Transcript: Andreessen on IPOs, Mayer, policy

July 16, 2012: 11:50 PM ET

Marc Andreessen, co-founder and general partner of the venture capital firm Andreessen Horowitz, joined Fortune's managing editor Andy Serwer for the opening discussion of this year's Fortune Brainstorm Tech conference in Aspen, CO. He talked about Marissa Mayer's new role, IPOs, and consumer electronics. 

Below is an unedited transcript.

Marc Andreessen

Marc Andreessen

ANDY SERWER:  Welcome, Marc Andreessen.

MARC ANDREESSEN:  Thank you.

ANDY SERWER:  How are you doing?

MARC ANDREESSEN:  I'm doing great.

ANDY SERWER:  Have some water.

MARC ANDREESSEN:  I'm going to get me some high-altitude water?

ANDY SERWER:  Yeah, we need it.  I don't know if that's the oxygenated water or not.  I can't even say the word.  But anyway, we need it bad.

So, I just want to let you all know that Marc did some sacrificing to come here.  He passed up, is this correct, a premier?

MARC ANDREESSEN:  Actually, that is true.

ANDY SERWER:  Premier of the new Batman movie.

MARC ANDREESSEN:  Yeah.

MORE: Andreessen: America's consumer electronics comeback

ANDY SERWER:  To be here today, so thank you.  Let's give him a hand for that.  (Applause.)  Very nice, I appreciate that.

MARC ANDREESSEN:  So, instead, I'm going to dress up like Batman and run around Aspen tonight solving crimes.  So, if you see --

ANDY SERWER:  Wow, so don't commit any crimes.

MARC ANDREESSEN:  If you see motion in the shadows, you know.

ANDY SERWER:  Yeah, be careful.  Chris Nolan really is kind of a genius, isn't he?

MARC ANDREESSEN:  He's fantastic.

ANDY SERWER:  I mean, he really is.  And I don't know whether this is the last Christian Bale movie or not.

MARC ANDREESSEN:  It is.

ANDY SERWER:  It is?  Okay.  So, we're not going to talk about Batman today, okay?

Why don't we talk about the news?  Does everyone know who the new CEO of Yahoo (YHOO) is?  (Applause.)  Yeah, okay.  So, the new CEO of Yahoo! is going to be announced -- Patty Sellers, Fortune's Patty Sellers just talked to Marissa Mayer, who is going to be the new CEO of Yahoo!.  Patty just talked to Marissa on the phone, and the announcement will be at 4:00 Pacific Time today, and Marissa will be starting tomorrow.  Unfortunately, that means she's not going to be coming to Brainstorm, which is too bad, but she told Patty, quote, "There's lots of work to do."  (Laughter.)  I thought that was a pretty good quote.

MARC ANDREESSEN:  It's a good first step.

ANDY SERWER:  Yeah, absolutely.  So, what do you think?

MARC ANDREESSEN:  Number one, I'm super happy for Marissa.  You know, I think it's fantastic for her to be able to step up and take on a CEO job this big.  So, it's great for that.  It's great, you know, for the Valley to have product-centric -- you know, I'm a big Ross Levinsohn fan, and I think he's going to do very well no matter where he goes, but it's a big statement on Yahoo!'s part to go with a product-centric CEO technologist, because if they had wanted to go with a sales-centric or content-centric CEO, Ross was the obvious choice.  So, I think they decided to -- in fact, they announced -- in the press release, they say that they intend to make Yahoo! a product innovator again.

ANDY SERWER:  Really?

MARC ANDREESSEN:  So, it's a very big sort of commitment on the board's part to pursue a product-centric strategy.

ANDY SERWER:  What sort of products could or should Yahoo! pursue?

MARC ANDREESSEN:  Give Marissa a little bit of time to think about that maybe.

