00:00hi everyone welcome to the a6 & Z
00:02podcast I'm sonal and today's episode
00:04features Marc Andreessen and a six in C
00:06board partner and co-founder of 21
00:08Balaji Srinivasan this conversation took
00:11place in front of a group of Stanford
00:12engineering students as part of the
00:15engineering Honor Society Tau Beta Pi
00:17distinguished annual speaker series with
00:19thought leaders and technology let's
00:20talk about one of the things I'm sure
00:22every student here at wants to or bring
00:24up everyone but a lot of them you know
00:25think about startups think about
00:26technology as an entrepreneur as a
00:27founder as a potential employee how
00:29should students today you know graduate
00:31from shampoo think about startups as in
00:33both the founder context in an employee
00:34context yeah so the traditional venture
00:37capitalists all have like a secret sauce
00:38kind of formula of how they think about
00:40what they want to fund and then it turns
00:41out I think the formula is all reduced
00:42to the same handful of factors with the
00:44exception maybe a Peter Thiel who has
00:46like six other factors in his head that
00:47he hasn't told anybody about but for
00:48everybody else basically always reduces
00:50down to some combination of market
00:52product and team if you talk to people
00:54have been adventurer for a long time
00:55what they'll tell you is basically the
00:56difference between venture firms you
00:58know in a lot of ways is based on how do
01:00they rank the importance of founder of a
01:02market product and team you know as an
01:04example Sequoia was legendary and
01:06prioritizing market over team right at
01:08Valentine has if you if you go online
01:10and Google divalent eyestalks he talks a
01:12lot about how the key to success of a
01:14startup is to land yourself in a giant
01:15market like land yourself in a market
01:17that's about to become explosively large
01:19and basically once the startup is in a
01:20position where it is the leading company
01:22and in an explosively large new market
01:24the people become somewhat fungible like
01:26you can swap the people out and then he
01:27would cite Cisco as a one of the great
01:29case studies of that which you know is
01:30actually a Stanford spin-off
01:31husband-and-wife team very sharp
01:33founders but they got booted very
01:35quickly actually by Don Valentine who
01:37brought in professional CEO John
01:38morgridge who who was phenomenal and
01:41actually built you know built built
01:42Cisco the company and so that's one
01:44model the diametrically opposed model is
01:46prioritizing team over market basically
01:49saying that you know the right market
01:51whatever can you really even know what
01:52the good markets are gonna be like how
01:54well can you predict really what you're
01:55doing is you're going into business with
01:56people either going to business with
01:58really good people or not if you're
01:59going into business with really good
02:00people one of the things that should
02:01make really good people really good is
02:03they should be able to find themselves
02:04you know a good opportunity right a lot
02:06of startups end up you know they succeed
02:07based on something different than what
02:09they started doing and so if you get in
02:10business with the right people they'll
02:11be able to sniff out the opportunity
02:13Peter I don't put words in his mouth but
02:15I think he'd probably prioritize product
02:17over market and team which is you have
02:18to be doing you have to be making a
02:20fundamental advance in technology we can
02:22tolerate a lot of flaws in the people
02:24and he can tolerate a lot of uncertainty
02:25around the market if the product
02:27breakthrough is big enough he'll make
02:29those other bets it's kind of an Angels
02:31dancing and a head of a pin thing you
02:32kind of is a VC you sit around talk
02:34about this a lot and then if you want an
02:35investment company you kind of figure
02:36out some rationalization I guess your
02:38formula to do it so I don't want to
02:40overstate it but I wanted to go through
02:41it because I do think that is the
02:42framework you know as you think about
02:43startups as some is either a startup
02:45that you might start or as you think
02:46about a startup you might go to I think
02:49that's a pretty good framework in terms
02:50of where if you're here is a student if
02:52you're gonna be graduating my personal
02:54recommendation would be to focus much
02:56more on team and the reason is just
02:58because I think we struggle from a
03:00distance to evaluate market and we also
03:02actually startled evaluate product but
03:04if you can get yourself in business with
03:06really good people I think number one
03:08like if it works it's great because
03:09those are really good people to be a
03:10business with and they can they with you
03:12can build something great but even if it
03:14doesn't work even if it's wrong market
03:15or the wrong product you'll still learn
03:17so much working with the right people
03:18and you'll build such a valuable network
03:20for whatever you do next
03:21it would also apply if you start a
03:22company like who do you start a company
03:23with you may end up in this situation or
03:25it's like do you start the company with
03:27a super genius who's cantankerous and
03:28hard to get along with or do you start
03:30the company with the person who's like
03:31maybe not quite as incandescently bright
03:33but maybe it's much more collaborative
03:34by the way I don't know there's a right
03:36answer I do know it helps a lot early in
03:38your career to be working with really
03:39good people because it really gives you
03:41guys gives you a sense of what good
03:42really means and gives you the ability
03:44to learn yeah I would say one thing that
03:46we've all talked about is that it should
03:48be exceptional at least one dimension
03:49yes it can't be like just pretty good
03:51and all these different things at least
03:53one dimension needs to be like truly 10x
03:55and you know amazing to make the bet
03:57that's exactly right we talk a lot in
03:58our firm about we have this concept we
04:00say we invest in strength not in lack of
04:03weakness and again it's one of these
04:04things that sounds obvious but it's
04:05proof to us to be a pretty big deal so
04:06there's a lot of there's a lot of
04:09startups you'll run into or you probably
04:10have friends who are at these companies
04:11or know people at them and it's like
04:12teams good products good market seems
04:15good they're making some progress I got
04:16some customers the customers are pretty
04:18happy ok where is that really gonna go
04:19and where is it really gonna go because
04:21what spectacular about it right what's
04:22the thing that's gonna cause it to jump
04:24out from the other hundred or other
04:27where you can say the exact same thing
04:28so then you say okay great now I want to
04:30I want to I want a vest a strength okay
04:32that's easy the probably investing is
04:34strength of the problem is running a
04:35company is that the strongest startups
04:36at the point of contact which you
04:37discover is the strongest startups
04:38aren't strong at everything they're
04:40strong at something and then they often
04:42have the term we internally use
04:43ironically as they have hair on them
04:45which people always kind of surprised
04:47when I start to use that metaphor but
04:48they often have serious team issues many
04:50successful startups have a founder
04:51divorce at some point likely the
04:53founders go to war and you would think
04:55that would be a very bad indicator and
04:56actually sometimes it's a really good
04:57indicator because it means that things
04:58are really starting to work and like
04:59it's time to get serious and one founder
05:01rush to get serious another one doesn't
05:03or you'll have these we some of our best
05:05companies are like stellar at product
05:07and engineering and cannot go get a deal
05:09with a customer to save their life and
05:10like labor for years under the illusion
05:12that the way the world works is that if
05:13you have that you know if you have the
05:14mousetrap everybody beats a path to your
05:15door and then three years later they're
05:17like oh we have to get salespeople to go
05:19sell things and so there's these things
05:21and they'll just drive you nuts but if
05:22the strength is strong enough they can
05:24really punch through and and so much
05:26about this another another thing maybe
05:27worth we're saying is the default state
05:29of every company is just dying in
05:31obscurity and so so much of a is how do
05:33you punch through how do you punch
05:35through in the minds of the of the
05:36people you're gonna have to recruit how
05:38do you punch through in the minds of the
05:39investors how do you punch through in
05:40the minds of the customers how do you
05:42punch through to the press like how do
05:44you actually get yourself visible such
05:46that you can start to attract the kinds
05:47of you know business and momentum and
05:48talent and money that you need to be
05:50successful and so that sort of model of
05:52strength versus like a weakness I think
05:53is pretty important every startup and
05:55every project starts as a as a
05:56hallucination right like it's a it's a
05:58word on a napkin it literally doesn't
06:00mean anything and if you have to believe
06:01it can become much bigger than it is and
06:03always at every stage it has to become
06:05you have to believe it's bigger than of
06:08but by the way it means in our business
06:09it means in our business we're doing
06:10something right there's something
06:11basically horribly wrong with every
06:12company we fund one of the reasons like
06:14investment banks are the headphones
06:15don't just come in and do venture
06:16capital is because they're just
