrangements, where they’re all renting big houses with like 20, 15, 10, 8 people. And our friends would never have thought of doing that when I was in grad school and undergrad. It would have been, like, two roommates at most.
John: Yeah. I think when I see a lot of the start-ups coming, I mean that’s what they’re doing. They go rent a house and squeeze more people into it than you ever thought were possible, right?
John: But it doesn’t matter because they’re working 60, 70, 80 hours a week, so…
Humility and empathy in leadership
Sonal: One question on the note that Marc was asking about the next Silicon Valley. So, the network effect of it becoming more valuable the more people that are there — the other part of the ecosystem is obviously people who are, you know, like yourselves, ex-founders, ex-salespeople, ex-marketing heads, etc. — who can then help these companies as they grow and get to the next level. That’s the biggest argument I’ve heard for why there might not ever be another Silicon Valley.
John: That’s a great argument. I remember a start-up founded at Marc’s alma mater, at University of Illinois. And — great group of people, they could hire great young engineers, because it’s one of the best engineering schools in the country, but they couldn’t get the kind of middle and upper-level management there.
Sonal: Right, exactly.
John: And so they ended up moving the company to the Valley, because there was lots of depth there.
If you look over history, I mean, Hewlett-Packard was there, then talent from Hewlett-Packard helped build Sun, talent from Intel helped build the first generation of fabless semiconductor companies, and that spread out over time. And that’s one of the great things that happens in the Valley.
Sonal: I agree. And I know this sounds so hokey, but I’m going to say it because I don’t think people really appreciate how unique it is. The generosity of mentorship. And, you know, a big theme of your book is about mentoring and molding the next generation of leaders, so let’s transition to talking about what some of [those] mentoring and molding principles are.
So, each chapter is devoted to a specific principle — humility, empathy, you know, honesty, transparency. There’s different levels of that. But they’re things that everyone says about leadership. So, I’m going to challenge you to convince me — what is the nuanced take on why humility matters? And by the way, on that one especially, I don’t know of that many humble leaders, quite frankly, that are really successful.
John: I think you can succeed while being humble if you’re also ambitious at the same time. Classical person who’s humble and ambitious is Abraham Lincoln. He’s just got to maneuver things over an extended period of time, he has to go to war, but he was a very humble person. I mean, and I think that combination — what humility does for you, is it removes the barrier to asking for help, to admitting that you’ve made a mistake. Which, for many people, that’s a fundamental thing. Look how many of our leaders won’t admit that they made a mistake, right? And won’t ask for the advice of others.
Marc: I think the challenge that leaders confront on that is, “If I show weakness, my people will start to lose faith in me.” And so what do you advise a leader who’s worried about that?
John: I think there’s a difference between being humble and being indecisive.
John: And I think it’s a question of making that decision. You know, when Abraham Lincoln finally drafted the Emancipation Proclamation, the majority of his cabinet didn’t want him to publish it, didn’t want him to release it. And yet he knew that that was the moment — that that was the time he had to do it, that he had to make that decision and move forward. And I think that kind of decisiveness is crucial.
So, you’ve got to take responsibility for making the decision and moving forward, but that doesn’t mean you shouldn’t gather all the input and be open. If you’re humble, then your staff, your team can come up and say, “You know what, Hennessy? That’s a really stupid idea. And if you do that, it’s going to come out bad.” Then you say, “Okay, well, you know, you’re probably right, I need to rethink this.” That’s fine.
Sonal: It’s kind of like our “strong opinions, weakly held.” Which feels like a very a16z value — it really seems to define the place. I love this phrase that you use in your book, “It’s not enough to understand how many people are depending on you, it’s just as important to realize how you are depending on them.” And I thought that was a very neat thing to think about — mentally inverting the org chart.
John: Yeah. I like to think of my org chart upside down. I’m the person supporting the rest of that team and serving them.
Sonal: I always think of how this plays out when it comes to things like equity, though, because you have to share the success. But, you know, quite frankly, some people do more, some people do less, some people are less fungible, others are more, and you have to take that into account. And I think that’s sort of an interesting calculus that people tend to sort of balance.