You know, it's a challenge.  There have been -- you know, on the one hand, there have been very few Web turnarounds.  There just haven't been very many.  I don't know if there have really been any big ones.  On the other hand, you know, I'd like to say it's hard to overestimate how screwed Apple was in 1997.  And then to see the huge turnaround under Steve proves that tech companies can, in fact, be turned around.  In fact, they can be turned around spectacularly, which I think was not actually known before Apple.

You know, the problem is there aren't a lot of Steve Jobs characters running around.  So, it's a big job to do what Marissa's doing to step into a company like that and to take on a challenge like that.

ANDY SERWER:  Right.  Yeah, there's a lot to talk about, Marc, obviously, software eats the world and Andreessen-Horowitz's strategy, and we're going to get to that, but I wanted to just drill down very quickly to I think your newest investment because it's emblematic of some of the things we'll be talking about, and that is GitHub.

MARC ANDREESSEN:  GitHub.

ANDY SERWER:  Can you say it like that?  Git?

MARC ANDREESSEN:  GitHub.

ANDY SERWER:  Because it's G-I-T?

MARC ANDREESSEN:  G-I-T.

ANDY SERWER:  Okay.  And I'm not sure how many people know what GitHub is.  I didn't until very recently, but Andreessen-Horowitz just invested $100 million in this company, which have been rather stealth in San Francisco until now.  So, why don't you talk a little bit about what they do, because it's not a straight shot, at least for me, and what you were thinking when you decided to invest.

MARC ANDREESSEN:  Yeah.  So, it's a fascinating company in a lot of different ways.  So, what it does, it's actually two things in one.  It's actually kind of -- it depends which angle you look at it from.

So, what it is, we call it "software eats software development."  So, it's an online platform, it's a software as a service company that provides a hosted environment for programmers to develop software.  And the heart of it is what's called a source code control system, which is one of the main tools programmers use.  So, it tracks all your software and all your updates and lets people work together on software.

It's the main software repository online for all the Open Source software.  So, all the Open Source software, a very large amount of the Open Source software lives in GitHub.  And then there's a huge advantage to putting your software into GitHub because the way people develop software these days is they tend to use a lot of Open Source software, even in commercial software projects.

So, if your software lives in GitHub, then you can automatically include all the Open Source software and the whole thing.  It's just the best way to develop software these days.  So, it's a very effective tool for that.

It's a new generation product into a business today that's a multi-billion-dollar business.  So, there's companies like Rational that are deriving billions of dollars of revenue for these kinds of things, so that's a big business.  That's one side of it.

The other side of it is it's a social network for coders.  And it's a social network in the sense of you have an account on GitHub like you would on LinkedIn or on Facebook, but it's not just things that you say or things that you post, it's actually your account on GitHub correlates to all the work that you do as a programmer developing software.

So, what happens is programmers are building their reputations on GitHub.  So, as a consequence, if you want to hire programmers, increasingly, you're going to go to GitHub because that's where the programmers are going to be, and that's where you're actually going to be able to see their work.  So, it's almost like a verticalized LinkedIn for coders, but with a tremendous opportunity to basically be the place where coders live.  And I think that's a big deal.

The other really cool thing about the company is how it got built.  It was founded by four guys who actually -- emblematic of something happening in the Valley right now, which is they came to the Valley -- three of the guys came to the Valley in 2000, 2001, right as everything was about to fall apart.  They went to work for some truly horrible companies.  They had terrible experiences.  So, they decided to start their own company several years ago to be the company that they would actually want to work for.

So, number one, they took no outside capital, completely bootstrapped.  Number two, very unique and special culture that really is just a fantastic place for programmers to work, extremely high productivity, huge vision.  They have very big dreams.  But they have taken no outside capital, literally zero outside investment capital.

So, we did a $100-million Series A investment, which certainly is a record for us, and I think that might be the record of all time for venture capital.

MORE: Follow here for all #FortuneTech coverage.

ANDY SERWER:  Right.  Fortune's Dan Primak pointed out, sort of related to that last point you made, that A16Z is maybe known for either ponying up a higher valuation or putting more capital in.  Is that kind of a conscious strategy that you guys have?