06:17horrified at every single investment we
06:19do the one saving grace that we have
06:21with that model is with portfolio so we
06:22get to make you know basically 30
06:24grossly irresponsible bets right in a
06:26portfolio and then basically the math is
06:28if we're doing our job right 15 work and
06:2915 don't and in almost any other area of
06:32investing or any other area of business
06:33if you have that kind of failure rate
06:35right with that kind of risk level per
06:36decision you would just throw up and go
06:38if there's one edge that we have is the
06:40ability to kind of indole
06:41in these situations where the strength
06:42is crazy but the the weaknesses are also
06:44frankly crazy yeah I mean like the thing
06:46is if it gets dearest all the way then
06:48it's just a safe investment and there's
06:50very little upset but I think it also
06:51holds for technology in the sense that
06:52if you read about something in the Wall
06:54Street Journal or the New York Times and
06:55Technology it's on everybody's lips it's
06:57probably not always but it's probably
06:59started to you know have some of the
07:01value taken out of it in the sense that
07:02there's a lot of companies that already
07:03built in the space is very competitive
07:04and the technology to look for are often
07:06the ones that haven't got a lot of press
07:07yet you know that are your inception
07:09that are in the labs of places like
07:11Stanford if it's a buzzword if it's
07:12something that some people's lips if
07:13there's a magazine articles about
07:15ordering newspaper articles about it or
07:16god help us if it's on TV
07:17yeah well the time has passed like we
07:20better look for something new so relate
07:22to this if we just talked about how
07:24people should think about pursuing
07:24startups but what does it mean for so
07:26folks who are you know employees what
07:29does it mean when companies Steve
07:30private longer and what do you think of
07:32as the root cause of this relatively new
07:34phenomenon really last 10-15 years or so
07:36so the model for Valley startups right
07:38used to be very straightforward which is
07:39you'd raise an around then you'd raise a
07:41be round to kind of build out your sales
07:42force once the product started working
07:44you raise a serum to maybe expand in a
07:46couple other countries maybe add another
07:47maybe do a little acquisition or
07:49something and then within you know four
07:51five six years gift about you know 30 40
07:5350 million in revenue and you go public
07:54it was sort of you know that was sort of
07:56the rite of passage and then a bunch of
07:58things became possible once you were
07:59public that weren't possible before so
08:01one was liquidity which is early
08:03investors and employees could start to
08:04sell stock but there were other very
08:06one was it was viewed as a legitimizing
08:08event especially for companies that sell
08:10products to other companies it was
08:12viewed as an event that basically was
08:14you know a lot of big customers of
08:16technology would much prefer to buy
08:18technology for public companies because
08:19they feel like they can understand the
08:20vendor they're buying from whereas these
08:22private companies they don't know if
08:23they're still going to be a business or
08:24not and then also M&A mergers and
08:27there you know is considered a great
08:28virtue of being public is to have an
08:30acquisition currency right to be able to
08:32issue stocks and a lot of the great tech
08:34acquisitions over the years we're done
08:35with stock because you know you get
08:37basically when you can use that value to
08:38buy things even if you don't have the
08:40cash the stereotype is that everybody
08:42wants to go to work for a start-up in
08:43the valley I think the reality is a very
08:44large number of people actually don't
08:46want the true early-stage risk they want
08:48to go to a company that's doing
08:49interesting things but they don't want
08:50to have to like go in for another job in
08:52six months if something goes wrong
08:54they've got like a family it cannot like
08:56a spouse and they've got a mortgage and
08:58they've got kids and they've got they've
08:59got bills they have to pay and so
09:00there's actually a lot of talent that
09:02got unlocked once you became public that
09:04you could actually recruit and so those
09:06are the old days interestingly in the
09:07u.s. the number of public listed
09:09companies in the US peaked in 1997
09:11weirdly enough and you might you might
09:12think it peaked in take 2000 or 2002 or
09:14something but actually peaked in 97 and
09:16basically the number of public companies
09:17in the US has now dropped by two-thirds
09:19since 1997 and that has coincided with a
09:23bunch of other things I mean one was you
09:25know we had that we had the stock market
09:26crash and we had the credit crisis but
09:28it's also coincided with some other
09:29changes one of the big changes for
09:31example a lot of tech IPOs actually were
09:33individual investors right a lot of
09:35historical investors and small tech
09:36companies were individuals who would
09:37read about these things and get excited
09:38and invest if you just look at the
09:40statistics on this the percentage of
09:42ownership of tech stocks by individuals
09:44has dropped like a rock since 2000 it's
09:46basically now all funds right and funds
09:48are inherently more conservative than
09:49individuals because funds have you know
09:51if they feel like they have
09:52responsibility to be sober and so
09:54they're not that excited about the next
09:55hot IPO and so the public market like
09:57just a lot of the enthusiasm has been
09:59drained out of it the Merc has changed
10:01dramatically and so it's it's sort of
10:03you know typologies question has kind of
10:05become in vogue or in style to not
10:07either not either not go public or at
10:09least not go public as fast as before
10:11the good news about staying private
10:12longer is that there is something about
10:14going public that puts you on a
10:15treadmill with quarterly results they're
10:17like well you know I'm not gonna get on
10:19this treadmill the quarterly results
10:20where I have to hit all these quarterly
10:21earnings targets I'm still gonna be able
10:22to do long term things so the good news
10:25about staying private is that you can do
10:26these big ambitious projects over long
10:28periods of time and you know you either
10:30get them right or you don't but you're
10:31not under any specific quarterly
10:32pressure to deliver any particular set
10:34of financial results my view is that the
10:36pendulum has actually swung too far now
10:37in the direction of not going public
10:39like too many companies are now staying
10:40private too long it used to be that it
10:43was a contrarian view that you should
10:44stay private it's now becoming
10:45contributes now becoming a contrarian
10:46view that you should go public and my
10:48argument of why more companies to go
10:49public is at some point it's good and
10:51not just have all of your results be in
10:53the future but to actually you have to
10:55deliver in the present and at some point
10:56it's good to have an organization that
10:58actually like knows how to how to work
11:00properly and knows how to sell things to
11:02people and knows how to like have
11:03financial plans and hits and knows how
11:04to make money and it's all hypothetical
11:06until you have to prove it
11:07and I think a lot of companies that are
11:09staying private for too long risk
11:10getting sloppy and undisciplined and in
11:12the beginning that's fine but at some
11:13point you have to get serious and if you
11:15can go for 10 years without getting
11:16serious I think there's a real risk that
11:18you never get serious so that's one of
11:20the number two I it's become massively
11:22differentiating to go public because you
11:24get these big advantages you still can
11:25then tap the public markets for more
11:27money people talk about Elon Musk and
11:28you know SpaceX is still private but
11:30Tesla is a public company so Ellen must
11:32puts out this thing that the Tesla Model
11:333 the pre-orders and they gets happened
11:35knowing pre-orders and all of a sudden
11:36everybody hated Tesla before cuz nobody
11:38wanted to buy the car now we're you know
11:40now all the investors say Tesla because
11:41now there's too much demand for the car
11:43right which is apparently equally bad
11:44and so he just now said he's gonna do a
11:47two billion dollar secondary offering
11:49right in the stock market and like even
11:51in modern like venture capital it's hard
11:53to raise two billion dollars at a shop
11:54not very many people can do it and so he
11:56can actually like raise that amount of
11:57money publicly he can access debt and
11:59then you know you go back to the
12:01acquisition currency like we've probably
12:03been in a slow period for M&A for a
12:04while but there is no question there's
12:06gonna be a lot of M&A in the years ahead
12:08and the companies that have public
12:10currencies they're gonna be able to be
12:11the acquirers and able to get big and
12:13become much more important so I think
12:15the pendulum is gonna swing back in the
12:16other direction there's a cropper
12:17companies good companies definitely
12:19I think another part is also Starbucks
12:21and all the rules and then dodd-frank
12:23and so on has made it quite difficult to
12:24be a public coming from a compliance
12:25perspective and the biggest cost
12:26associated that yes so there's this
12:28thing sarbanes-oxley which I see
12:29somebody in the audience yawning and
12:30this topic is gonna make everybody yawn
12:32and so I'm not gonna go into detail you
12:33can google it if you really want to
12:34learn about it but it's the regulatory
12:36kind of threshold that public companies