John: Well, you have to think about the value of the individuals. Everybody’s work has value, but obviously some of it is more crucial to the success of the organization than other work. So, everybody should be rewarded, but that doesn’t mean all the rewards should be equal.
Sonal: Let’s talk about empathy. Because you’re one of the pioneers, in your tenure as president, of the largest increase in financial aid ever, which allows more lower-income families to experience Stanford. And this is incredible. But you talk about how it was hard for you to actually make this happen, because empathy needs to be balanced with fairness. And that really resonated. So, tell us about how you sort of navigated that thorny issue.
John: So, we decided that one of the challenges that people who came from disadvantaged backgrounds faced is just getting through the whole process of applying to a highly selective school. You know the federal financial aid form is 23 pages long? Often you get people — they may not even speak English because they’re an immigrant family. And so that’s a major barrier. We decided we needed a very simple message, right? Your family makes less than $100,000 a year, your tuition at Stanford is $0. The next thing that happened, though, was somebody came in and said, “Well, I make $110,000 a year and my tuition is $30,000 a year. This doesn’t make any sense.”
So, we concluded you had to balance this with fairness. You had to ask the students to have some skin in the game.
John: So, we said, “Even though your tuition is $0, you have to work for the university 10 hours a week during the year, and 20 hours a week during the summer, and contribute that to your education.” And then everybody said, “Well, that’s fair, that’s reasonable.” So, balancing that was really key.
Marc: So, can I ask you the obvious follow-up question?
John: Yeah, sure. Yeah.
Marc: So, how many 18-year-olds a year — how many kids come of age to be 18 in the world each year right now?
John: Oh, a gigantic number. I don’t know, Marc.
Marc: About 100 — I don’t know.
John: Yeah, yeah. A very large number.
Marc: 100 million, some large number like that.
Marc: How many undergraduate freshman slots does Stanford have each year?
John: About 1,750 this year.
Marc: Yeah. And how many total university slots are there globally in Stanford-scale institutions, or Stanford-quality institutions, for the freshman class?
John: Well, let’s say — I mean, then you’d have to put all the elite publics in. I mean I’d say, probably there are maybe 200,000 slots in the entire United States.
Marc: So, take 100 million 18-year-olds to 200,000 slots. You know, the obvious question, right? Which is, like, it’s fantastic, obviously, what Stanford is doing for the kids who then end up in Stanford, but most kids don’t, most kids don’t end up in anything resembling a Stanford-quality education.
John: I came to the view that the university had a moral imperative to increase the size of the student body. Now, there’s a limit [to] how far you can increase it before you change the quality of the experience, right? We house all our students on campus, things like that. But we could certainly do more. And the provost and I made an argument.
So, in the end what happened — the financial crisis came along, we had to put that on the back burner. But then it came back later, and we’ve engaged in the gigantic expansion of undergraduate housing so we can house students on campus.
Marc: This does sound a little bit like the Director of the Globe Theatre in, you know, 1550 or whatever, kind of, saying, “More people should get exposed to Shakespeare’s plays. And so therefore we should build a balcony, right? And we should, you know, double the number of people who can come to London and see the play.” But, like, most people in the world are never going to be able to get to London and see the play. Like at some point isn’t the right answer to invent television?
John: No, the right answer is to change the way we educate people. I mean, I think if you were to make an accusation against higher education, it’s that they haven’t really done very much to bend the cost curve. And part of this is understanding what it means to bend the cost curve. Think about Vivaldi writing “Four Seasons” and having four musicians play the “Four Seasons,” right? It takes 23 minutes. It took 23 minutes in, whatever it was — 1790s, it takes 23 minutes today. What’s the big difference? Those musicians get paid a lot more today than they got paid then. So, actually, there has been no productivity gain in the presentation of the “Four Seasons” piece, right? And universities are somewhat in that, it’s still a craft to some extent.
Now, that has to change. That has to change. We’ve got to figure out how to leverage technology in an appropriate fashion to get the cost of education down. Otherwise, it’s simply going to become more and more expensive for American families, we’re going to load up with student debts going through the roof. And part of the reason it’s going through the roof is families are less able to save than they used to be, and so we see student debt going up.
