00:00welcome to the a 16z podcast I'm Michael
00:02Copeland the old constraint when it came
00:05to technology was hardware
00:07how many CPUs can I get my hands on
00:09today spinning up compute can be done
00:12from any smartphone with an AWS account
00:15or something similar
00:16the current constraint is software and
00:19since software is written and operated
00:21by people tackling that constraint comes
00:24down to making people as informed
00:26enabled and efficient as possible three
00:29CEOs and cofounders of three companies
00:32that serve software developers Chris Wan
00:35Jeff Lawson from Twilio and Ben Uretsky
00:38of digitalocean take part in a
00:40conversation with a 16 ZZZ Peter Levine
00:42about the needs of software developers
00:45what are the emerging platforms
00:47ecosystems and tools that help
00:49developers succeed at what is
00:51increasingly the most important job in
00:53any company writing and running software
00:56Peter Levine begins the conversation we
01:00wanted to hold this panel on the sort of
01:05future of software development and one
01:09of the interesting sort of rationales
01:12behind the panel was I I often look back
01:15and it's a software development has
01:18software development prior to let's say
01:21five years ago there's always been
01:22software development as long as there's
01:25been computers but it feels to me like
01:27after the over the past five years
01:30software development has become a market
01:34in and of itself like there are many
01:36companies that get built to serve
01:38software developers and if you look at a
01:40number of early stage and mid stage
01:43companies they often start out selling
01:46to the developer and that's a very
01:48unique phenomenon in sort of the
01:51industry today as different from several
01:54years ago and what we wanted to do is
01:57bring up companies github Twilio and
02:01digitalocean companies that have that
02:04particular focus so maybe I'll start out
02:07with Chris on that and you know Chris
02:12you know software development has been
02:14around forever why is this you know
02:16what's so new and exciting about now
02:18I think software I mean and recent a
02:21couple years ago had the software is
02:23eating the world piece and I don't think
02:25that even begins to touch the
02:26transformation that we're seeing I think
02:28you could make the case that the 1900s
02:31were all about hardware and so when you
02:32wanted more processors when you wanted
02:34more bandwidth you just built bigger
02:35factories or more assembly lines now
02:37we're getting to a place where you can
02:38on your phone go get CPUs and bandwidth
02:41and the real constraint is software and
02:43really you can't order more software
02:45what you have to do is get more people
02:47to create software because it's all
02:48powered by humans so I think that's
02:50what's really interesting about our
02:51businesses and a lot of the
02:53transformation right now is we're going
02:54from a world where we're trying to
02:55create more physical goods to widen the
02:58amount of processing we can do to we
03:00need to figure out a way to get more
03:01people involved and make the people that
03:03are currently involved more productive
03:04because ultimately the world is thirsty
03:06and hungry for software and the only can
03:08get more of it is by creating more
03:09developers you know so then you guys at
03:14digitalocean focus on the developer
03:17what's unique about building a company
03:19you guys have deliberately focused on
03:21you know sort of serving the developer
03:24what's unique about what you do versus
03:27others and how do you sort of how do you
03:29service them well just to kind of
03:33piggyback on on Chris's point the the
03:35social aspect is very important to note
03:38here so building communities of software
03:41developers turns out as a is a very good
03:44way to engage a much broader audience
03:46and one of the things that we've done is
03:49not only focus on software developers
03:51but build a community at at the heart of
03:54digital ocean so to date we've published
03:561,500 tutorials that talk about how to
03:59leverage open source software we're
04:02educating developers worldwide on how to
04:04write code how to deploy applications
04:07and we're drawing in four million unique
04:10visitors a month to this community and
04:12it's growing month over month and I
04:14think it just speaks to the excitement
04:16that people have over the ability to
04:18create you know from from the ground up
04:21you can literally take a vision and and
04:23deliver it that's that's the power
04:26software and that's why it's attracting
04:27so many people obviously it's a great
04:30job market to be in and it's also very
04:33much a level playing field with with the
04:36internet so a software developer you
04:38know somewhere in asia-pacific is
04:40technically as capable and as as
04:43powerful as a software developer in in
04:45the US so one of the things that we
04:48realized is this intersection of the
04:50developer the software that they're
04:52creating and then ultimately what we do
04:54is provide the infrastructure that it
04:55runs on and bringing very much a social
04:59aspect to what we're deeming in the near
05:02future collaborative infrastructure and
05:04really empowering software developers to
05:06figure out the best way to run
05:08infrastructure in the cloud because at
05:10the end of the day what they actually
05:12care about is the application and not
05:14the resources that are behind it so
05:16we're trying to make it as transparent
05:17as possible you know when when I see the
05:21three companies up here I just want to
05:23kind of frame it out a little bit github
05:26is where you build software develop and
05:29build software digitalocean is where you
05:31deploy software and Twilio is how you
05:35communicate between software so maybe on
05:39that GF how do you how do you guys look
05:42at sort of startup companies versus
05:46legacy companies in terms of their
05:48development processes and are there
05:50differences that you see yeah I mean I
05:53think every company is becoming a
05:55software company or else they're not
05:57going to compete effectively yeah and
05:59and every company is figuring this out