12:37need to hit on how they deal with risk
12:39and they do reporting and all this stuff
12:41and the knock on sarbanes-oxley has been
12:43exactly what Balaji said which is it's
12:44it's basically a burden that falls
12:45disproportionately on small companies
12:47because big companies have huge staffs
12:48of lawyers and finance experts and so
12:50forth who can do all this stuff but
12:51small companies at the burden falls
12:53directly on the management team our
12:54partner Ben Horowitz now argues the
12:56opposite side of this having seen a lot
12:58of companies which he argues if you're
13:00good enough as an operating team to
13:02actually comply with Starbucks then
13:03you're good enough basically to do
13:04anything like basically it not
13:06everything in it makes sense but it sets
13:08a bar for what it means to be an
13:09operating business that's operating in a
13:11responsible way so he I think he's
13:13actually flipped a little bit on that I
13:14think he would argue it's actually part
13:15of being a responsible company at some
13:17point interesting it actually kind of
13:18gets into our next question you're gonna
13:20you important technologies one thing
13:21I've thought a lot about is that the
13:23ultimate kind of solution to this is
13:25gonna be something related to the
13:27Bitcoin / aetherium crowd funds that are
13:29happening now on the Internet where you
13:32have the regulatory stuff has to be
13:33worked out about that but you do have a
13:35very large potential pool of capital
13:36that people can use for this kind of
13:38thing and that might be you know the
13:39this is an essay that navall and I wrote
13:41a couple years ago about like an app
13:42coin so you'd actually start a company
13:44and actually issue a coin that could be
13:46used to redeem for calls of thats a
13:48service so that's one model that my
13:49matches mentioned Bri is this is a whole
13:51new model for how to think about sort of
13:52crowdfunding to another level we just
13:54mention the Dow and what that is yeah so
13:56this is a pretty interesting concept on
13:58where so in theory it's something that
14:00was based on Bitcoin initially and is
14:02sort of like a more programmable version
14:03of Bitcoin in some ways there is a thing
14:05called the Dow which raised almost 130
14:08million dollars online in a purely
14:10distributed way just with digital
14:12currency without any stock market or
14:13what have you there's all kinds of
14:15regulatory hair on this animal and
14:16people can pull their money out of it so
14:18it's sort of like a VC fund where the
14:20LPS don't actually commit until they see
14:22the first investment and so I think
14:24there's gonna be all kinds of stuff that
14:25happens with it nevertheless I think
14:26there's a very interesting experiment
14:28and something which will probably be
14:29relevant for you guys not not this year
14:32not next year but in maybe five to ten
14:34years in terms of potentially an
14:35alternate way to get financing for
14:37something so actually that leads us into
14:39important technologies right so let's
14:41get a quick riff on them one by one
14:43so starting with maybe you know talk
14:44about Bitcoin and blockchain then
14:45FinTech more broadly yeah so I'm gonna
14:48turn the first one around so Balaji is
14:50the founder of one of our two big
14:51Bitcoin investments so sure
14:52Balaji how's Bitcoin doing how's it been
14:54doing yeah so you know like the Gartner
14:56hype cycle right something we think
14:57about a lot we think of is this
14:58fundamental thing in technology that is
15:01there you've got this trigger and then
15:02people get really amped about a
15:04technology and it's everyone's doing it
15:05oh you know bots are at that stage right
15:07now and then you you try to actually do
15:09it and you find it's actually hard and
15:11there and gets demoralized and they quit
15:12and you've got the trowel and then it's
15:15those guys who stick it out in the
15:16trowel and pull up over here that you
15:18know things actually happen so that
15:19happens like the dot-com bubble it
15:21everyone was hyped about in 2000 and
15:22crashed and then actually you built all
15:24these massive businesses and it happens
15:26on like larger and larger cycles as well
15:28Carlota Perez she's got this whole
15:30theory about why that happens and it
15:32kind of happens a different ski
15:34and we sort of think that's happening
15:36for Bitcoin in the sense of there's a
15:38huge amount of excitement like 2013-2014
15:39you know oh my god new paradigm then you
15:42know like oh the price crashes and now
15:44it's coming back up with a lot of like
15:46the micro payment stuff interesting
15:47things happening this year
15:48I think the blockchain stuff is actually
15:50right the top the Gartner hype cycle and
15:51I think it's gonna crash down like
15:53towards the second half
15:54you know this year when people actually
15:55try to implement it that's where I kind
15:57of think Bitcoin and blockchain is and I
15:58would say that you know in addition to
16:00our kind of point earlier about like you
16:01know getting technologies that are that
16:03nobody knows about at all
16:04that are in the lab right now I think
16:06other kinds of technologies to really
16:07look at are those that people have
16:08written off right like you know VR after
16:11Second Life and so that's the kind of
16:13thing to look for the stuff that people
16:14think of as you know dead or didn't work
16:17or what-have-you and find out why so
16:18it's very funny you don't remember the
16:19first time VR got written off oh no
16:21that's true you only remember the second
16:23time everyone the second deaf guy yes
16:24that's right no actually you remember
16:25the third time I got written off I
16:26remember the previous two you got
16:28written off after BPL you got written
16:29off after the VR there's a whole view
16:31our wave in the late 80s one of the
16:32great all-time hacker movies one more
16:34man no it was kind of a peak of that
16:36cycle and then we bought up via our
16:37company it escaped in 95 to do VRML does
16:41VR on the browser you may you may you
16:43may note that that didn't work right
16:46there were a Second Life which was like
16:47the third cycle hey one of the things we
16:49talk a lot about is say to operating
16:51principles and how we think about
16:52technology well in things I'm come to
16:53believe that there are almost no actual
16:55new ideas right basically everything
16:57that is gonna be a big deal in the next
16:5830 years is in a lab somewhere probably
17:00here in a lab at Stanford and so the
17:02Eureka moment is like an almost
17:03non-existent thing maybe every once in a
17:05while but there's almost always a 20 or
17:0730 year backstory of research that often
17:10by the way turns out to be 50 60 80
17:11years back story of research before
17:13something pops and then the second thing
17:15is just yeah things take time there's
17:17this concept go to AI winter and it's
17:19literally there have been surges of
17:20enthusiasm and crashes in AI and I think
17:22we've counted they're like we're 5 AI
17:23winters yep between 1950 and basically
17:2610 years ago even the term AI has only
17:28come back recently after neural networks
17:30of themselves came back because everyone
17:31is like oh hey is all rule based and ml
17:33is the new thing and having another mini
17:36cycle within that we're like Chris 6 and
17:38I joke that so many AI companies are
17:39just a collection of if-else statements
17:41and you know it's like okay which are
17:43very compelling on first demo
17:45very first yet but it's always on Rails
17:46alright and then we're gonna try to get
17:48it a little bit off then it's like well
17:51yeah and so I think it's about you got a
17:54very important kind of fundamental point
17:55which is it's it's not it I mean what's
17:57new is is important but it's often
17:58what's new that where there's a where
18:00there is a track record of intellectual
18:02depth that's gone into it over a long
18:03enough period of time that people really
18:05have thought hard about it and it turns
18:06out that track record it's almost always
18:08multiple decades and then whatever
18:09happens to be hot or not in any
18:11particular moment is really not
18:12predictive of what's actually going to
18:14happen yeah exactly I think you know in
18:17particular there's two things if you ask
18:19me you know what like to look at for
18:21start-up ideas and so on like so first
18:22I'd say don't do startup unless you're
18:24ideologically driven to make it succeed
18:26beyond the economic motivation because
18:28it's actually very hard but if you do
18:30want to just find start-up ideas there's
18:33this book the sovereign individual it
18:34came out in the late 90s it's the most
18:36prescient thing in the world
18:38most bestsellers you can take the 300
18:40pages and compact them into like a
18:41one-page summary there's actually
18:42websites that do that right whereas this
18:44book is the opposite you can take like a
18:46page and turn it into a PhD thesis and
18:49what's awesome about it is you know we
18:50kind of think Satoshi read through the
18:52sovereign individual and actually made
18:53Bitcoin in part on that basis because
18:56the description of is so lucid but it's
18:57interesting is there's other pages of it
18:59which haven't yet been implemented so
19:01it's so like the book of prophecies and
19:02just flip through it all let me do that
19:04line right so I've been the kicker be
19:06you know that book ripped off another
19:07book an order boy was that older book
19:09called the twilight of sovereignty
19:10interesting I mean she was written by a
19:12guy named Walter Wriston founder of
19:14Citibank who spent 40 years in 40 years
19:17in banking 40 years in like big New York
19:18has situational banking and his