MARC ANDREESSEN:  Yeah.  So, I'd argue with the first one, which is I don't think we actually go in at very high valuations, especially relative to some of our competitors.

Notably, by the way, we didn't do any sort of expensive valuation rounds, what you'd consider to sort of be north of $200 million.  Between basically July of last year and July this year.  So, we've basically been out of the market for all those things for the last 12 months.  And there are a variety of reasons for that, but it's just the reputation is a little bit overblown on that front.  The big one that we've had is GitHub.  We think we've paid a very good price both for the company and for us.

The more interesting thing for me is the size, which is, you know, $100 million is a big investment for a private company.  The main thing that we think is happening is that many of the best very special new companies are being built as private companies for a much longer period of time.  They're getting to be much larger as private companies and a longer period of time than used to be the case.

ANDY SERWER:  Why is that?

MARC ANDREESSEN:  It's because it's so hard and unfriendly to be a public company, right?  So, basically, it was very easy to be a public company in the '90s.  Then the dot-com crash hits, then Enron and WorldCom hit.  Then there's this huge amount of retaliation against public companies in the form of Sarbanes-Oxley and RegFD and ISS and all these sort of bizarre governance things that have all added up to make it just be incredibly difficult to be public today.  So, the logical response has been many of the best new companies don't want to go public.

So, this is a very interesting public policy question.  The number of publicly listed companies in the U.S. is down by half over the course of the last 12 years, right?  Because every year, right, companies drop out of the public market because companies get sold and they go bankrupt, and you need IPOs to keep replenishing the pool.  And that flow has basically stopped.

So, the number of companies that are public has been cut in half, right, which means that all of the investors who invest in public companies are now fishing in a much shallower pool, much shallower pond.  And those include all the pension funds, all the mutual funds, all the actual investors that actually represent ordinary people now can't invest in a lot of the interesting new things, they can't invest in the private companies.

So, we're creating this kind of two-tier investment world where if you're a public market investor, there are very few things for you to invest in, and if you're a private market investor you can do whatever you want.

And in a sense, that's really good for a guy like me because we can finance these companies.  It's not preventing the companies from getting financed because we can do it privately, but it is delaying the IPO.  It does mean the pension funds and a lot of the public participants don't get to benefit from the early growth of these companies, which is historically a big driver of returns.

So, the public market is basically being protected from itself to the point where it's being choked off.  And I think it's an unintended consequence.  Like I don't know that that's healthy.

ANDY SERWER:  Well, I don't think that's what they had in mind when they passed those laws.

MARC ANDREESSEN:  Right.  Unintended consequences.

ANDY SERWER:  Right.  Yeah.  That's interesting, and there are a lot of places to go with that point that you just made.  In fact, I mean, you're on the boards of several large public companies, but as a director you can't talk about what goes on at those companies, which is unfortunate because you have a deserved reputation as being one of the most open people in Silicon Valley when it comes to talking about what you do and what your company does.  But when it comes to being on the board of these public companies, it becomes selective disclosure, right, and you can't talk about that.

Having said that, let me ask you this question, though:  Suppose you had a fast-growing, very hot soda pop company and you were looking to go public.  And the company decided to increase the size of the offering and then increase the price a couple times.  And the company went public and they were able to sell at this high valuation, but the stock didn't do terribly well and it pissed a lot of people off and there was a lot of sturm und drang.

Would the fact that the company got all this money into its coffers make that a successful IPO or would the fact that there's all this sturm und drang mitigate that?

MARC ANDREESSEN:  I don't know.  (Laughter.)  The soda cans are blue, the color blue?  (Laughter.)

ANDY SERWER:  I don't want to disparage Indra Nooyi in any way.

MARC ANDREESSEN:  With white lower-case font?

ANDY SERWER:  Yeah.  Yeah.