06:01what's fascinating is that software in
06:04most companies has moved from you know
06:07the back office it used to be you had
06:08ERP e and there were people with
06:10spreadsheets and that was the extent of
06:12how software was used in your business
06:14to now software being the business right
06:16and you see this with with banks but a
06:18Bank of America is now closing branches
06:20and moving into the app and I think the
06:24fascinating trend that has happened is
06:26that you know it used to be when we
06:28started to leo people said well you know
06:30why are you going after developers they
06:31developers on a market developers don't
06:34you have a checkbook they don't buy
06:38Metsu sorry there's no money in
06:40developer tools I hope yeah I heard that
06:41yeah the interesting thing is if people
06:44who said that we're really thinking back
06:46to this age like 1990 right when every
06:49software that a company used caught you
06:52know cost you know three million dollars
06:54and took three years to install it was a
06:57big deal and you bought all your
06:58software from Oracle and Microsoft and a
07:01few others and in that world you know I
07:03like to call high-stakes software right
07:05CIO is bet their careers on some big
07:07Oracle implementation right and in those
07:10days yeah the developer didn't cut the
07:11three million dollar check absolutely
07:13true they just worked with what someone
07:15told them to do but the big thing that
07:17has changed similar to like you think
07:20about well software as-a-service came
07:22along in the early 2000s and you had
07:25essentially multi-tenancy
07:26changed the game because now you could
07:28cost-effectively onboard thousands or
07:30millions of customers and every
07:32deployment wasn't a one-off well that's
07:34now migrated to the developer where
07:36multi-tenancy and api is have come to
07:39play so that a developer can come and
07:41self service provision infrastructure or
07:43any other form of component of a
07:47software application and they're not
07:49spending millions of dollars they're
07:50spending a dollar right and it's not
07:53taking thirty six months it's taking
07:54thirty six minutes to build the
07:56prototype and so this is putting the
07:58developer now in control right because
08:00while other parts of the business might
08:02be debating which vendor is going to
08:03take the three-year contract the
08:04developer in less time than it took to
08:06schedule that meeting has built the
08:08prototype and it's showing it off and
08:10I've literally had developer customers
08:12of ours say thank you so much you made
08:15me into a hero I walked into a meeting
08:17with the CEO of my company and you know
08:20some code that I had hacked together day
08:22the day before in this demo yeah I made
08:24everybody in the room phones ring right
08:26to show off what we could build and
08:28people are like oh my god this amazing
08:30like I didn't even see the the seventeen
08:33million dollar budget for this and if it
08:35developers like you know yeah I built it
08:36yesterday yeah and that's the thing
08:38that's changing and we also say that you
08:40know the the prototype is worth a
08:42thousand sales calls right and that's
08:44what the developer has that's the magic
08:46that developers now have is that they
08:48can just waffle ones debating the merits
08:49of this and that and which
08:51and someone's out playing golf with
08:52someone else you know the developer is
08:54just like here I I solved the problem
08:56and that's happening at companies big
08:57and small yeah yeah that's it really is
09:01a transformation in the industry along
09:04those lines Chris you know by the way
09:07all of these folks up here are all the
09:10founders and CEOs of their companies so
09:12it's really interesting I mean they all
09:13started when the company was just them
09:16and they've built very large
09:18organizations around them so Chris along
09:21those lines like you guys started with
09:23this thing called git and then you added
09:26hub and became github and like what is
09:29that and what is what's get and what's
09:31hub and kind of how'd it all get started
09:34yeah I read a blog post once that was
09:37like all you need to do is find a unix
09:39tool and make it into a website and
09:41you'll be rich so I was like okay well
09:43get is a UNIX tool we'll just make into
09:44a website no not really but git is a
09:46version control tool that was written by
09:48Linus Torvalds who created the kernel
09:49and it's super good super fast
09:52it sort of unlocks a lot of different
09:55workflows and it assumes that there's a
09:57network connection and that's the
09:58professor thing to me personally and I
10:00think to every developer in that there's
10:02almost as paradox where we are working
10:04on computers all day but it's very hard
10:06for us to work with each other and this
10:07is why you see I think things like polio
10:10and even slack blowing up is like
10:11finally there's an easy way for me to
10:13work with someone sitting right next to
10:15me and for me that was what git was all
10:17about is I had this tool I could easily
10:19see the history and it gave me a very
10:21Wikipedia like control over my codebase
10:24but there was still this huge hurdle for
10:26me to work with someone who was sitting
10:27right next to me and so what github did
10:29is it really created that hub is it
10:31built on top of open-source software all
10:33the existing workflows all the open
10:35protocols but it gave you sort of a
10:36communication platform to coordinate and
10:38just really discuss what you're working
10:40on with someone whether they're next to
10:42you or across the globe so we call it
10:44github and it is really we think a hub
10:46for people using git or maybe that's
10:49what it was originally it's evolved into
10:50much more and maybe this is something
10:53that everyone learns as they've worked
10:54with the internet for a long time but
10:55the communication part is the hardest
10:58part and the most important part and the
10:59technical part you have unlimited amount
11:01of time to figure out later and so we're
11:03really focused on the communication
11:04but we're still super involved in the