19:19conclusion at the end of it was it was
19:20all and he basically wrote a
19:22book predicting basically the rise of
19:24networks and just distributed finance
19:26distributed money this is like 30 years
19:28yeah so I mean what's interesting is
19:29those a lot of those guys got the
19:31general direction right and then there
19:33is some aspect that actually turned out
19:34to be much more difficult than they
19:35thought for example like autonomous
19:37robotics well actually that's really
19:39hard because of the number of degrees of
19:40freedom and the probabilities it's one
19:41but it's doable with enough training
19:43data I think the other thing that that
19:44you know I think of it like a Back to
19:46the Future thing that's very important
19:47is thing called t bow sorting so like a
19:49while back we found this guy done in
19:511956 and he had a bunch of assumptions
19:53for this model of how people could
19:55sort into like basically many
19:57governments around the world and he
19:58assumed like okay you have search you
20:00have perfect information you have
20:01perfect mobility of this year that and
20:03he basically like assumed the smartphone
20:04I'm gonna put it that way at that time
20:06but 1956 II assume the smartphone is
20:08like oh wow you can solve all these
20:09problems with governance and so on it's
20:10like literally 60 years later you can go
20:12back you know dust off this Raiders of
20:14the Lost Ark stuff and just you know go
20:15with it right and you'll sound really
20:17smart because you can just like read off
20:18the book of prophecies right but okay so
20:20other important technologies all right
20:22so AI all right we just kind of talked
20:24about this a little bit so autonomous
20:25cars drones ml and software what is your
20:28take on this yeah so magic is happening
20:30and I think everybody here probably
20:31knows this by now but it's something
20:32something has changed and actually what
20:35that something is as a matter of some
20:36debate and it's probably multiple
20:37something's but an entire battery of
20:40techniques that people have known about
20:41for a long time plus some new techniques
20:43have in machine learning and deep
20:45learning have really started to work
20:462012 was kind of the tipping point for
20:48that and you know it's really building
20:49steam and then it also feels like
20:50something changed part of the passage of
20:52time in our industry is just Moore's law
20:54allowing processors to kind of catch up
20:55with our ideas and the rise of this new
20:58generation of GPUs that are able to run
20:59these are able to run neural networks
21:01and deep learning algorithms is a really
21:03and then you know we know the existence
21:05proofs of you know fully running
21:07autonomous cars using deep learning
21:09we've got a ton of his drones with deep
21:10learning we've got you know alphago the
21:12great accomplishment that Google
21:13recently had the deep mind had we're
21:15like significant breakthroughs are
21:16happening I would say something about
21:17very dramatic happening but also
21:18something very real happening yep I
21:20would add to that actually just data
21:22like cuz you know like many of these
21:23algorithms you just put ten Exadata at
21:25them and they work and one one tenth and
21:27they don't and so like just the easy of
21:30collecting massive amounts right yeah so
21:32VR and AR so you know oculus and
21:34magically you've been stuff like that
21:35what are your thoughts on that area yeah
21:37so very exciting so a VR right is the
21:38idea of the headset of you basically are
21:40in a completely computer-generated world
21:41like to say the world's now divided into
21:43two groups of people people who haven't
21:45tried the shipping consumer version of
21:46oculus who think they are stupid and
21:48then people who have tried it who think
21:49it's the future of everything and so if
21:51you haven't tried it find somebody they
21:53just started shipping find somebody who
21:54has won and tried it it's it's a it's a
21:57really profound thing the other idea of
21:59people are playing with is augmented
22:00reality her AR which is the idea of you
22:03still see the real world but you have
22:04computer-generated imagery kind of
22:06populating it and there's
22:07Nicolle magic leap in florida that's
22:09doing this and Microsoft has a thing we
22:11actually argue there's two kinds of AR
22:12there's the kind that people are talking
22:14about because they find VR too scary
22:15that's why although although all the
22:17news articles in VR are all like very
22:18emotionally loaded because it's
22:19invariably a picture of somebody with
22:20this thing strapped to their face right
22:22you don't actually get to see what's
22:24inside the VR you just get to see the
22:25idiot sitting there in the chair with
22:26them you know the alien facehugger like
22:28this and then everybody thinks it's
22:30funny to a lot of people who find VR too
22:32AR feels like it must be more normal
22:34because I still get to see everybody and
22:35I think it's actually a little bit of an
22:36intellectual crutch for people who just
22:38can't quite come to gross as they are
22:39that said there's the other form of AR
22:42which is like if we can get AR to really
22:44work right and if we can get to the
22:46vision that I think everybody in the
22:47industry has which is get a pair of you
22:48know very light eyeglasses or even
22:50contact lenses an overlay computer
22:52imagery on the real world like that is a
22:54big deal yeah and there are teams there
22:56are teams that there are handful of
22:57companies now that have teams that are
22:58super focused on this and two thoughts
23:00one on the PR and one OVR one thing of
23:02the thing about AR is if that kind of
23:03thing can work I think you can have what
23:05we think of as like the Instagram
23:06efficacious of many more things in the
23:08sense that what is Instagram so yeah
23:10it's a photo app but it's also is
23:11something that takes somebody who has no
23:13skill and photography and gets them to
23:14like an eight because you got a
23:16programmer on your shoulder and you know
23:18he's like oh you know put the f-stop
23:19there and whatnot and don't you know you
23:21know generate and so on there's always
23:22at least one filter that makes any photo
23:23look good exactly that's right I
23:25actually think like the next version of
23:26miss Graham will make people prettier
23:27right like I call it tinder for
23:29Instagram so just keep swiping until you
23:32get attractive well yeah you exactly you
23:34just got a filter that just morphs it
23:36just a little bit right exactly the
23:39thinking is though that Instagram
23:40vacation you could apply to many other
23:41areas where they are right like so the
23:43classic examples are your mechanic and
23:45you put on the glasses and now you know
23:46every part lights up and you see the 3d
23:48schematics and you tap here to order the
23:50replacement from Honda and so on or
23:52you're a surgeon and you can actually
23:53see the person's x-ray superimposed on
23:55them and so it's like you've got a
23:56superpower right in that sense which
23:58actually you know tweets from you had a
23:59while back and then on the VR end of
24:01things you know one thing when people
24:02you know kind of dismiss VR I always ask
24:04them okay how much time do you spend
24:05looking at a screen how much time you
24:07spend looking at like a laptop or a
24:09phone and they'll say you know okay
24:10maybe you know six hours a day and still
24:12say okay well that's like 50% of your
24:14waking hours and we're probably gonna
24:15replace a significant percentage of
24:17monitors with VR with something to the
24:192d world right and there's gonna
24:20new windows that's based in the 3d
24:22universe which has totally different GUI
24:23metaphors so that's interesting kind of
24:25company to build that it doesn't exist
24:26yet but that company okay so when you're
24:28wearing this VR thing to do work not
24:30just to play video games well actually
24:31most your life is in the matrix so
24:34that's gonna be kind of interesting in
24:35like five or ten years everyone's
24:36wearing these kind of things it's coming
24:37great great okay what should Stanford
24:39students be thinking about doing after
24:41graduation or dare I say instead of
24:43graduation that's question number one
24:45and then related what advice would you
24:47give if you're at Stanford right now and
24:49which is to walk now this Hall do right
24:51now yeah so I used to people that people
24:53used to ask you know should you know
24:55Saul VC go fits examples Mark Zuckerberg
24:56and all these founders have dropped out
24:57and so therefore you know everybody
24:59should drop out and start a company and
25:00people used to ask you know should i
25:02expand should I drop out what should I
25:03do and it used to be a very I used to
25:06feeling like a real moral challenge
25:07answering that question as I felt like
25:08if somebody was meant to if somebody
25:10really should drop out and start a
25:11company and I tell them not to I'd be
25:13committing a moral crime but most people
25:15probably should stay in school and
25:16actually get degrees and I feel immoral
25:18to suggest otherwise so I felt trapped I
25:20thought about it and the absolute
25:23straight advice 100% of the time you
25:25should stay in school finish your degree
25:26not drop out and I've concluded that
25:28because the people who are gonna drop
25:29out and start a company are gonna do it
25:31regardless of what I say or what anybody
25:33else says and so it's a it's it's it's
25:35by definition it's good advice
25:36I can't possibly steer anybody wrong in
25:39general actually not always a good idea
25:41to get the degree the thing that it's
25:42the most underrated right now I think
25:44the archetype / myth of the 22 year old