MARC ANDREESSEN:  Unfortunately, this whole topic is so --

ANDY SERWER:  Because of what I'm saying, it's so obviously --

MARC ANDREESSEN:  Metaphoric.

ANDY SERWER:  -- metaphorical about a certain company that you can't really discuss?  I don't mean to be cute.  You know, it's just you know the answer to that question.

MARC ANDREESSEN:  I will say, let me take the general case.  There's an embedded issue in IPOs, and this again is part of the hostility that the public market has for the public company.  There's an embedded conundrum in IPOs which is that if they are priced in a way where there's a 100-percent first-day pop, they get reported as huge successes, right, and everybody's all happy.  Like the press is all happy because the stock has gone up a lot, the early investors, people who bought the IPO are happy because they made a lot of money overnight.

The down side of that, which I've had with my companies a couple times, is that the company has left an enormous amount of money on the table.  And sometimes that's fine and everybody's happy, but sometimes later on in life you wish you had that money.  Because what that means is the offering was underpriced relative to the true demand.

There are other times when people price these IPOs much closer to the actual demand that's in the market.  And you try to peg it so it's more or less right.  You try to get a little bit of a first-day pop at least if you can.

But then there's this sort of weird conundrum where a lot of people who want to invest in IPOs think that they're doing it because there's going to be a --

ANDY SERWER:  It's a right.

MARC ANDREESSEN:  -- 100-percent first-day pop.  There's a right to basically double your money in the first day.

ANDY SERWER:  Right.

MARC ANDREESSEN:  And so when those people get disappointed, the question would be like, okay, was that a good thing for those people to get disappointed or a bad thing?  And I think it was probably a good thing because that's a bad pattern.

ANDY SERWER:  I understand that.  It will be very interesting to see two years out when you compare LinkedIn, which did have that pop, versus Groupon and Facebook and Zynga, because right now one's outperformed the other three, right?

MARC ANDREESSEN:  Yeah, sure.  Well, I would also say number one, LinkedIn did a very good job setting expectations on its financial performance going into the IPO.  And then number two, LinkedIn has like knocked it out of the park in terms of results.  The thing you've got to give LinkedIn enormous credit for is the performance in their business, and they've just done an extraordinarily good job on that, which is not to say that others haven't, but LinkedIn has been really good.  Stock prices do, at the end of the day, correlate to how well the business is doing.

ANDY SERWER:  And then when you talk about consequences, you talk about the consequences of the Facebook IPO are a chilling of the IPO market, right?

MARC ANDREESSEN:  I think the IPO market was pretty chilled to start with.  I mean, look --

ANDY SERWER:  If that had been a moon shot, maybe not, but who knows?

MARC ANDREESSEN:  We live in unpleasant times, right?  So, number one, right now, the public market hates tech stocks, right?  So, assumption number one, public market hates tech stocks.  The reason I say that is because tech stocks are trading at a 30-year low relative to industrials.

ANDY SERWER:  And you mean that broadly, software and hardware?

MARC ANDREESSEN:  The whole sector, the entire tech sector.  The public market stocks are trading at rock-bottom PEs relative to any other time in history.  Like you have to go back to the mid '70s oil crisis, horrible stock market.

ANDY SERWER:  Why do you think that is?

MARC ANDREESSEN:  It's we're still living through the hangover of the 2000 crash.  So, the 2000 crash was a generational psychological event that still weighs very heavily on people's minds for anybody who lived through it.  And just like when it looked like we were starting to get back up on our feet again, then the financial crisis in 2007 and 2008 hit.

Now, of course, you know, Europe blowing up, China slowing down, huge amounts of money being pumped into the U.S. without job growth.  Everybody's just in a horrible mood, public market investors.

ANDY SERWER:  And people are talking about high-yielding stocks, which are not tech stocks generally, and they're talking about multinationals, which sometimes are tech stocks.

MARC ANDREESSEN:  Yeah, and you know, gold, I mean, everybody wants -- you know, it's gold, and it's -- in any area you're in where commodities are the hot investment topic it's like a very depressed environment because it just doesn't make any sense.  It's not where any of the growth for the future is.  So, we're just in one of those times where just everybody hates tech stocks.