11:06community and so really what we've done
11:07is we've created a social network
11:09looking to Facebook Twitter all the
11:11prior art around developers and
11:13developer workflows and so we heavily
11:15make use of and participate in open
11:17source communities but we are our own
11:18business and that's where it get and
11:20github sort of collide yeah Jeff how did
11:23you guys pick communications as sort of
11:25the area to win developers and why you
11:28know Chris just alluded to that and
11:30maybe explain kind of some of the API
11:32interfaces and that that you guys have
11:34but communications seems like a
11:37interesting place to start yeah yeah it
11:39wasn't it wasn't like we said we don't
11:41start a company for developers and like
11:42here are all the different things we
11:44Tullio really came from frustration of
11:47mine of things that I had wanted to
11:49build so I'm a serial entrepreneur my
11:50developer the serial entrepreneur had
11:52started three companies before Twilio it
11:54had been at AWS as a product manager uli
11:57um and when I left I was looking at a
12:00wide variety of ideas you know things
12:02had nothing to do with developers
12:03there's a sort of thing about like what
12:03are problems that are interesting and
12:04one solve and I had this realization
12:07that and every single one of my previous
12:08companies at one point or another we had
12:11needed communications and not like you
12:14know the desk phone for the employee
12:15communications but actually
12:16communications that was a part of our
12:18application part of our workflow part of
12:21our customer experience like something
12:22that was like deeply integrated into the
12:25software product that we were building
12:26and and every one of those companies I
12:29said well I'm a software developer or so
12:31I don't know anything about
12:31communications you know let me go find
12:33out how to build this and you go talk to
12:35the industry and industry you know
12:37telecommunications industry and they'd
12:38say oh yeah yeah we can help you build
12:40that we're gonna you know bring in all
12:42these lines from carriers to your data
12:43center we're gonna rack up all this
12:45telco gear and that's gonna be a four
12:47million dollar project and it's gonna
12:49take three years to complete with an
12:50army of professional services yeah sign
12:53here and I said like first of all we
12:56were startup so I wouldn't have the four
12:58million dollars to spend that it but I
12:59also looking at these people saying hmm
13:01yeah I'm a software person in the world
13:04of software you're agile right you are
13:07keeping a backlog you're shipping every
13:09two weeks you're iterating constantly to
13:11find the right solution for your
13:12customers that's the speed at which
13:14software operates and here are these
13:16people saying it's gonna take three
13:17years to deploy this
13:18I'm saying that's insane you know
13:20nothing in software takes three years
13:22right because you're assumed everything
13:24will be different three years from now
13:25and there's no way you ever embark on a
13:27project that looks like that and so
13:29after having that experience three
13:30companies in rail I said huh you know
13:32it's whole world of communications which
13:33is one of the largest industries in the
13:35planet has actually been stuck in this
13:37very hardware centric this physical the
13:40physical items of communications
13:42actually dictate what you can do in the
13:45world but if you think about it
13:47Alexander Graham Bell made the first
13:48phone call 150 years ago about wearing
13:50together a speaker or a microphone and
13:52some copper wire and you know the in the
13:55150 years that have progressed
13:57everything in technology has gotten
13:58amazing right we've invented vacuum
14:01tubes and transistors and integrated
14:02circuits miniaturized it all we've had
14:05copper wires turned into fiber optics
14:06and satellites and and Wireless and
14:09we've wrapped the whole planet like
14:10literally in all this technology yet if
14:13you want to make a phone call the only
14:15change that's really happened in 150
14:17years was dialing with an operator to
14:18dialing with a rotary - dialing with a
14:21push-button now open up the phone app on
14:24your iPhone supercomputer in your pocket
14:27it's still just the touch screen
14:29representation of the same push-button
14:31telephone that John F Kennedy would be
14:33familiar with using right and that's
14:36insane and that's because the world of
14:37communications has been essentially the
14:40tightly coupled to the physical nature's
14:43of that network and so what you could do
14:45with communications is all about yeah
14:47like let's rack more boxes and pull more
14:49wires and all this and you're saying
14:50what well let's bring this into the
14:51world of software let's make it
14:53programmable let's make it agile and
14:56let's make it scale up and down and make
14:58it global and remove the social you know
15:00the political boundaries that telecom
15:02always has and just let's turn this
15:04thing into software and that's really
15:06where we where we started the company
15:07yeah awesome so Ben digitalocean
15:12an amazing story you point to the growth
15:15recent studies show that you guys are
15:18the second largest cloud provider by
15:21number of server instances or instances
15:24that you guys serve out second only to
15:28and even though I'm on your board like I
15:32still find that hard to believe so like
15:33can you explain that like how can it
15:35start up actually come in and grow so
15:40rapidly in the face of let's say Amazon
15:42as your Google you know sort of all the
15:45big guys who you would expect a really
15:48dumb you know win and dominate the cloud
15:50world how does a startup do it
15:51yeah I mean all those companies at one
15:54point were startups tamiya but yeah so
15:57you know from from humble beginnings
15:59Giants emerge for us you know I started
16:04my first company in in 2003 was a
16:06managed server provider and so I was
16:08already in this industry and you know
16:10having done that for eight years you
16:13learned a lot about the competitive
16:15landscape and what emerged is you know