Marc: The one form of debt that is not discharged through bankruptcy?
John: Yeah, correct. But it’s also — look at the default rates. Now, part of this is the for-profit industry, unfortunately, in the higher education space doesn’t deliver a lot of value. So you end up with lots of students who are not able to use their education to get ahead. We’ve got to figure out how to deliver a high-quality education. Not decrease the quality in order to just get the cost down, but hold the quality up while reducing the cost. And the only way I know how to do that is by using technology.
Marc: Have you read Bryan Caplan’s book, “The Case Against Education?”
John: No, I haven’t read it.
Marc: It’s probably not a common book on the Stanford campus. Although he is a tenured professor of economics, and so he is an instance of what he is talking about. And so, I’ll just focus on one aspect of the book that he talks about. The sheepskin effect, if I recall correctly, is basically if you take somebody — if you take an undergrad who’s completed seven out of eight of their semesters, right? So, they’re three and a half years into their program and they drop out. You might think that they would get seven-eighths of the income in their first job as somebody who does all four years, and it turns out that’s not the case at all.
Marc: Which then, basically, means that the value of that four-year education program is primarily in the signal of the diploma, as compared to the actual education. I think statistically, I think, this is in the numbers. So anyway, you might interpret that in different ways. I’d be curious how you would interpret that.
John: I think there’s some truth to this observation. And I think one way of interpreting it is that the drive and the determination to finish that degree is actually the key signal that employers are looking for, not just what courses you took.
Now, I should say, post-bachelor’s degree, this is changing dramatically. But if you think about other kinds of post-bachelor degrees, we’re moving very quickly towards a certification type model, where you take a course, or a sequence of courses, right? So, you go and take the sequence of courses on cryptography and blockchain, and you become an expert on that. And by demonstrating that you’ve mastered three, four, five courses, then that all of a sudden becomes the key to getting a new job opportunity. I think we’re going to see more and more of that as we go along.
Marc: So that’s, like, an alternative to a master’s degree?
John: Yeah, it’s an alternative to a master’s degree, you actually have to demonstrate mastery of the material. I think that’s the key thing, and that’s what an employer wants to know, right?
Sonal: It’s like Udacity with the Nanodegrees to some extent, too.
John: Yeah, it is like that.
Interdisciplinary studies and degrees
Sonal: Actually, on this very note, like, I would love your take on the interdisciplinary side of things. Because to me, the one unique thing that universities can do that a lot of these other institutions cannot do is break down barriers between disciplines. And you guys have tried experiments, or legitimate degrees, like symbolic systems, etc., that cross across, you know, multiple disciplines. But I’ve yet to see examples of true successes in multidisciplinary degrees or entities. Like, maybe Xerox PARC would be the best example, but I really can’t think of any others.
John: Happens a lot more at the graduate level and the research level. Partly because I don’t believe that multidisciplinary or interdisciplinary things are a substitute for some deep domain knowledge. I’m a firm believer that you start with deep domain knowledge, and then you build on top of that.
You know, one of the challenges with these small courses that certify you in an area — those work well for a professional. They’ve already got an undergraduate degree, there’s a clear connection between the value of the education program and how they’ll be rewarded.
Take an undergraduate coming in without some of the advantages that you’d have if you want to an elite high school. They’re not going to thrive very well in that kind of online setting, where they don’t see how that directly translates to getting a job at Facebook, for example, all right? They’ve got a long way to go before they’re there. So, they need a rather different educational system than somebody who’s already got their degree. They see, “If I take this course, I’ll get this new opportunity.”
Martin: I also think computer science is a little bit unique in this, in that, you know, so we call it a science, but, I mean, ultimately it’s an engineering discipline. And while there is, like, pure computer science, almost all of it is applied. And so, when I did my Ph.D at Stanford, we had people that would work in graphics, and they worked very, very closely with, you know, computational physics, for example, solving very real problems. Same thing with biology, right? One of my best friends, I mean, he did some really core work in DNA sequencing. And if you squinted at him one way, he looked like a biologist. If you squinted another way, he looked like a computer scientist.