Now, historically, that's been a very good indicator in terms -- you know, this is the time when you tend to get the headlines, you know, magazine headlines like "death of equities" which are sort of the famous contrary indicators.  I do think we live in that time.  It's hard to call it whether it's this year, next year, or the year after, but the psychology could get much worse.

ANDY SERWER:  I saw that.

MARC ANDREESSEN:  And then to your point on LinkedIn, right, most of the other tech IPOs in the last two years, the stocks, you know, forget whatever first-day pop they had, they then crashed, which, again, is a sign people just generally -- people are not willing to take any risks on these things, they just want out.

ANDY SERWER:  I saw the short sells, saw that crossing the tape, they were in a high sense going back to last September, so that's always the same kind of thing.

Shifting gears completely, what are you and your partner Ben Horowitz most excited about right now?  Like what's the thesis?

MARC ANDREESSEN:  Yeah, so the big thing we're excited about is it just feels -- you know, I've been working in the industry now for about 20 years, and it just feels like -- we have a lot of people in the room -- it feels like a lot of the work that we put into building -- you know, originally before me but the PC industry and then the Internet and now the smart phone industry, it feels like a lot of work got put into getting us to the point where we actually have the Internet in everybody's hands, and both on PCs and now increasingly on smart phones.  So, it feels to us like now the game is actually beginning of, okay, what are the true Internet killer apps, what effect is the Internet going to have on a very broad swath of businesses and industries, what are the great Internet franchise businesses that are going to get built.

It's funny if you talk about that, because this was the big topic of conversation in 1999.  It was just we were all just early in '99, because it turned out like 1999, right, 50 million people on the Internet, 25 million on AOL.  It was just early.  Whereas now the global addressable market is 2 billion people on PCs on the Internet and pretty soon 5 billion smart phones, and broadband fully deployed.

So, now we have a chance to actually build all the businesses that we thought we were going to be able to build in the late '90s, right?

So, when everybody was euphoric it was too early, and now that everybody is depressed now is the right time, which is sort of, you know, par for the course.

But we're excited about it, because we're just -- you know, the way I put it is the things that are working, the new businesses that are working are working really well.  I mean, the really astonishing thing -- I mean, it's, you know, we're in venture capital so not everything works, but the stuff that works is just on fire.

ANDY SERWER:  Such as?

MARC ANDREESSEN:  I mean, you go through our portfolio, you just go through a whole sequence of things.  But companies that aren't in our portfolio, it includes Dropbox, Evernote, and Spotify -- specifically naming ones that we're not invested in that are just on fire and spectacular companies and they're growing at light speed.

And we actually have this phenomenon now where we have a pair of companies that can go from zero to $1 billion in revenue in just very little time.

ANDY SERWER:  So they really cracked the code.

MARC ANDREESSEN:  And then that goes back to the investment thesis we have, which is software eats the world, which is basically that because of the world getting connected and because of PCs and smart phones being in everybody's hands, you can now build Internet-style businesses or dot-com style businesses in many industries and basically software can colonize other industries.  Air B&B is software eats real estate.  GitHub is software eats software development.

We just started a company called Solum, which is cloud analysis of soil samples for precision agriculture, which is an Internet company that focuses on soil, so we call it software eats dirt.  And we're basically just seeing this pattern across a very wide range of businesses, and it's a very exciting time to build these companies.

ANDY SERWER:  You talk all the time about software and you're Mr. Software, but you're also on the board of HP.  I know you can't talk about that, but hardware still exists.  Lenovo is here, Michael Dell is here, you're on the board of HP, you've invested in some hardware companies.  Does hardware have a role in this going forward?  It seems like it does.