16:18the old world was all about dedicated
16:20servers kind of going back to Jeff's
16:22point where you know we're just racking
16:24and stacking more equipment and then
16:26Amazon came it came out in 2007 and kind
16:29of really brought cloud computing to to
16:31the front and center and the the big
16:34shift there is that as a as a business
16:38you no longer needed to worry about the
16:40metal those resources were now
16:42abstracted and there was a provider who
16:45was able to deliver them to you via a
16:47web interface an API call no more human
16:51interaction required in order to build
16:53out a massive web infrastructure and so
16:58what we what we realized you know
17:01haven't been in this industry for for
17:03nearly a decade is that as Amazon had
17:06grown over time their features that had
17:08also become much more complex because
17:10they were they were serving an ever
17:12larger customer and you know we we kind
17:16of associate and feel that word
17:18developers ourselves kind of the type of
17:20people that would use Twilio or would
17:22commit code to github and actually do
17:25Twilio is integrated into our two-factor
17:28authentication and our entire
17:30engineering team is is on github and so
17:33we you know as high five pair five you
17:38know as kind of I have this tag line
17:42CEO right and I think what we're seeing
17:45is these software-defined companies are
17:46looking at the world in a very different
17:48way and the problem is that Amazon Web
17:52Services or Azure all all of those
17:55businesses essentially started off in a
17:57very different environment they figured
18:00out like in the case of Amazon they
18:02needed to ship software continuously
18:03they needed to iterate in the e-commerce
18:05space and that's what actually led to to
18:08the creation of Amazon Web Services they
18:10were finally able to productize this
18:13internal system and and push it out to
18:15the rest of the world the the problem
18:18became that there was a huge disconnect
18:19with the user base the customer was now
18:22forced to use a tool that was not
18:24technically built for for their use case
18:26it was Amazon's a use case and and then
18:29generalized for for mass consumption and
18:31what we started with is this notion that
18:34if we believe the developer is really at
18:37the heart of the ecosystem going into
18:40we should really put their needs first
18:42which is why we focus on the product
18:44experience or you can you know sign up
18:46with an email and a password in 55
18:48seconds you've already provisioned your
18:50I mean sometimes we have customers and
18:52kind of group studies that we've seen
18:54there they still look at the screen and
18:56we're like what are you doing they're
18:58like well I'm still waiting for the
18:59server to be provisioned and we're like
19:00no it's it's ready you could you can
19:02actually log in and and start using it
19:04and I think you know what this speaks to
19:06is a lot of companies are engineering
19:09driven and it becomes very difficult to
19:11differentiate in this space because can
19:13you truly claim to be you know faster
19:15than than Google or more engineer and
19:18driven than Amazon it's like it becomes
19:20this murky mess whereas no one is really
19:22focused on the on the user and and the
19:25experience that they're that they have
19:27using these resources and and that's
19:30really where where we started first and
19:32foremost and we're trying to actually
19:34abstract this away because at the end of
19:36the day you know the the customer
19:38doesn't care what lies underneath the
19:41application what they want is a reliable
19:43service to get the application out into
19:45the hands of their users to iterate on
19:47their feature set to actually ease
19:49software development and and that's why
19:53platform services have become so popular
19:56you know going back to digital oceans
19:57origin it was putting that user
20:00perspective first and and building the
20:03entire company around the customer and
20:06and you know we are one of the only
20:08pure-play I think we might be the only
20:10pure-play cloud out there digital ocean
20:13does nothing other than infrastructure
20:15as a service so we've been able to
20:17curate this entire experience from the
20:19API with a single call to create a
20:21server single call take a snapshot
20:22transfer to a new region to the web
20:26interface to all the tutorials support
20:28and documentation that we provide and
20:31what winds up happening right the reason
20:33why we've we've become number two is
20:36when a user actually goes through the
20:38process and consumes that resource they
20:42fall in love with the service it's it's
20:44exactly what they expect that it's
20:46performant it's cost-effective and then
20:50they can't help but share that
20:51experience so they'll go on Twitter on
20:53Hacker News write a blog post or even
20:55just tell a friend and even to this day
20:57over half of our customer signups come
21:00through a word-of-mouth recommendation
21:02keeping our cost of acquisition really
21:03low and essentially creating a viral
21:07component to to to our business which is
21:11really unheard of to think about you
21:12know a utility provider in some aspects
21:15to be able to garner a viral Network
21:18effect and I think it just speaks to
21:21once again you know not enough companies
21:23are actually placing enough time on what
21:26the customer needs and what we've
21:29realized is that simplicity is a great
21:31long-term differentiator and everything
21:34that we do is through that lens and I
21:35think that's obviously what what speaks
21:37for the traction that we've been able to
21:39attain it's really in a few unique
21:41dynamics to the developer market right
21:44so big companies who try to market to
21:46developers like I have to imagine in my
21:48head when a big company says oh we have
21:49a strategic idea let's go after
21:50developers Israel and they like send
21:52their marketing and sales teams after
21:53the developer audience like developers
21:55literally like will look at a sales
21:57person and like imagine them in a clown
21:59suit I really just walked in and he's
22:04gonna try to sell us some stuff I mean