The thing that I love about computer science, and I’ve always loved, is if we wrote a program that solved grand unified field theory, physics would go away as a discipline, and we’d be like, “Okay, that was more application. Let’s go on to biology,” right? So, in some ways it doesn’t exist without, like, the other disciplines, in another way it really is kind of this meta-discipline. And so I do think it’s pretty unique in that way.
John: It is unique and it is this meta-discipline, I mean, I think. And it’s become the new meta-discipline that everybody needs to learn.
John: Because algorithmic thinking is such a fundamental thing about how the world operates these days.
Sonal: Right. Like math, reading.
John: Like math, right? It’s just like that.
Sonal: You know, computational literacy should be just one other form of that. I was thinking, there was this debate with Vitalik Buterin — who’s, like, the inventor of Ethereum — and this professor, who’s a former editee of mine. And the debate they were having was whether there should be a dedicated degree for blockchain. So, the professor was saying, “We don’t need this, you should have fundamental basic science, and that’s good enough.” And Vitalik’s point was, “Well, actually, this is a really interdisciplinary, multidisciplinary, unique case where you’re layering economics and computer science and lots of other — finance and lots of other things, in a very intersected way.”
So, I thought that was fascinating, that there was a sort of tug of war. And this, to me, is the wave of the future. Like, I could even see the blockchain as a laboratory for people learning on their own in the future, especially if you think about what Marc mentioned earlier, about all these kids coming online around the world who don’t have access to these universities locally and are learning from YouTube. I could see programmers in my parents’ village in India becoming people who become such experts in this world. I mean, you’ve been the president of a university for 16 years that I greatly respect, but I wonder if it means that maybe the university model might have to really evolve in a different direction.
John: Well, I think there’s about to be a great test of this, because [of] the wide applicability of machine learning to all kinds of problems, all kinds of problems. I mean, you know, you should see breakthroughs in biology, in chemistry, in astrophysics, coming out of various forms of machine learning. So, all of a sudden it becomes this tool that is applicable to a whole range of things and is changing those fields. What do the scientists, the people who think of themselves as astrophysicists or as organic chemists — how much do they need to understand, how do they deploy this technology?
And this is a big gap right now, because the senior people in the field — it’s highly unlikely that most of them are going to take a year or two out and go back and learn a bunch of things about computer science and statistics and machine learning ideas. We’re really going to have to build a new breed of people who, kind of, fill up this interstitial space and become the key innovators in the disciplines.
Sonal: Well, I would argue that it needs to be more applied. We have an executive briefing center with a lot of big companies coming in, and the number one challenge they have when it comes to ML and AI is production-ready, industry-applicable machine learning. It’s actually, like, what’s happening in academia is not at all connected to what they need to actually do.
Martin: Yeah, it’s not only that. Which is as you move to AI and ML, more and more of the value is the data.
Martin: And more and more it’s, you know, almost serendipitous understanding of the data prior to manipulating it, right? It’s almost impossible to remove the context of the domain understanding from data. From programs, maybe. From data, almost certainly not. Which is why we’re seeing such, kind of, a confluence of CS, statistic and data understanding, and domain expertise.
Sonal: Right. It also goes to your views about the end of theory. You should share that.
Martin: Or not.
John: So, you’ve got to look at that. You’ve also got to look at how and who establishes ground truth in these. I may have an AI program that can recognize some medical condition, but who decides whether or not it’s right on the basis of that? ML is the ultimate “garbage in, garbage out” technology. Because if the data isn’t good and properly validated and the learning process isn’t — you’re going to get assumptions and outputs that are ridiculous.
Martin: So, this is something that we have to deal with a lot, you know, in venture capital, which is a number of constituencies and entrepreneurs actually view AI or ML as almost, like, the end of theory. So, it’s almost like, I don’t have to know what I’m doing — the AI and ML will figure it out for me. So, like, they’ll come in and they’ll say, “Listen, there’s all of this data in enterprise X or whatever, we’re going to apply AI and ML, and then the net result is going to be value.” And, like, “Well, what’s that value?” “I don’t know, the AI and ML is going to tell you and it’s going to be valuable because we’re going to apply this.”