MARC ANDREESSEN:  Yeah.  So, I think the fact that software is becoming so important is actually leading to a new kind of hardware renaissance.  I think that consumer electronics might be in the process of coming back to the U.S.  And that's a pretty big deal for those of us a little bit older, like Andy and myself.  You know, 30 years ago when I was growing up it was a huge topic of all the consumer electronics businesses were leaving the U.S. because it was all commoditized manufacturing.

ANDY SERWER:  Right.

MARC ANDREESSEN:  Well, you know, it's interesting today you've got the iPhone.  And so it's assembled in China, but all the components are built in Dubai and the United States, all the profit blows back to the United States.

You've got the entire smart phone industry now software is being built in the U.S.  And you've got all these new categories.  We have two investments we're very happy with, one is Jawbone and the other is Lytro, which is the new camera company.  You know, both of which are hardware companies, designing everything in the U.S., all the profits come into the U.S.

And the thing that's happening is basically these products are extremely sophisticated software wrapped in special hardware.  But as a consequence, companies that can only do hardware are going to have a very hard time.  So, one of the great stories of our time is what's happening at Sony, right?  Because a lot of people remember Sony used to be the innovator in consumer electronics.  Now, the new CEO, who is a very talented guy, has a huge challenge on his hands because Sony never became a great software company.  The disconnect between Sony and Apple is largely a story of software, not hardware.

So, we think that there's this model for basically extraordinary software wrapped in special hardware that is going to be a very big deal, it's going to apply in more and more categories, and then American companies in many sectors are going to be the leading companies because the American companies tend to be better at software.

ANDY SERWER:  And you indicated to me that there are some on your radar screen now in fact?

MARC ANDREESSEN:  Yeah.  We have another stealth investment which I'm very excited about.

ANDY SERWER:  And the name of that company is?

MARC ANDREESSEN:  I can't say.

ANDY SERWER:  Really?  Okay, I understand.

MARC ANDREESSEN:  Hopefully we'll be able to announce it -- how about next year on stage?

ANDY SERWER:  Right here.  Okay.  You have some witnesses.

MARC ANDREESSEN:  It will be something everybody will be very excited to take with them and take home to their kids.  Let's put it that way.

ANDY SERWER:  Wow.

MARC ANDREESSEN:  Very cool new hardware.

ANDY SERWER:  Hey, I want to take one second out because we have our first AT&T Mobile polling question.  (Tones.)  That was amazing, that was genius.

Okay, here's the question:  The question is -- and it's open to all domestic carriers.  What will future generations remember Marc Andreessen for?  Is it, A, co-founding Netscape?  Is it B, Andreessen-Horowitz's pledge to donate half of venture capital earnings to philanthropic causes?  Or is it C, none of the above, whatever it is hasn't been invented yet?

MARC ANDREESSEN:  How about option D:  Misspelling my name on the quiz.  (Laughter.)

ANDY SERWER:  That's a real problem.  I apologize for that.  (Laughter.)

MARC ANDREESSEN:  It's okay.  I'm not allowed to vote.

ANDY SERWER:  I've gotten my name misspelled too, but not by your company.

MARC ANDREESSEN:  Okay, good.

ANDY SERWER:  Anyway, so go ahead and vote.  Sorry about that, Marc.

MARC ANDREESSEN:  No problem.

ANDY SERWER:  So would anyone like to ask Mr. Andreessen some questions?  (Laughter.)

(Break for direction.)

QUESTION:  Marc, I'm curious, you just mentioned you think that the consumer electronics could be coming back to the U.S.  You talked about how things are being designed here, still assembled in China.  Do you see the assembly potentially coming back?  Does it get to the point where for those companies it becomes cost effect to build where you're designing?

MARC ANDREESSEN:  Yes.  You know, this is the whole thing -- this is a lot of the fallacy around China.  Manufacturing is a very loaded word, as you know.  Manufacturing applies that something is getting created.  But a lot of what happens in China is assembling components that are built elsewhere.  Or assembling components that are manufactured in China that are based on American IT.