22:05like that's how its think about it like
22:07you have to be credible right you have
22:09to be a part of the community you're
22:11serving and it's through that
22:12authenticity that you actually build
22:14your ecosystem your community and if you
22:17are successful at doing that yeah I have
22:20been say the developer market is like
22:21you're doing customer acquisition like a
22:24consumer business like a massive funnel
22:27right lots of developers coming kicking
22:29the tires like a consumer business but
22:32unlike any other consumer business they
22:34have the ability to later spend like an
22:36enterprise mm-hmm like because they know
22:38about your service and all it takes is
22:41the use case to come up at work like oh
22:43we need communications we need service
22:45like oh I know how to do that let me
22:46spend something up with this tool that I
22:48know and now suddenly that can grow into
22:50an enterprise size revenue stream for
22:52the company yeah interesting
22:54I agree we'll talk about business models
22:57in just a second here because I know
22:59many people often ask how can companies
23:02that either run open source support open
23:05source how do they actually make money
23:06and we'll talk about that in a minute
23:08but Chris wanted to get to you on China
23:11there were reports earlier this year I
23:14guess a couple of months ago that github
23:17had a security attack from China
23:20allegedly allegedly that's what I said
23:22apparently allegedly why would anybody
23:26care about hacking into github why do
23:32you allegedly think that somebody would
23:34kill why do I allegedly think someone
23:35would allegedly hack and I think that
23:38you know we're entering this age where
23:40whether you want to write code or write
23:43text or write a poem or share a photo a
23:45lot of people want to put something out
23:47there in the internet and make it
23:48available to everyone and I think when
23:51maybe we're getting used to with
23:53Facebook the idea of posting a comment
23:56or a photo everywhere but it's still
23:58less sort of engrained in the general
24:00psyche that you could post code for
24:02everyone right and when code is still
24:06seen as very much like infrastructure as
24:08data as something very tangible maybe
24:10more so tangible than a blog post or an
24:12idea and so I think what github lets you
24:15do and one thing we've always done
24:16really well and cared about is you just
24:18have a couple clicks and your stuff is
24:19up there and just like the basic YouTube
24:23you sent for an account you can post
24:25whatever you want up there and I don't
24:28know I think that that has been a
24:30challenge for us in that a lot of really
24:32amazing ideas have been posted out there
24:34there's never been more open source than
24:35there is today in the world and I really
24:37believe get up is a big part of it and I
24:38think a big part of that is you know
24:40it's the sharing economy all of this
24:42code was sitting around on people's
24:43laptops little scripts three lines of
24:46code a little enhancement they made to
24:47their JavaScript or whatever and before
24:49there was this arduous process for you
24:51to share it with the world you had to
24:52publish it and I'm not going to publish
24:53the three lines of code that's absurd
24:55but I'm gonna share it I'm gonna put it
24:57in my bucket or on my profile or
24:58whatever right I'm just gonna throw it
24:59over the wall and as that happens a lot
25:01more and more you get a lot of great
25:03stuff that was never visible to the
25:04world before but you also get weird
25:06stuff or bad stuff and so we often have
25:09to deal with people writing code
25:11specifically that's malware people
25:12writing code specifically that helps a
25:14botnet and that's a challenge that we
25:16have to face in particular regarding the
25:18United States laws and so I think when
25:20you start having a global community and
25:22we are only 30% of get-ups traffic comes
25:24from North America you start dealing
25:26with different mindsets and different
25:28governments that have a different
25:30perspective on what is okay to share and
25:32what's not okay to share and when our
25:33our mission is make it as easy as
25:35possible to share everything that comes
25:37in conflict with the way certain people
25:38see the world so I think it's it could
25:42be a github problem we could get into
25:43conversation about China and github in
25:44freedom of speech but I think it's just
25:46larger issue happening on the internet
25:47right now is that everyone's lagging
25:50behind I mean as we talk about the
25:51developer honestly I think everyone in
25:53this room including us we lag behind the
25:55concept of the modern developer of the
25:57new developer we don't have a clear
25:59understanding of what that is yet it's
26:00happening so quickly so the idea that a
26:02state government or a state actor would
26:04be understanding the nuances of sharing
26:06information and what that means and
26:07what's harmful or not harmful it's
26:09ridiculous there are at least 10 to 20
26:10years behind on that so yeah we I think
26:14as a social network as they as a
26:15publishing platform have to deal with
26:17the same sort of challenges that any
26:19publishing platform deals with for us
26:21it's particular to code but I think that
26:24what we're experiencing related China
26:26our lever is not dissimilar from what
26:28Facebook or Twitter or all these people
26:30are facing with I think it's gonna get a
26:31lot worse before it gets better
26:33yeah and it's not about hacking or
26:35stealing information or trying to get
26:37it's about suppressing the speech yeah
26:40interesting so there's often I've often
26:44associated open the open-source movement
26:47as also being sort of developer oriented
26:52meaning that the companies that build
26:54open-source are the ones that are going
26:56to have the most developer traction and
26:59it's kind of interesting we have company
27:02I mean Twilio is not an open-source
27:04company and digitalocean sort of touches
27:08github of course supports all
27:09open-source but isn't really an
27:11open-source company so maybe we can just
27:13talk about the business models I've