25:47founder it's been blown completely out
25:50the thing that is underestimated now in
25:52the valley and frankly Stanford is a
25:53grown siguro of this
25:54I think skill acquisition literally the
25:57acquisition of skills on how to do
25:58things is just like dramatically
26:00underrated people are over valuing the
26:02value of just jumping in the deep end of
26:04the pool because like the reality is
26:05most people jump have you been at the
26:06pool drown like there's a reason why
26:09there are so many stories about Mark
26:10Zuckerberg is because there aren't that
26:11many Mark Zuckerberg like most of them
26:12are still floating facedown in the pool
26:14and so for most of us it's a good idea
26:19to get skills you know your degree or
26:21whatever but then there is a lot to
26:23learn if you want to like ultimately to
26:25start a company or go to startup there's
26:27a lot to learn about how companies
26:29operate right there's a lot to learn
26:30about how to manage how to deal with
26:33about how to manage there's a lot about
26:35you know leadership there's a lot about
26:36by the way finance there's a lot about
26:38legal there's a lot about marketing
26:40there's a lot of what sales HR like
26:43there's a whole skill set like if you if
26:45you meet you know the really great CEOs
26:46if you spend time with them and if you
26:48would find this to be true of mark today
26:49or of any of the great CEOs today or the
26:52past like they really aren't cyclope
26:54dick and their knowledge of how to run a
26:55company and it's just very hard to just
26:56kind of Intuit all that in early 20s and
26:58so I think that the path that's makes
27:01much more sense for most people is to
27:02spend five or ten years getting skills
27:03so the problems going to Ross start
27:05about school of sounds great but like
27:06most startups are like really screwed up
27:08like I said most of just died in
27:09obscurity and I don't know exactly what
27:11you learn from dying in obscurity but
27:13it's not very much a lot of people are
27:15at startups that don't work well they
27:16actually don't carry away a lot of
27:17useful skills conversely you know you
27:19leave school you go to a big company a
27:20lot of what you learn in a big company
27:21is how to function at a big company
27:23right but the problem with people have
27:25been a debate company too long is in the
27:27cold light of day when they go out to be
27:28their own thing they literally don't
27:29know how to function without all the
27:30infrastructure and support of a big
27:32company and so I think there's a sweet
27:33spot I can you high-growth company or
27:35the company that's scaling that's
27:36probably the best place to go and of
27:37course your Stanford you have a huge
27:39advantage of being in the environment
27:41you already know who those companies are
27:42and you you know you have a pretty good
27:44chance of getting jobs there so I think
27:46that's generally really good advice the
27:48other thing that I would say is I have a
27:50favorite book I've never read and I'm
27:51actually I'm worried about reading it
27:53because I think it can only disappoint
27:54me at this point because I like the
27:56title so much and the title of the book
27:58is smart people should make things and
28:01like as far as I'm concerned like that's
28:02the entire value of the book like I
28:03don't care what else he said is like
28:04just mmm for engineers it's very obvious
28:07like engineers engineers should build
28:09things should build should build
28:09products and that could be open source
28:11it could be you know working with with a
28:12cop you know with a friend of something
28:13but like going to a company that's
28:15building something but but I think the
28:17same thing is true of everybody else
28:18right and people build all kinds of
28:19things and by the way the things that
28:21people build might be art right the
28:23things that people built might be you
28:24know businesses the thing that people
28:25built might be you know an organization
28:27inside a company or it might be you know
28:29a great you know explanation of
28:31something but tangible output it just
28:33always kind of really encourage people
28:34like when in doubt fall back on building
28:36something tangible yeah and like we've
28:38got that thing at Anderson ours right
28:39like works in practice not in theory
28:40yeah so much stuff that I saw you know
28:42as a scientist a PhD at Stanford worked
28:45in theory but just not in practice and
28:46lots of stuff that's just the converse
28:48and only if you actually build it you
28:49see that yeah why did you and Bend then
28:51decide to start a VC fund rather than
28:53doing another startup
28:54yeah so we were customers of venture
28:57capital or at least I thought about it
28:58that way they thought they were giving
28:59us the money I thought I thought we were
29:01the customer we had maybe occasional
29:03disagreements about that and so we were
29:05customers of venture capital I first
29:06raised venture capital in 1995 with my
29:09partner Jim Clark from jondura who was
29:11actually you know an excellent VC for us
29:13at Netscape and then we raised money
29:14from benchmark in in 99 for a lot of
29:16cloud that went really well and then
29:17between Ben and I we also helped
29:19probably a hundred friends of ours over
29:21the course of sort of a 15 year period
29:23raise venture capital you know we were
29:25angel investors that we would just would
29:26help our friends go through it and so we
29:27kind of you kind of viewed like I was
29:29like almost going to save the partner
29:30store every day for 15 years or
29:31something after a while you're like you
29:33know I think maybe I could do this and I
29:36think maybe I have a few ideas from
29:38being on that side of the table so we
29:40started really thinking about entering a
29:41business and then we thought really hard
29:42about you know the traditional way to
29:44enter venture capitalist to join an
29:45existing firm because the history of
29:47venture capital is that the successful
29:49firms have all been around for 30 or 40
29:50years and we consider that and then we
29:53basically got bit by the startup bug me
29:54for the four and a half the time and we
29:56decided that it was actually a good a
29:57good idea for a startup we spent about a
29:59year and a half actually thinking about
30:01andreessen horowitz as a startup and we
30:02spent a lot of time studying the models
30:04and talking to people who had been in
30:06the industry for a long time and we
30:07ultimately resolved on what we thought
30:09could be two big differences one was
30:10actually a little bit of a Back to the
30:11Future thing which is we decided that
30:13the general partners and increase in
30:15Horowitz would all be people who have
30:16been founders or CEOs or both of tech
30:19startups and that kind of sounds like it
30:21might be obvious like if you're gonna if
30:22you're gonna have somebody in your board
30:23and they're gonna give you advice on
30:24what to do in your company but maybe it
30:26would be helpful if they had actually
30:27done it before it actually turns out
30:29first of all it had been a good idea in
30:31the 60s and 70s the top pcs in the 60s
30:33and 70s when venture capital was created
30:35had for the most part all been operators
30:37and they had been legendary characters
30:39Jean kleiner had been one of the
30:40famously one of the Fairchild one of the
30:42original fairchild people one of the
30:43famous traitorous eight left Shockley to
30:46start Fairchild left Fairchild to start
30:48Intel Tom Perkins had actually been a
30:49general manager at Hewlett Packard which
30:52was actually at the time a source of a
30:53lot of the CEOs of the new companies in
30:55the valley and actually himself had been
30:57a founder he started a laser company and
30:59which was the kind of thing
31:00did in the 1960s and he actually raised
31:02venture capital himself and was a
31:04founder down Valentine you guys had I
31:06think Mike Morris here last year the
31:08founders of Sequoia Capital wrote
31:09Valentine Pierre Lamont
31:10both of whom are famous chip executives
31:13and entrepreneurs and so it actually was
31:16how venture capital got formed our
31:17analysis was basically over the course
31:19of time venture capital added that
31:20traditional venture capital firms had
31:22evolved where the successors to the
31:23founders were were in many cases very
31:25successful investors but were people who
31:27had not started and built companies
31:29themselves and so we kind of decided to
31:30bring that idea back the other big idea
31:33that we had that we've really pushed
31:34hard is the idea of giving founders and
31:37especially founders who have not been
31:38CEO before we used the term sort of give
31:41the founders superpowers in the form of
31:43basically the world's best network and
31:45this is an observation that we made when
31:47we you know we've seen over the years
31:48we've seen founders start companies and
31:49then at some point the founder gets
31:51fired and you bring in a professional
31:52CEO one of the questions we've always
31:53had is what what's the catalyzing thing
31:55that causes the founder to get fired and
31:57then what is the newt what is the
31:58professional CEO then professional CEOs
32:00always it's always a type right it's
32:02always like you know square shoulders
32:03blue suit six-foot-two
32:04great hair fantastic teeth like it's a
32:07type and what do these professional CEOs
32:10have that the founders didn't have and
32:11actually some of it is they have
32:13experience running a company and we
32:14think we can help with that but the
32:15other part is they have these in the
32:18industry for 20 years longer they've got
32:2020 years worth of basically network
32:22built up right and so they know