And so, like, it’s a very important toolset, but I think you have to understand the domain. To your point, “garbage in, garbage out.” You have to have some way of getting the expertise or whatever in the prior to get the answer. It’s not like this has become the end of theory, and we don’t have to know what we’re doing anymore and we’re going to get valuable results.
John: And the space where that works, sort of unsupervised learning, is such a small part of the giant ML space. It’s relatively small. And most of its interesting applications are in the natural science world, not in real-world applications.
Martin: Where there is actually a truth and way to test the truth, right?
Martin: And so for me the most difficult thing about moving from academia to industry was that in academia, you look at a problem domain and, kind of, your job is to think very, very clearly and, like, pull out, like, these kind of, you know, global truths, and they have to be very elegant. And very rarely do you write a paper where you’re like, “Here’s this problem domain and here’s, like, my litany of 50 fixes. And read through every one of my heuristics and, oh, look how elegant it is.” Right? It’s almost the exact opposite. What you learn about starting a company is, it’s actually the opposite — which is, almost every solution is dealing with a heavy tail of complexity, and it’s a bunch of patches and the real world and everything else.
And so mentally you’ve got to go from, “I’m going to look at a problem space and extract elegance,” to, you know, “I’m going to deal with all of this complexity and master it.” But where I did find this energy very useful is, a lot of leadership is thinking simply. And so if you start a company and you can extract that elegance, you can use that to really lead a company, and you can convince a customer, and you can talk to an investor — because you’ve really distilled what’s important about it. But you can’t let that constrain you, because ultimately you have to build something that solves a real problem, and the universe is a messy, messy place.
And so, if you can get beyond that kind of ability to have everything be incredibly elegant, I think you can have both the leadership and kind of, like, the actual complexity.
Sonal: That’s fascinating.
Academia vs. the corporate world
John: Yeah, no, I think you’re absolutely right. I think in the academic world, we like things that really look elegant. And we often actually delay publishing a paper or getting a result out there until we get it all gelled just right, right? That doesn’t work in a start-up company.
I think the one thing that is common is — focus really does help in both cases, right? I mean you’re relentless in a start-up company, you’ve got to focus, you’ve got to drive, you’ve got to decide what’s peripheral, and [that] you’re not going to do now. And the same thing is true in academia. If you want to do really great work, you need to focus, you need to kind of — somebody once told me, they gave me some good advice. They said, “You know, you ought to be working on three or four things, but you ought to have one or two of them that are really important, where you’re really putting your energy. And these others are your backup in case those really great things don’t work, and you don’t get tenure for those.” And that was good advice about how to think about a research career, but it doesn’t work in a company. You’ve got to get rid of those things that are not the home runs.
Sonal: When I think of examples like Xerox PARC, which honestly, despite the mythology, they actually did put a lot of repeat successes out into the world. It wasn’t that they had, like, a carte blanche to just invent whatever they wanted. They had a very specific mission, and they invented towards that mission. When you talk about the differences between academia and industry, academia is about ideas and industry is about implementation. And you believe that there’s an interface that VCs and others carry across those two. Do you think, though, that that’s sort of a false divide in some ways? So, it was actually not just ideas versus implementation, it was ideas in practice, in industry settings — because it was for a corporate research lab. So, I just wonder how you’re thinking about this — was then and now, and how it’s evolved.
John: So, I think there was a time when IBM Research, Xerox PARC, and Bell Labs were the great giants.
John: What they had — they were not devoid of application and things. I mean the work on the transistor was really begun to solve a fundamental problem that a telephone switch built out of tubes. What they did have was, they had the advantage of a long investment horizon. It’s harder to find that in industry nowadays. It’s harder to find that patience. Partly because of the observation that, if you discover something really big, lots of people have to eventually benefit from it, right? Bell Labs and AT&T were not the major beneficiaries of the discovery of the transistor. Xerox was not the major beneficiary of the discovery of modern personal computing, right? That’s why universities are the ideal place to do this kind of work, because society benefits. Universities do technology transfer in a very natural way. It’s called graduation.