27:15often you know we're investors in in two
27:18of the companies up here and you know
27:21part of our analysis and investing in
27:23digitalocean and github was the way they
27:27monetize the use of open source right
27:30open source is a is a very challenging
27:32business model and if you don't get it
27:34right you are chasing a downward spiral
27:38in terms of you know everyone racing to
27:41zero with open source and yet here we
27:43have these incredibly profitable
27:44companies that are two of them and then
27:48we'll talk about Twilio x' model but two
27:50of them that are flying we're not
27:52profitable no no no no I I'm implying
27:54that you're not open source yet you're
27:57appealing to the developer and so how
27:59does that work if you're not open source
28:00so we'll talk about that in a second but
28:02then maybe maybe um you know how do you
28:05guys make your you use open source and
28:08yet you monetize it how do you guys do
28:10that yeah absolutely I mean at the heart
28:12of what we're doing is is virtualization
28:14and today we're using KVM prior to that
28:18we use then and it was a very you know
28:21very concerted effort to ensure that we
28:23don't bring some proprietary closed
28:25source software a because their license
28:29fees involved and we just wanted to
28:30really deliver the most cost-effective
28:33solution to our customers and then be
28:35ultimately we actually want to have full
28:38control over the the environment now we
28:40haven't made any modifications to
28:42virtualization at this point but the
28:43good news is that in the future if we
28:45want to go down that route we have the
28:48capability to do that so
28:50that you know that kind of open source
28:53capability is really what empowers that
28:55it'll ocean from from from its origin
28:58but I think what's even more interesting
29:00is that the virtual machines that run on
29:03our cloud the majority of the software
29:06that's loaded on them is also open
29:09source and so it's you know it's
29:12providing this underlying foundation for
29:16everyone else to run a successful open
29:19source project that makes digitalocean
29:21successful because if you take a look at
29:24the more complex projects out there like
29:27you know Hadoop is pretty complicated
29:29Cassandra some of the Big Data stuff I
29:32mean we offer a lot of like one-click
29:34installations that really simplify and
29:37and reduce the barrier to entry into
29:40those environments and so you're able to
29:43load this open source software onto a
29:46open source cloud and essentially you
29:49know with with the pass-through in terms
29:51of cost so I think we're very
29:54complimentary to to open source a lot of
29:57the work that we're doing right now
29:58we're actually contributing back so in
30:00in the metric space where we're getting
30:02heavily involved with with some of those
30:06open source projects which were using
30:07internally and contributing back but
30:09even more importantly is you know I
30:12think if there was a closed environment
30:15like if we were running VMware for
30:17instance it would it would I'm not sure
30:21exactly the right way to like phrase it
30:22but there's there's a community that's
30:24created around the open source and and
30:27they they kind of impact they they go
30:30hand in hand right so they're I don't
30:33know I can't really finish the thought
30:35but they're they're self reinforcing in
30:38nature and if we were running a VMware
30:39cloud it wouldn't make sense to run
30:41open-source on top of it which is why I
30:44think VMware's like V cloud initiative
30:46hasn't necessarily had the traction that
30:48they wanted to see so kind of going back
30:51to Chris's point you do have to stay
30:53authentic and original with the
30:55developer community so if they're
30:57writing open-source and your cloud is
30:59open in nature and actually I think this
31:00will be our differentiator over the long
31:03and I think about a little bit more
31:04Amazon is known as a closed cloud with
31:08you know vendor lock and is kind of
31:10written all over about in the press and
31:12the idea for us is that we actually want
31:15to empower that open source movement
31:17because it's not the proprietary
31:20implementation around the jewel ocean
31:22it's the full package it's the full
31:23experience that will actually allow us
31:25to to succeed over the long term
31:28so we want to empower open source
31:30projects to become as useful as powerful
31:33as they can on top of our infrastructure
31:35and if you want to rip that out and take
31:38it to Amazon you actually can and we
31:41think by staying open that we will
31:43actually win more customers in long term
31:45because they'll choose us for the right
31:47reasons rather than being locked into
31:48into that environment so Jeff I'll jump
31:52to you now then just made the point that
31:55developers follow open source if you're
31:58an open source developer you want
31:59something built on open source you guys
32:01are not open source yet you have this
32:03incredible developer following and
32:05you're selling your you know you sell to
32:08developers so that feels counter almost
32:12to the momentum in the industry right
32:15now how do you guys how does that happen
32:17well I think I think we're very actually
32:19very similar to digital ocean and github
32:21in that we use open source software
32:24extensively sometimes we release code
32:28back to our customers usually you know
32:30non-core assets that we've built that we
32:33think the community will enjoy but
32:35fundamentally one of the things that we
32:36are all doing as our businesses is
32:40taking open source software and actually
32:42making it so a developer doesn't have to
32:44figure out how to operate it and you
32:46know there's been so much open source
32:48software written and it's getting it's
32:49very sophisticated very complex software
32:51and using that software you know just
32:55writing a thing like that exercises
32:58Cassandra or whatever well it's gotten
32:59pretty easy right but actually operating
33:02in scaling it making it reliable
33:04redundant is harder than ever because of
33:09the nature of global real time always on