32:24customers and they know other investors
32:26and they know all the big tech companies
32:28and if the company used to get sold they
32:29know all the buyers and they know all
32:30the reporters who cover the space and
32:32they know all the if it's a regulated
32:34business they know all the government
32:35regulators and so they just they have
32:36these these giant networks that they
32:37built so we decided to do in our firm is
32:40basically essentially pre builds the
32:42best possible network that any that any
32:43startup could have and then basically
32:45let our founders plug into it and
32:47basically get the super power of having
32:48a giant network the way that we did that
32:50is we actually have we have a very kind
32:51of non-traditional structure we have
32:52full-time professionals in our firm who
32:54are not general partners or investing
32:56partners who are operating partners in
32:58six teams that build and run networks
33:01customers investors acquirers executive
33:04talent engineering talent PR and now
33:08policy and regulatory affairs so we've
33:10got 85 people in the office every single
33:11day and what they're doing is they're
33:12basically building and grooming
33:13now we're a network on behalf of the
33:15firm which that works on behalf of all
33:17the portfolio companies and snart's is
33:19actually a network as a service
33:21yeah so then one one interesting point
33:24II 16z was actually started in you know
33:26Oh 809 it's been like seven years now
33:28right and industry has changed you know
33:30the firm is change VC more broadly has
33:32changed what are your thoughts on kind
33:34of that evolution I would say there's
33:36been more change there's no more change
33:38in venture capital in the last seven
33:40years than probably in the preceding
33:41twenty and I'd also argue there's
33:42probably been more change in the tech
33:43industry in the last seven years then
33:45probably the preceding at least 15 or 20
33:47there's a bunch of new firms now that
33:48people are starting that are exciting
33:50another thing as seed investors a seed
33:52investors angel investors have always
33:54been important like a big part of the
33:55history the valley is the willingness of
33:57people who have made you know some
33:58amount of money to write a check and
34:00sort of fund the next idea and you know
34:02a lot of the original a lot of the
34:03original companies in the valley there
34:04was angel money involved so I mean
34:05angels have always played a very
34:07critical role in the last seven or eight
34:09years it feels like a lot of the Angels
34:11actually have become perfect for
34:12professionalized and when they do that
34:14they renamed themselves angel investors
34:16to seed seed investors because angel
34:18kind of implies an individual whereas
34:19seed kind of represents a sort of
34:21investable asset class and so a lot of
34:23the best angels have now actually raised
34:24funds instead of just investing out of
34:26their own pocket and they actually run
34:27these seed firms and so actually we see
34:29kind of a restructuring happen in the
34:30industry where a lot of companies
34:31companies used to just raise venture
34:33capital as their first round they were
34:35just go straight and raise a Series A
34:36and you could either raise a Series A or
34:37you couldn't but it was only a very
34:39small percentage of founders could raise
34:40an a round right out of the gate you
34:42know these days it's much more common to
34:44raise the seed round you know raise five
34:46hundred thousand or a million or even
34:47two million dollars as a seed round and
34:49then go for a year or two or three well
34:51before you actually have to raise full
34:52venture capital in fact the the seed
34:55phenomenon has now gotten so widespread
34:56that now the seed investors are trying
34:58to differentiate against each other so
34:59now there's another seed there's also
35:01precede there's also the seed extensions
35:04there's post seed there's early a and
35:07then actually below all of that there's
35:08the incubator accelerator
35:09kind of phenomenon that so will actually
35:11some will actually sometimes meet
35:12companies that have raised like five
35:14rounds of seed capital in different
35:16forms and so there's just a lot more
35:17support in the infrastructure for a much
35:20larger number of new companies I think
35:22that maps to what's happened in the
35:24industry over the last seven or eight
35:25years which I think has really
35:27remarkable and either we're just taking
35:28it for granted or we haven't really
35:29wrapped her heads around it which is the
35:31valley the history of the valley for you
35:33know 50 years from so 1960s through the
35:35mid-2000s you know the valley was kind
35:37of the best place in the world building
35:38you know literally computers but so
35:40chips and then computers and then
35:41software that runs on computers but
35:44fundamentally building tools right
35:45computers or software as tools and then
35:47we would you know these giant companies
35:49Oracle and Sun and Cisco and and so on
35:51would build these great tools and then
35:53would sell them to customers and the
35:55customer might be a consumer at home but
35:57the customer more often was a big bank
35:59right or a big insurance company or you
36:01know a hotel chain or somebody like that
36:03our car company in the last seven or
36:06eight years post the financial crisis
36:08something has changed either the valley
36:09is about to grow to become a lot bigger
36:11and more important than the valleys ever
36:13been or we are completely smoking crack
36:15many valley companies still build
36:16technology and sell the technology as
36:17tools but a lot of the best new valley
36:19companies build technology and use it as
36:20a wedge to enter an end market right and
36:23so as an example the predecessor company
36:26to uber was not you know a ride-sharing
36:28service that failed the predecessor
36:30company was a little boutique software
36:31company that built dispatch software
36:33that got sold to taxicab operators right
36:35and there actually were companies that
36:36were in that business
36:37it just it was a tiny little business
36:39because it turns out taxi cab operators
36:40actually aren't that excited about
36:42adopting new technology they don't live
36:43very much IT they don't buy very much
36:45software if they did buy software they
36:46wouldn't know what to do with it
36:47and so that was just never a very big
36:49business and so uber and lyft just come
36:51in and basically say let's just do it
36:52let's just provide the ride let's take
36:54completely complete responsibility for
36:55the customer service Elon Musk of course
36:57has pushed this to its logical
36:58conclusion which is you know one I just
36:59built the car I think that you UNK is
37:02tremendous credit both for the car
37:03company and the rocket ship company both
37:05of which are things that nobody 10 years
37:06ago thought was possible to build either
37:08kind of thing as a new company and it
37:09turns out that it is it feels like the
37:11valley is really expanding basically
37:13certainly expanding an ambition and
37:14quite possibly we believe expanding a
37:16capability to be able to actually go
37:17directly into a lot of markets that
37:18historically you would have viewed as
37:20you know much more the province of
37:21existing banks or existing car companies
37:23or existing incumbents
37:24I think a big part of that is actually
37:26the fact that if you're selling IT to
37:28somebody versus actually using it
37:30yourself you can just recognize the
37:31benefits you know more obviously like oh
37:33if you've got your entire thing in a
37:34database well you can push out like a
37:37report of all ride times and so on and
37:39so forth and they can understand and
37:40but the customer wouldn't necessarily do
37:42that it's a major efficiency shoe
37:43selling technology to a company that's
37:45in implementing it's a layer of
37:46indirection and there are companies I
37:47mean look there's you know Oracle got
37:48built to do this and a lot of horrible
37:49customers have gotten great results with
37:51Oracle and salesforce.com just had a
37:52great quarter and sure you know they you
37:54know they they sell their stuff to lots
37:55of companies with big sales persons who
37:57do great with it so it works but yeah
37:59there we see this we have this or take
38:00the term we use this full stack which is
38:02we sort of see there's a particular
38:03magic exactly to biologies point there's
38:05a magic that kicks in when you actually
38:07have complete responsibility for the end
38:09customer experience and how the product
38:11or service is delivered and then
38:12especially these days right in the in
38:14the in the era of big data and machine
38:15learning and all these things there are
38:16things that you can do to optimize both
38:18experience and then ultimately the
38:20economic model of the business it's
38:21become a very open question or its topic
38:23okay so how many industries are opening
38:25up where you could possibly do you know
38:27the equivalent of an uber Airbnb or a
38:29Tesla in these industries from the
38:31I guess questions yeah hi so for a first
38:35time founder who's bootstrapping a v1
38:38product when do you think is the most
38:39appropriate time to first improve
38:41approach investors and at what level is
38:45having a business plan and a team
38:46reasonable prototype to show potential
38:48or demonstrably customer traction thank
38:52you yes so it's hard to give general
38:55advice because it really depends but
38:57unquestionably it's better to have
38:58something working coming in with
38:59something working as a gigantic edge
39:01over coming in with nothing working
39:02right like a huge edge even by the way
39:04for people who have done it before