Martin: Marc and I are both dying to jump in. I think historically that’s certainly been the case. One could make an argument that this is shifting, and some of the most fundamental research contributions are actually happening in industry today. And not only that, that — you know, the academic system has actually moved towards short-termism, especially in incremental publishing. Like, I even feel like I’ve seen that dynamic shift in the last 15 years in just my, kind of, professional career. Where I would say Google and Microsoft are doing some of the more, you know, innovative fundamental contributions. And then I still sit on program committees — it’s interesting, they publish a paper, I’m in the PC committee, and then all of the professors are basically trying to do incremental work on top of Google’s work, right? So, are we seeing, like, an imbalance lately, or is this a momentary thing?
John: No, I think you’re right, I think there is a bit of a shift occurring here. It’s driven by not only the amount of resources that are available at Google, Facebook, Microsoft. It’s driven by data, and it’s driven by computational resources that are available in those companies that are much larger than is available to a typical university setting. So, I think we’re seeing a growth of, kind of, [a] new research environment in industry that’s quite a bit different than the old environment, and may be a harbinger of how things get invented in the future.
Marc: I’m kind of the skunk on this topic. So, I think the reason — the skunk at the garden party. So, I think the reason — I mean they did great work, Xerox PARC, Bell Labs, IBM Research. But here’s the thing, like, it’s always those three examples. They’re basically like they were rounding errors on everything. Like there weren’t 10, there weren’t 20, there weren’t 100, there were 3 or 4. And there were two preconditions for them. One is they all were offshoots of monopolies.
John: They were all offshoots of monopolies, you’re exactly right.
Marc: So your point on long-term thinking, the reason they had long-term thinking is because monopolies…
John: They could afford it.
Marc: By definition, all monopolies have this long-term thinking. Right?
John: They all were offshoots of monopolies, that’s an important insight.
Sonal: I never thought about that.
Marc: And arguably from a corporate, like, investment of capital standpoint, they were worth it just for the marketing value. Right? Of being able to demonstrate that they weren’t just, you know, sitting on their rear ends in the corporate office. And then, the other precondition was they were all pre-1975, 1980 — they were all pre-venture capture capital.
John: Yeah, yeah, yeah.
Marc: Right? And so when the monopolies cracked, and then venture capital pulled the talent out, like, that was basically it. And the downside case would be, that removed this kind of long-term commercial research. But the upside case would be, that led to what I would argue as just an explosion of R&D at far greater scale, right? Across the corporate landscape than ever existed in the 1960s, 1970s. And so, we’ve kind of mythologized these things. But they were tiny, they were tiny relative to what’s happening today.
John: So, there’s a lot more happening today, to the extent to which I can’t imagine a start-up, kind of, thinking about the length and the amount of money that was invested to build the Alto. I mean that’s a major, major undertaking by any measure. On the other hand, I think you’re right. There are now a much larger number of players doing interesting things. And in the software-driven world that we live in, the cost of experimentation and development is not the same amount in terms of capital.
Sonal: Right, you don’t need that amount of capital anymore.
Marc: Well, and I agree with all that, but I’d also say — even with what you just said, even that — like, yes, the Alto, but, like, also, like, look, Apple made the iPhone, right? Like, that was, what, a $150 million-dollar project? Like, you know, over the course of its — like, they were able to do that. Google, as you’re well aware, like, basically invented the self-driving car. Those are on par with the Alto.
John: I mean, if you look at the self-driving car, the tipping point was when the DARPA Grand Challenge was won. And that really was a key tipping point, because it demonstrated the technology was considerable. Considering that the previous contest before that, the car had not driven very far at all, and all of a sudden boom. So, there’s a tipping point in that. And when you see those tipping points, that probably is the time when you say, “Let’s move it from an academic setting that’s kind of more freewheeling, and operates more incrementally, to a different environment.”