33:11cloud applications and so while building
33:14software is getting easier operating it
33:16is getting hurt and that's actually
33:18where the as-a-service world is adding
33:22value to its customers yeah which is to
33:24say you no longer have to worry about
33:27how to scale it how to make it resilient
33:29reliable secure etc you just get the
33:33benefit of it yeah and you get the
33:35benefit of what open-source software or
33:37proprietary software do you get the
33:39benefit of what it promises but we're
33:41the experts and scaling it that's why
33:43you pay us you know we're gonna get the
33:46expertise like no one in the world is
33:48probably better at running get thing
33:49github right so if you want to install
33:51it yourself great but you're not gonna
33:52run you know an instantiation of it as
33:56well as a service provider who that's
33:58all they do and they build up that
33:59expertise and when you think about the
34:01economies of scale that go into
34:03operating these services when you do the
34:06math you're like yeah it just make sense
34:07I'm gonna pay someone who's an expert
34:08who's operating at global scale they're
34:11gonna do it better and more
34:12cost-effectively that I can do it myself
34:13and you know you saw this a lot of
34:15companies jumped on no sequel bandwagon
34:17you know in the two thousand ten eleven
34:19twelve timeframe you know and stood up
34:21their own Kassandra clusters because
34:23they read the vendor arguments about how
34:25fast it was and like at the beginning it
34:27sounded good and then you know boom no
34:30die and hit a bug all the data was
34:32deleted down for 36 hours and it's like
34:35everyone realized oh yeah like there is
34:37no free lunch like operating cloud scale
34:40software is still really hard yeah yeah
34:42and you know in some ways like we're
34:45just taking open source and we're making
34:47it multi-tenant and when making it
34:48scalable and we're making it
34:50yeah so Chris let me you know jump over
34:54most people think github is an open
34:57source company and I remember I wrote a
34:59blog a couple of years ago entitled why
35:03there will never be another Red Hat like
35:05I really I'm not a big believer in sort
35:07of the traditional open-source models
35:09and I was very worried I had just doing
35:12the board of github and I was like panic
35:14that Chris was gonna be like all over my
35:16case about you know how can you be you
35:19know throwing open source under the bus
35:21and not supporting it and all that from
35:23a business standpoint and I showed him
35:25the blog and he's like oh yeah this is
35:26great this exactly we don't we don't
35:30thing open-source and I'm like mmm
35:31github is all about open source but you
35:33don't have open source like help me to
35:34understand that so you know how do you
35:36guys I mean I know how explain to the
35:38folks here how you guys monetize what
35:41you're doing and what you know how you
35:43feel like what's happening there yeah I
35:47mean I have been criticized and have
35:50been in this conversation for many many
35:52years now but I grew up as an
35:53open-source developer I have some
35:55luckily I'm like a little bit too old to
35:58be part of the generation where
36:00everything's on Facebook but I like
36:01wrote papers in college about how
36:03open-source everything is the way to go
36:05that I'm completely embarrassed about
36:06now I mean well written papers well
36:07research but I wouldn't put them in
36:09front of this group so I grew up is like
36:10I was installing Linux and when I was
36:12young I was going to install fast and
36:14Microsoft is evil and you know Gates's
36:17the Borg I came from that world of
36:19open-source but at the same time you're
36:22watching Red Hat you're watching
36:23Microsoft you're watching Apple and it's
36:25like I'm a capitalist I think capitalism
36:27fuels innovation and I think it actually
36:28feels altruism as well and so like what
36:30I want to do is make an impact in the
36:32world and I think that like history
36:34speaks for itself and really what
36:36history says to me is pragmatism use the
36:38best tool for the job and there are
36:40cases in which open-source is the best
36:41tool for the job and there are cases in
36:43and ultimately the end of the day I am
36:45NOT a crusader for the cloud and I'm not
36:47a crusader for open-source I am a
36:49crusader for you building the best
36:50software possible and you shipping the
36:52best product to your customers possible
36:53you having an awesome software
36:55development team part of that is having
36:57a software development team that is
36:58empowered that has autonomy that has
36:59agency that's able to use and make
37:01decisions about the software that
37:03they're using able to share able to
37:04communicate openly and part of that is
37:06about making money off of the service to
37:08providing and so for us I think github
37:10really is the sweet spot there is this
37:12not about open versus closed for us it's
37:14about open and close and I really
37:16believe that that's something that we
37:17embrace that has opened doors for a lot
37:19of companies where it's okay for you to
37:21sell proprietary closed source software
37:23and at the same time send a poor request
37:25to one of the biggest open source
37:26projects in the world it doesn't make
37:29it doesn't make you anything like worse
37:31or it doesn't make you a fraud in fact
37:32some of the biggest closed source
37:34companies in the world are built on top
37:36of Linux the most massive open source
37:38project in the history of the universe
37:39so for me the the whole discussion the
37:42whole philosophical discussion
37:43it's it's it's not anchored in the right
37:45sort of mind frame which is what's best
37:48for the end-user and ultimately I think
37:50developers are lazy and this is the
37:52thing like like Twilio is awesome
37:54because I'm sure there's a million
37:55Tullio competitors that are open-source
37:56but like they're difficult I don't have
37:59time for that I don't care I just want
38:01to use Tulio and digital ocean is the
38:02same thing that you yourself been in
38:04hosting for over eight years like what's
38:06different digital ocean is simple as
38:08hell to set up and really that's what I