39:05people who have successfully had ranked
39:07companies before coming or something
39:08working is a really big deal and then it
39:10is like absolute magic I mean it's like
39:12catnip to VCS if you can walk in and
39:14you've already got both the product and
39:15customers just rub it on us it'll drive
39:17us crazy this is another thing probably
39:19what's overestimated right now it's just
39:20raising lots of money to be able to save
39:22raise lots of money probably was
39:23underestimated is the bootstrapping
39:24process of getting in position with the
39:26core thing that you're doing and both
39:28the product itself and its value to
39:29customers before you start raising a lot
39:31of money and with that customer traction
39:34MVP already like what level angels seed
39:36a if you're a first time founder first
39:39time founder it's almost always better
39:40to start with angels or with the early
39:41seed investors it's again contrary to
39:43myth an archetype it's very hard for the
39:45first time found at a rate straight
39:46raise raise and straight around it's
39:48almost always the case that they're
39:49coming up through through seed I mean as
39:51an example you know Mark Zuckerberg
39:52raised literally angel
39:54computer tiel was how he got started he
39:55didn't go up he didn't go and raisonné
39:57out of the gate the Sergey and Larry
39:58same thing they raised angel money and
40:00so I think that that's almost always the
40:02best thing for a first-time founder
40:03thank you yeah you mentioned all the
40:05progress in AI a new input output and
40:08all the language process so I have a
40:10very if you have to pick in third years
40:14what's the chance that we have a bot
40:16that does a better job in picking
40:18companies than Andrews and all this
40:23because it'll be the last investment we
40:25ever make so I mean this idea is this
40:30ideas out there right and so there are
40:32actually people literally trying to do
40:33this and there's actually a venture firm
40:34called correlation ventures that
40:36literally is trying to do this or a
40:37version of this and then you know
40:38there's there people who are like data
40:39mining angel list and trying to figure
40:41out how to do this and there's there
40:42other people who are going about this
40:44the computer science computer scientist
40:47in me with engineer me would like to
40:48believe this is possible and I would
40:49like to be able to figure it out I
40:50frankly like us to figure it out
40:51the thing I keep running up against the
40:53cognitive dissonance in my head that I
40:55struggle with is what I just see in
40:57practice talk about in theory versus in
40:59practice like in theory you should be
41:00able to get the signals like you know
41:01founder backgrounds and this and that
41:03progress against goals or whatever a
41:04customer satisfaction should be able to
41:06measure all these things well we just
41:07find what we deal with every day is not
41:08numbers right there's nothing that we
41:11quantified what we deal with every day
41:12it's idiosyncrasies of people and under
41:14the pressure of a start-up like
41:15idiosyncrasies that people get magnified
41:17out to like a thousandfold like people
41:19become like the most extreme version of
41:21themselves under the kind of pressure
41:21they get under in a syrup and then and
41:24then that's either to the good or to the
41:25batter both but people have their own
41:27issues and then the interpersonal
41:28conflicts between people so the day job
41:30is so much dealing with people that
41:32you'd have to you'd have to have the AI
41:34bought that can like sit down and do
41:37maybe yeah I mean like my guess would be
41:40worse still wears off yeah like just add
41:42to marks the point of that I mean the
41:44fundamental issue from like a machine
41:45learning champion is and you have a very
41:48few events that are most your returns
41:50which are like these Facebook like
41:51outcomes right and so it's like almost
41:53like a rare event detector like the
41:55Large Hadron Collider right you've got
41:57all these particles coming through and
41:58you have to be able to predict okay
41:59which one of them is actually gonna make
42:00a lot of money that's number one number
42:01two is especially at the very earliest
42:03stages you don't have features in the
42:06traditional sense like you don't have a
42:07lot of really good data to work with in
42:09terms of prediction so the later it gets
42:11probably like Series C or thereabout you
42:13have enough you know systematic data to
42:14work with but early on it's actually
42:16pretty challenging yeah hi thank you
42:18um how are you guys thinking through
42:20your fund structure and the types of
42:21investments that you have to make as you
42:22raise more money and can you convince
42:24you be like a winner take all market
42:26there there are a bunch of challenges to
42:28it the central challenge is any top-end
42:30venture capital firm that has a
42:31reputation that it wants to maintain
42:32which is I think very important can only
42:34invest in one company in a category you
42:36can't practically speaking invest in
42:38competitors the company you've already
42:39invested in will feel it's betrayal if
42:42you invest in the new one and then the
42:44new one will think if you're willing to
42:45invest in them you must be very
42:46dishonorable that you're willing to
42:47betray your previous one so it just it
42:49doesn't work and so like the minimal
42:51number of venture capital firms has to
42:52be the number of firms required to fund
42:54the number of competitors right in each
42:56new market and and then we can debate is
42:58that 3 or 5 or 20 or 40 or 100 and you
43:02know certainly that we have too many
43:03venture capital firms like there we've
43:04got like 500 venture capital firms of
43:06the US and certainly there aren't 500
43:07competitors in every market or at least
43:08there need to be at least a half dozen
43:11dozen you know 15 you know good firms to
43:13fund to fund the competitors we would
43:16love to make venture capital on
43:17I have a question regarding block
43:20chaining and like the financial services
43:22industry so it seems like there's a lot
43:23of like low-hanging fruit and a lot of
43:24like far-fetched ideas that on one could
43:26foresee using blockchain now some during
43:28like um like how are you like what
43:31advice would you give for someone who's
43:32trying to see like what is the best I
43:34guess niche area to target um when
43:36you're given such a wide array of like
43:38potential use cases for the blockchain
43:39yeah so we actually shy away from giving
43:43advice like that so there's two reasons
43:45for it so one is there is a concept
43:46called product market fit right which is
43:48we become very fairly publicized now
43:50right product in the right market
43:51there's another concept bulk we call
43:53founder market fit which is is the
43:55founder of a company is that the person
43:57who is born to do that idea and so that
44:00question we tend to defer to the
44:01founders because we figure the really
44:02great founders are gonna figure that
44:03like part of what makes it louder
44:04greatest are gonna figure that out the
44:06other thing we found is that it's very
44:08hard if we have ideas for companies we'd
44:09like to fund but we try not to talk
44:11about them too much because we don't
44:12want somebody we don't want founders to
44:13pick up somebody else's idea and it goes
44:16back to what Balaji said which is it is
44:17so hard to make a start-up work you have
44:20to be so irrationally committed to it I
44:21mean people this is another thing like
44:22startups are over glorified in the sense
44:24of like people think they're fun like
44:25they're not fun like they're not even
44:27remotely fun like they're punishing is
44:28help I think is Bill Lee for Alamos it's
44:30like chewing broken glass it's like
44:32bring you to the best starting companies
44:34like cheering broken glass is like after
44:36a while you start to like the taste of
44:37your own blood yeah it's very different
44:39quote yeah but like it's so hard and
44:42it's so hard because people are saying
44:43no to you all the time it's just no no
44:45no no constantly being told no and you
44:48know and your idea is stupid and like I
44:50would never do that or I don't but do
44:51that so their company's gonna kick your
44:52butt I like it and your lead engineer
44:53quits and like there's knit it just like
44:55endless it's got to be an idea that they
44:57feel so deeply about if they're what
44:59goes to like Balaji Sturm ideological
45:01mission it's got to be something where
45:02people feel so deeply that they have to
45:04do it if they're willing to tolerate the
45:05oval of pain and in our experience most
45:07people aren't willing to tolerate that
45:08level of pain for somebody else this
45:09idea and so I respectfully decline
45:12answer the question okay I see you know
45:14it just seems like for blogging like
45:15there's there's so many use cases and
45:17like for many of them like the timing
45:18could be completely off whereas for
45:20example like for like remit his payments
45:22like one could easy to see how that's a
45:23very easily applicable thing like use
45:25case a block to it so yeah basically I
45:28think that remittances are to Bitcoin
45:30what VoIP was to the Internet in the
45:32sense of it'll work at some point in the
45:35first five years or ten years of the
45:37thing it's not high enough quality with
45:38the obvious alternative namely VoIP
45:40forces landlines or remittances versus
45:42legacy or immense system to win I think
45:44that you know Bitcoin like Bitcoin as
45:47opposed to blotchy of a Bitcoin is good
45:49for transactions that are very large
45:50very small very fast very international