So, the way you actually, in my view, have to look at it from an economic standpoint is gross margin, right?  So what is the actual embedded margin in those products?  And you just basically follow the profits.  When you follow the profits, you find that basically the people who actually design the IP get the majority of the profits, which is kind of how it should be.

So, the assembly part is sort of an arbitrage of labor against transport, right?  Which is it's cheap enough to ship everything and then ship it back.  And then it's cheaper to actually do the manufacturing.

You know, I don't know if on the one hand, if these products get more componentized and they become easier to assemble, then that argues you could bring them back because there's just less work to do during the assembly part.  You could also argue that if they get more complex and there's more parts that go into them, you should also bring them back because the manufacturing job gets harder and you might want to have it close to home.

You probably want to track the cost of shipping pretty carefully because that has a lot to do with it.  But I guess the final thing I'd say is I think it actually doesn't matter that much because as long as everything -- this is where I think the China thing is overblown.  As long as everything is being designed in the U.S., I don't think it really matters where it's built.  I don't think that really has anything to do with the economics, it doesn't have a lot to do with any of the high-value jobs.  Frankly, the other problem is there are very few Americans who would actually want to work in a Chinese manufacturing plant at those wages.

ANDY SERWER:  Question over here?

MICHAEL SCHRAGE:  Michael Schrage with MIT.  Marc, you've been a pioneer on the cloud.  You really, with GitHub, are very focused on the economics of coding.  I'm very interested in your perception of the evolution of cloud computing and machine learning.  How steep is the machine learning curve and how big of an opportunity is that for redefining the kind of surfaces and speed and agility associated with those surfaces for the Internet revolution you're describing here?

MARC ANDREESSEN:  Yeah.  So, machine learning might be one of the few areas my friend Peter Keel (ph.) and I might agree on.  I love talking to Peter, I love arguing with Peter because he always comes up with new ideas.  I almost always disagree with him, and then he's almost always right.  So, I don't know what the pattern there is.

So, he's taken a position that now is exactly the time to go long on artificial intelligence, as he puts it.  So, those of us who have been in the computer industry for a while, AI is sort of a maligned term.  Machine learning is the new term for AI because AI became such a term of bad repute.  People lost so much money on AI.  It's like when everybody lost money on pen computing, all of a sudden it became tablets.  (Laughter.)  Then we lost all the money in the dot-com era, then we stared building cloud companies.

We might be entering the golden age of machine learning or AI.  And the reason actually has a lot to do with the cloud, which basically is it turns out the approaches to AI that work are not the ones where we try to get a computer to think like a person, they're the ones where we basically use essentially extremely sophisticated brute force algorithms against extremely large amounts of data.

And the most interesting example of that is using Google as spell checker.  Google is, by far, the most impressive spell checker ever built.  Spell checking is sort of a classic AI problem of understanding a language and then being able to correct syntax and grammar and spelling.  It's always been hard to get spell checkers to work right because it's been hard to train computers how to speak English, or how to translate languages.

Google is the spell checker that works in like virtually all cases.  And what's interesting is Google doesn't actually have a spell-checking function, right?  It just simply sees all the different queries and it sees where people end up correcting themselves in volume, and so it's able to automatically predict that if you made this mistake, therefore you would end up doing this other thing.  So, it's an amazing spell checker.

And that same kind of approach is now being used in a lot of other areas in predictive analytics, in face recognition and all kinds of things.  And the whole basis of it is huge amounts of data.  So, the cloud aspect of having huge amounts of data in the cloud will basically power all those applications.  And now is probably a pretty good time for that.

QUESTION:  Hi, Marc.  In the '80s we saw a single-point solution like graphics on the PC software and WordStar and Lotus123.  Over time, it became the Microsoft suite.  Then we saw that again in the computer software on a single point for accounting, manufacturing, supply chain, HR and then SAP.  Now we have the Internet that's having the same problem, which is a lot of single-point solutions.  Where do you see it going forward?  Is this going to converge to an integrated solution with suites coming up?  Or do you think the integration cost is going to be low, so we're going to continue to see a single point of solution for a while before they integrate?