Sonal: Well, one could argue, in that example, that DARPA was a VC.
John: They were, they were.
Sonal: Because they were putting up the prize money and everyone was competing in the start-ups, i.e. the individual people trying to meet the challenge, etc.
Marc: But then Google has now put another, what, dozen years?
John: Oh, yeah.
Marc: And a lot more money behind it. And I think that, you know, the self-driving car, the Waymo project, is as glorious a success as anything that ever came out of Bell Labs or ever came out of IBM Research.
John: Yeah. I mean I think the gap between, “Okay, we can drive on this desert road in a fairly constrained environment,” to, “I can drive in a city environment, with lots of people who do wrong things,” including look at their cell phone while they’re driving, is a much harder environment to do it in.
Martin: I think another interesting example is a company that you sit on the board of, which is Cisco Systems. Which is, Cisco has long had this stated goal of no internal research. However, they really made modern networking in, like, no small sense of the word, right? <inaudible> in the universities. But when you actually go in Cisco and see what they’re actually doing, you’re like, “Wow, they understand the real problems, they understand the customer.” Like, so I think, like, actually they’ve taken a stance against research there, yet they’ve done a tremendous amount of innovation. However, they have done a good job collaborating. So, it’s a little bit of a spectrum in Cisco.
John: Yeah. And they’ve had a model for many years of, “We buy interesting companies, and we bring technology in that way, and then we grow it and use the rest of our ability to really make it successful.” So, it’s a different innovation model, as opposed to one that’s more organic.
Sonal: I mean why wouldn’t you? Because then you’re essentially betting on 1,000 experiments and figuring out which one is a winner, instead of trying to internally, captively figure it out yourself. Like, I just can’t see any alternative to that.
John: Well, the only downside is that once that company gets far enough along, that little start-up, that it’s got some great technology — there are often more than one company that’s bidding for it. Then you could actually lose out in that setting.
Sonal: Right, right. You don’t want to lose that. Right, right.
Marc: God bless America.
John: Good for the entrepreneurs.
Martin: I mean it’s actually a really interesting point. The thing I’ve been most impressed with Cisco over the years is, they really, I think, are probably the top company in making those acquisitions successful and doing spin-ins. I mean there are very, very few companies you can put in that have been so successful in acquisitions. So, it’s basically a core competency.
John: Yeah, it has been a core competency.
Sonal: “Spin-in” as in?
Martin: So “spin-in” is, they’ll take an internal team, they will take them out of the company, they’ll help fund them, and then they’ll bring them back into the company once the product…
Martin: Yeah, yeah, yeah.
Sonal: I didn’t realize that.
Martin: It really has kept them relevant, where many companies have actually not — you know, of the same vintage are no longer.
John: It has, and it’s injected new technology and new products into the space and things.
New talent in Silicon Valley
Sonal: Right. Last question. What do you think has changed with talent — like the whole talent landscape, over the last 30 years? Because we’ve talked a lot about tech trends changing, the availability of capital, the ecosystem, industry, collaboration, academia, etc. But the people themselves in this ecosystem, what is the biggest change that you’ve seen, or are they the same?
John: So, one of the changes I’ve seen recently, that really has me delighted, is to see the number of young women going into computer science. What’s funny about it is, computer science in the ’80s was one of the…
Sonal: Yeah, it was very female-dominated.
John: It was, there were a lot of women in it. And then it got wiped out with the growth of the field and the number of males grew. And now we’ve seen a resurgence — I think, begun by a group of very energetic women that started to build support groups and things like that. And then we got over the critical mass, you got enough women in the discipline that they didn’t feel isolated anymore, and that’s really great to see. The number of opportunities in the software space are so large, we need to bring as much talent in.
The other thing that’s been remarkable for me is, I thought 10 years ago that computer science was going to become second to the biological sciences, in terms of getting the best students, and that everybody — the really best students were going to go do the biological, biotech, things like this. Well, that’s changed. And now computer science gets the very best students in any of these fields. I mean, I’ve seen freshmen that know more mathematics than I knew when I was a senior getting my college degree now. That’s remarkable, and they’re going to build great things, I believe.