45:52or very automated and you have to try to
45:54envision transactions that are like two
45:56three four or even more of these kinds
45:58of things to think of things that cannot
46:00be done with the current system if you
46:01think of things that cannot be done the
46:03that are still useful well then that's
46:0510x right so that's one way to think
46:07about it the other great thing about it
46:08is like Evan Williams's thing which is
46:09sort of vague but it's actually very
46:11useful so on the one hand Oh a new
46:13technology 10x something what people
46:15haven't done before on the other hand
46:16Evan Williams this thing is take a
46:17behavior that humans want to do and
46:20allow them to do it faster better
46:22cheaper over and over take something
46:24that was once a rich man's thing and
46:25make it accessible to the middle class
46:26or take it for the middle class and make
46:28accessible to everyone right and so if
46:30you kind of combine those two things the
46:32technology allows you to go and in a way
46:34that was not possible so I'd seem you
46:36know hunt in that general area that
46:37might be something thank you
46:39I co-founded two companies that faded
46:41into obscurity too quickly you identify
46:43problems and issues and opportunities I
46:45might take a startup you know weeks it's
46:47not months if not years identified and
46:48I'm kind of curious why injuries and
46:50Horowitz and others don't explicitly
46:52identify opportunities and in problems
46:55or even issue challenges or competitions
46:57then so one and delve a little bit
46:59deeper one of the things you've been
47:00talking about apology more specifically
47:02is like the cloud versus the land and
47:04and being create you know software you
47:06in the world like the divergence of the
47:07cloud and I'm kind of wondering in that
47:09world where ownership seems to be more
47:11centralized there's there could be some
47:13risk associated that associated with
47:15that and wonder if you could speculate
47:16about ownership in the future the
47:19interest especially talking from a
47:20blockchain perspective on asset may be
47:22over so on so it kind of there's two
47:24separate questions Aaron I think the
47:25first one is why doesn't VC pursue like
47:28an X PRIZE style model that's one and
47:29then number two is what happens with
47:31like the future of ownership right kind
47:32of interrelated so the first one
47:33actually gonna be a very interesting
47:34model for a fund the reason I think
47:36that's interesting is one of the points
47:38mark made is and it's one the most
47:39counterintuitive points about VC no
47:41matter how innovative it is an idea that
47:43comes across your doorstep today
47:44there'll be two more like it my best
47:46example of that is Hyperloop right like
47:48stuff the company comes across our
47:49doorstep and like a few weeks later we
47:51have like two more that come in there
47:53and so what it means is that VC is all
47:54about filtering winner-take-all so the
47:57more that you can kind of push the
47:58tournament to inception the more you can
48:00push the tournament earlier and earlier
48:01before you invest the better so prize
48:04model I think could work the problem is
48:06of course grading the prizes judging the
48:07prizes is all the type of stuff that's
48:09one the number two in terms of the
48:10future of ownership I do think that
48:11basically the interface to every
48:14be ultimately digitized in the sense
48:16that you won't own a car you won't have
48:18you already don't have a book show of
48:20Kindle right and you don't have a house
48:21you have near B&B and so on and server
48:23and all that so it becomes extremely
48:24mainstream and what that means is that
48:26actually your mobility is vastly
48:27increased and right now we think of
48:29mobile as oh I can just go to Starbucks
48:30and I can work from there and says much
48:32as I could work at home but I think in
48:33the next five 10 years it's gonna be as
48:35easy to just jump up and move to another
48:37country as it is to just go down the
48:39street what that means is the more
48:40internationally flexible you are so one
48:42of the big aspects of that by the way is
48:43the bank accounts that impacts a
48:45blockchain aspect one of the big things
48:46it's a pain moving between countries
48:48your Gmail works your Facebook works all
48:49your internet services work
48:51those are IP based right but all the
48:52nation-state based things like your bank
48:54account are not quite as portable or as
48:56easy and so those kinds of things I
48:58think it's useful to identify all the
48:59prerequisites so as a thesis for kind of
49:01startups to look at chop the things that
49:02anchor people tool and I think you'll
49:04have some interesting things there hi
49:05this question is for Balaji I'm a
49:07freshman studying physics at the
49:09University of Illinois and I was just
49:10wondering what convinced you to continue
49:12on to do a PhD and wore the skills that
49:14helped you on like in regards to
49:16entrepreneurship and whatever you're
49:17doing today so I would not do a PhD
49:20today that's my quick answer so why did
49:23you do a PhD why did I do a PhD because
49:25I wanted freedom in the sense of I
49:27wanted to do math and you know computer
49:30science and so on on my own time right
49:32but what I would have done instead is I
49:34think the single most important metric
49:36for you guys to measure is your personal
49:38runway in Silicon Valley people think a
49:39lot about you know okay how do I get an
49:41exit and get the money on top but they
49:42think much less about how do I minimize
49:44my personal burn so today in the world
49:46it is possible to just find a
49:48jurisdiction that is amenable to your
49:50preferences that is a worm that is safe
49:52that has good internet and it's really
49:54really cheap and so you know what I
49:57would do instead of getting a PhD if I
49:58was just doing it today the first I'd
50:00worked for a year at at Google or
50:01Facebook or at github I would have a job
50:03that permitted remote work I would
50:05sacrifice the advancement to be able to
50:07work remote for the next three years or
50:09so on and I've just saved enormous
50:10amounts of money and live very very
50:12cheaply every year that you work you've
50:14got three years of runway and so that's
50:17actually freedom once you have the
50:19ability to have like three four ten
50:21years of runway and you have the
50:22discipline of the grad student but the
50:25here right so that's what I would have
50:26done instead um so I wouldn't do the PhD
50:29you can I think you can learn and self
50:31learn faster on the internet then you
50:32can you know in grad school I think a
50:34bachelor's degree is fine like you know
50:36like I'm not saying drop out or it would
50:37have you right now at least but I think
50:39you can do better than a PhD today so
50:42I've got a question for biology yeah
50:43so biology for those of you who don't
50:45have my reputation or know I'm tonight
50:48he's done this very big advocate for
50:49entrepreneurship outside the valley very
50:51big advocate for a developing world
50:53entrepreneurship very big of a kind I
50:54still hear in this case for for actually
50:56literally every computer someplace else
50:58I can't help but point out that biology
51:00lives where do you live yeah I know I
51:01unfortunately I'm in San Francisco now
51:07literally like if you drew like a circle
51:09around San Francisco he's like right in
51:10the middle let's say that sometimes you
51:13have a goal that you have it takes a
51:15while to get to because they're a bunch
51:16of prerequisites that have to be met
51:17right I was married and he has a lovely
51:23little baby I do I do that those are all
51:25anchors the two of them are gonna have I
51:27think I think they get votes they get
51:29those it's my understanding how this
51:30works they get to contribute to the
51:33I'm from China I work for Google China
51:36and now a current student in the
51:38Stanford GSB I'm really inspired by the
51:42Entrepreneurship here but I know there's
51:45a lot of a lot challenges for the
51:47immigrant entrepreneurs to start the
51:50company here so I've been one - your
51:51advice for the immigrants entrepreneurs
51:54key especially for the first time right
51:57I'm gonna turn that question over sure
51:59okay yeah sure - the immigrant
52:02entrepreneur on the stage yeah yeah sure
52:03so I thought about this a lot and I
52:05discussed this with with Mark and Ben a
52:07lot what comes after the Durham
52:08entrepreneur is the developing world
52:10entrepreneur and amacrine entrepreneur
52:11and but especially in the developing
52:13world and I think you know one thing you
52:16know depending on what country one is
52:17coming from and so on obviously there's
52:18a wide range but for someone coming out
52:20of India for example frequently making
52:22$100,000 is like making a million
52:23dollars in the sense of like the impact
52:25on quality of life and so on right and
52:27there's actually much lower risk ways to
52:30make $100,000 than to do a start-up
52:32which is just extremely stressful and
52:34you're going for infinity and so on
52:35right and so I think that we're gonna
52:38new kinds of things particularly as you
52:40get another billion two billion people
52:42with cell phones right like that then
52:44we're gonna see new kinds of business
52:45models that are based on knowledge that
52:47folks outside the US and in the
52:49developing world have about their local
52:51economies and also have maybe less than
52:53we have upside more predictable returns
52:55and they're not quite as much of a you
52:57know roll the dice kind of thing in some
52:59ways if you start at zero you you're
53:01it's easier to get to infinity because
53:02you just have no illusion some good
53:04thanks everybody for coming