MARC ANDREESSEN:  Sure.  And you're talking more about cloud applications, business applications?

QUESTION:  Could be business and B2B and B2C because the user has the same problem.

MARC ANDREESSEN:  Yeah.  In the long run, there will probably be consolidation.  I think it's still really early.  So, there are some companies that are being able to expand their footprint.  On the business side, Salesforce.com and Workday would probably be two that would immediately jump to mind that will probably add more functions over time.

But it's still really early for cloud applications in general on the business side.  I think we're still -- a couple things.  One is, we're still figuring out what the killer cloud application can be, and there are lots of new ideas being experimented with.

Number two, cloud companies are able to go after much larger markets than historical enterprise software has been able to go after, right?  So the market for Salesforce.com is a lot bigger than the market for Oracle because Salesforce.com can go down market much more effectively.  They just can go after a much larger number of businesses.

And so we don't yet have a real sense, I think, of what all the different categories are in the new world and what the market sizes are and how big the companies can get.  So, I think it's going to be years before there's sort of a big consolidation.  That's like ten years out or 15 years out.

Integration also is actually pretty easy on the cloud, and there are a bunch of companies that specialize in that.  We've got a couple in our portfolio that are really good.

ANDY SERWER:  One more quick question back here.

CHRIS FRALIC:  Hey, Marc.  Chris Fralic from First Round Capital.  Can you talk about how venture capital is evolving to help startups beyond just writing a check?  In particular, what Andreessen-Horowitz is doing on that front.

MARC ANDREESSEN:  Yeah.  So, I might take slightly longer than 30 seconds to answer that question.

You know, one of the fun things about venture capital right now is there are a bunch of new models.  And so what you guys are doing is one new model, what Wyc is doing is another new model.  There are a whole bunch of them.  You know, I'm excited because I think it's a good thing to experiment with.  You know, if you're going to be in the business of funding innovation, actually doing some innovation might actually be a good thing.

So, it's a very exciting time.  Our approach to it is sort of based on our backgrounds, our experience.  It may or may not be right, but we think it works, which basically we had great experiences with the VCs we worked with, but we were always wishing that they would do more for us in certain categories.  And so we basically said as a firm let's build up the capabilities to do those things in those categories.  Not doing things where they would have just interfered with us, but the things where they could really help us.

The big thing turned out to be network.  We always wanted more connections.  So, we always wanted more customer introductions, we always wanted more business development relationships, we always wanted more international help, we always wanted more talent, we wanted more referrals, help hire programmers, executives.  We wanted help working with the media.  All these things -- we had a round of investors, acquirers, all these things where you would think venture capitalists have relationships that have been built over 10, 20, 30, 40 years that you could kind of plug into as the next startup.

And what we found is venture capital firms, in general, just have not systematized the network.  So, the big thing that we're trying to do is systematize the network.  So, we're doing it across categories, we're doing it for customers, we're doing it for partners, we're doing it for acquirers, we're doing it for future financing rounds, we're doing it for press, we're doing it for talent.  And we have dedicated teams in each of those areas.

So, when you work with us, you get a GP, and then you also kind of plug into the network.  And at this point, it's becoming a very large network.  We're going to run 600 corporate briefings in our office in the next 12 months.  So, our entrepreneurs literally get an agenda that basically says if you want to meet with the CEO of GE, it's next Tuesday at 4:00.  And it's like that every week.  And that turns out to be good for the big companies, and it also turns out to be good for the startups.

ANDY SERWER:  Great.  All right.  We are at time, we're going to have to leave it at that.  Please join me in thanking Marc Andreessen.

MARC ANDREESSEN:  Great.  Thanks, everybody.  (Applause.)

Posted in:
Join the Conversation
Current Issue
  • Give the gift of Fortune
  • Get the Fortune app
  • Subscribe
Powered by WordPress.com VIP.