Sonal: And those are merging, actually. Like a lot of the comp. sci. folks are now starting bio start-ups.
John: Yes, they are, they are. And bringing computer science knowledge to the bio space.
Sonal: Yeah. Do you guys have thoughts on any big talent shift you’ve seen?
Marc: I think the big one I see, that I think is probably under-remarked on, is engineers are so much more productive today, especially in software, than they were 20, 30 years ago. The tools are so much more sophisticated and powerful than all the infrastructure technologies. And then all the — the ability to learn. Kind of to your point on the undergrads, but, like, the ability to go online and learn. Right? It’s like, I’m an engineer and I don’t know how to do something, like, I don’t have…
Sonal: Stack Overflow it.
Marc: Boom, boom, boom. I know it in 10 seconds.
John: Yeah. You may actually be able to find the piece of code, because code sharing has become such a big part of what we do, reuse.
Sonal: Right, right, right. I mean MIT was a pioneer there with the MIT license and open source. What’s your biggest shift?
Martin: I think the biggest shift that maybe has impacted me is, like, I just remember the transition where pretty much everybody was in computer science for the love of it, because it wasn’t really clear where the industry was going. Often they were doing it to get something else done — to basically the professionalization of an industry. Meaning, it is a real discipline, people are in it to make money, people are in it for a future. Which is not a bad thing, it’s just required. And I think it’s actually quite good, because it requires us to really think about what it is, what people do.
And so, kind of on the negative spectrum there’s a — you know, people are lot more mercenary about it than they were before. And on the positive end, I do think we have a lot of framing around it — what does it mean to have a workforce in computer science that will come and go, and to handle that in a way. But for me, it’s been a very, very stark difference to people that I used to work with 20 years ago, when we were literally all there, you know, for the love of solving these great problems — to now it’s like, you know, this is your job.
Sonal: I think my favorite thing is seeing the intersection of art and humanities and code. And people used to keep them as separate in their heads, and there’s a whole new wave of talent that’s native in both. And that’s really exciting to me because, you know, art is code, code is art. So, to me that’s, like, the biggest, or most exciting, talent shift.
Well, John, just want to say thank you for joining the “a16z Podcast.”
John: Thank you, delighted to be here.
Martin: Thank you very much.
Marc: Cool. Thank you, John.
Find them wherever you listen to podcasts.
The views expressed here are those of the individual AH Capital Management, L.L.C. (“a16z”) personnel quoted and are not the views of a16z or its affiliates. Certain information contained in here has been obtained from third-party sources, including from portfolio companies of funds managed by a16z. While taken from sources believed to be reliable, a16z has not independently verified such information and makes no representations about the enduring accuracy of the information or its appropriateness for a given situation.
This content is provided for informational purposes only, and should not be relied upon as legal, business, investment, or tax advice. You should consult your own advisers as to those matters. References to any securities or digital assets are for illustrative purposes only, and do not constitute an investment recommendation or offer to provide investment advisory services. Furthermore, this content is not directed at nor intended for use by any investors or prospective investors, and may not under any circumstances be relied upon when making a decision to invest in any fund managed by a16z. (An offering to invest in an a16z fund will be made only by the private placement memorandum, subscription agreement, and other relevant documentation of any such fund and should be read in their entirety.) Any investments or portfolio companies mentioned, referred to, or described are not representative of all investments in vehicles managed by a16z, and there can be no assurance that the investments will be profitable or that other investments made in the future will have similar characteristics or results. A list of investments made by funds managed by Andreessen Horowitz (excluding investments for which the issuer has not provided permission for a16z to disclose publicly as well as unannounced investments in publicly traded digital assets) is available at https://a16z.com/investments/.
Charts and graphs provided within are for informational purposes solely and should not be relied upon when making any investment decision. Past performance is not indicative of future results. The content speaks only as of the date indicated. Any projections, estimates, forecasts, targets, prospects, and/or opinions expressed in these materials are subject to change without notice and may differ or be contrary to opinions expressed by others. Please see https://a16z.com/disclosures for additional important information.