38:09think it's about so when open source
38:11solutions can provide that that's great
38:13and when they can't that's great too and
38:15I think ultimately it's about software
38:17development and not open and close and I
38:19think four years from now there won't be
38:20a single business on the planet that
38:22only uses closed source or only uses
38:24open source I think even the companies
38:26like Red Hat even the MongoDB s even the
38:28Hortonworks of the world they're using
38:30some ERP software or some Intuit
38:32software that is itself closed source
38:33right and it's disingenuous to say and
38:35I'm not accusing anyone but there's no
38:37only open company it's it's a mix and
38:39really it's about what is the best for
38:41your business your developers and your
38:42customers right now that's what I
38:43believe and that's what we've tried to
38:45push with github and that's why the
38:46business model has been open source is
38:48totally free and closed source cost
38:50money because we've always believed that
38:52the more we can get people using open
38:53source the more we're gonna have closed
38:55source businesses making money profiting
38:57and ultimately concerning back to get up
38:58so that's sort of what we'd come from it
39:00right Chris I'll extend something that
39:03you just said you said it's about
39:04software development right and I think
39:06that's one of the key things is that
39:08developers love building it's like it's
39:11inventing it's the human spirit of
39:13building that's existed forever but once
39:17it's built they don't want to operate
39:19right that's that's just annoying that's
39:21busy work because what happens you
39:22deploy it and then you know three months
39:25later you know a pager goes off and
39:28you're like oh I got to deal with this
39:29now that's annoying and so what a lot of
39:31company is in the as a service space
39:33you're doing is essentially we're moving
39:34the non fulfilling the thing you usually
39:37get yelled at by your manager for or
39:40your customers by and the thing that's
39:42annoying and getting rid of that so that
39:44all you do is have that joy of building
39:46and once you've built it it's someone
39:48else's job to operate it know and it's
39:50that separation of concerns that I think
39:52has really come to a maturation in the
39:57ten years because you can do it now
40:00because of multi-tenancy not because a
40:03lot of the things that have advanced
40:04that allow us to say yeah we can you
40:06know ship that code to somewhere else
40:08and let it run in a way that many
40:10customers can run on the same
40:12infrastructure without affecting each
40:13other and that's like you know joy for
40:16the developer right I don't worry about
40:17it anymore yeah remember my last company
40:19we were installing we needed you know
40:21trouble ticketing you know for our
40:23customers so we're installing an
40:24open-source mantis if anyone remembers
40:26that right and like we needed a blog so
40:29I'm installing WordPress and we needed a
40:31CMS we're installing Drupal like and we
40:34have this collection I had one server
40:35that had like 40 open-source PHP
40:38applications on it and they were all in
40:40various stages of you know too many
40:42versions old and getting exploited
40:44because they were insecure or not
40:46scaling right and like starting Tullio
40:48it's like we need ticketing we use
40:50Zendesk right we need you know CMS or we
40:54need WordPress we use WP engine like you
40:56need and everything is not as a service
40:58so that we don't have to operate it
40:59great there's still a big obsession I
41:01think with code and I don't know why I
41:04mean we're a part of it but writing code
41:07running code maintaining code debugging
41:09code scaling code that's the hard right
41:10people ask me I'm in this fortune
41:12position I've been a developer my whole
41:13life and I no longer write a lot of code
41:15people say do you still write code and I
41:17do like I it's fun for me like a hobby
41:19is like some people play guitar for
41:21money so you'll play guitar in the
41:22weekend and they're different but my
41:24idea of writing code now is like a
41:26little sip it that no one will ever use
41:28but me or a plugin for my editor that no
41:30one will ever use but me so my answer is
41:32no because now I think of writing code
41:34as debugging it getting code review
41:36running it watching the exceptions and
41:38so so much of the software development
41:40lifecycle is not writing the code but in
41:42maintaining the code and making sure
41:44it's bug free and running the
41:45maintenance on it I think that's really
41:46the next thing um I think that github is
41:49actually gonna get less interesting in
41:50the future and it's gonna be less about
41:52I can grab an open-source project and
41:54run it and more about where do I run it
41:56what are the systems it integrates into
41:57and so gate up is really awesome I think
41:59for many reasons but it's going to
42:02become more of a commodity more of a
42:04sort of table stakes that oh yeah
42:05there's this place where I can get all
42:07this code that's not I'm not trying to
42:09hire for that I'm not trying to
42:10someone that can download something from
42:12github or write some software that
42:13integrates some truly oh I'm trying to
42:15hire someone that can sort of manage
42:17this holistic process and run the
42:19service and see when things are wrong
42:20and in fact debug them when things go to
42:22crap at 2:00 a.m. that's the real
42:23business value you get out of it and I
42:25think in the future that's going to be
42:26the thing that schools start teaching
42:28and the things that we start hiring for
42:29more and more is not the aptitude test
42:31around what java functions do you know
42:33and do you end up between static and
42:34dynamic but can you debug a piece of
42:36running software how do you scale a
42:37piece of running software and how do you
42:39work with someone else across the planet
42:41when you're both working on a piece of
42:42software that's going wrong I think
42:43that's the real thing that we still have
42:44figured out yet yeah great Chris Jeff
42:47Ben thanks a lot and thank you all