Cloudflare TV

Bringing the Next Billion People Online

Originally aired on 

Best of: Internet Summit 2016

Session 1

Bringing the Next Billion People Online

  • Nathan Eagle - Co-Founder & CEO, Jana
  • Evan Marwell - Founder & CEO, EducationSuperHighway
  • Alex Dyner - Head of Special Projects, Cloudflare

Session 2

Hardware Companies that are Really Software Companies

  • Jayshree Ullal - President & CEO, Arista Networks
  • Michelle Zatlyn - Co-founder and COO, Cloudflare
Internet Summit

Transcript (Beta)

Music I think this is the post-lunch sort of low, potentially.

So we'll get started. We're going to pick up on a theme that we were talking about this morning, connecting more people around the world to the Internet with our next two guests, who have connected tens of millions of people around the world, both in the United States and in the emerging markets.

Evan Marwell is the founder and CEO of Education Superhighway.

The organization's mission is to upgrade Internet access in every public school classroom in America so that every student can take advantage of the promise of digital learning.

Its first three years, the organization helped shape President Obama's Connected initiative and served as a catalyst for modernization of the FCC's $4 billion e -rate program, which helps schools and libraries obtain affordable broadband.

These achievements earned Evan the 2015 Visionary of the Year Award from the San Francisco Chronicle.

Welcome. Thank you. Nathan Eagle is the founder and CEO of JANA, which is currently the largest provider of free Internet in emerging markets.

JANA's technology has been integrated into the billing systems of over 300 mobile operators that have a coverage area of over 4.5 billion people.

In 2015, JANA provided over 30 million people with free ad-supported Internet access, in addition to running JANA, Nathan teaches at Harvard and is a technology pioneer at the World Economic Forum.

So you guys are both solving big problems. Evan, you're improving connectivity and the Internet for 50 million plus K-12 students in the United States.

Evan, you're connecting even more people globally. We talked a little bit this morning in a different session about the private sector and the public sector and governments and how they work together.

Even more broadly, I mean, talk about the problem that you saw four years ago when you started the organization and how you've worked with the government to try to solve that problem.

Yeah, great. So four years ago when we got started, there were less than 5 million out of the 50 million kids in America's public schools that actually had enough Internet access in their classrooms to use technology.

And we saw the revolution that was starting to happen in classrooms where teachers had access to enough bandwidth.

And so we set out on this mission to connect the next 45 million of those students.

But what we realized very quickly was just like any of the great infrastructure projects in America's history, whether it was the Erie Canal or the Transcontinental Railroad or rural electrification, this was not something that the private sector alone was going to accomplish.

In fact, we've had Internet for a really long time and only 5 million of our kids in 2012 had enough bandwidth in their classroom.

So we realized we had to leverage the government and the public sector to actually achieve this goal.

And so all of our work has been about really bringing together the public sector and the private sector, the public sector both on the demand side, the schools, they had a real lack of expertise, a lack of knowledge about how to go through these procurements, how much broadband should cost, who they could get it from aside from their phone company or maybe their cable company that they'd been buying forever.

And it was really about connecting them and their needs with the private sector, but with the government bringing the funding to make that happen, especially to make the buildout of fiber networks to places that don't have it happen.

Because today, you know, telecom companies, cable companies, they've built fiber networks to probably 70% of the population, if you will.

But that last 30%, it's really hard for them to make an economic case to build out those networks and make those investments.

And that's where I think the public sector has a really great role to play.

On top of that, the public sector can be really helpful in getting to scale and building awareness, right?

So the way that we've managed to connect 20 million kids over the last two years has been largely by partnering with state governments and taking resources from the federal level, but the relationships that those state governments have with the schools and with the communities to make this all happen.

That's great. And so Nathan, you're also dealing with governments, you're dealing less with fiber, more with mobile operators.

Talk a little bit how you're going about connecting more people in the emerging markets.

Sure. I think there's an interesting analogy between our two models.

I mean, we have to work with service providers as well. I mean, our service providers generally aren't just kind of the ISPs of the world, but rather they're the mobile operators.

And there is inherent costs associated with providing connectivity, whether you're providing that connectivity in a school or in a village in Africa.

And you can't expect these service providers to basically just be rolling over and providing that connectivity without thought to their bottom line.

And so someone needs to be footing the bill here. And for us, it's not about the governments.

The governments are a potential source. But what we've found is that advertisers are a potentially massive group of organizations who are increasingly interested in trying to get mindshare of these emerging market consumers.

And so maybe backing up and talking a little bit about our business, we provide free Internet to people exclusively in emerging markets.

We are now become the second largest mobile advertising platform in India next to Google.

And we've gotten that way on the backs of advertising. So we provide people with free Internet and in exchange, they are exposed to ads.

And that model isn't particularly new, right?

I mean, just as the PNGs of the world created the soap opera to make broadcast television free, we're enabling these big global brands to essentially make broadband mobile Internet free in some of the regions where people simply can't afford the price per megabyte to be able to consume it any other way.

So an obvious question for you and sort of a juxtaposition is and what Facebook is trying to do, I think, in a different way in terms of how they enter into some of these countries and interact with the operators.

Can you talk about that a little bit?

Absolutely. I mean, so, they are our largest competitor.

But the model is fundamentally different. What we do is we essentially put money into someone's prepaid account, money that can be used to get access to any kind of content.

They can get access to Facebook, but they also can get access to Google and watch a YouTube video.

And we get that money through advertising. The Facebook model is one where they partner generally with one carrier per country and convince that carrier that they should give away Facebook for free.

And both models work.

I mean, we've been working in India before their project kind of got regulated away.

We were working hand in hand, and our users found benefit in both things.

I mean, they would go through Facebook to, or Freebasics to get access to Facebook.

But when they wanted to get access to the rest of the Internet, they would go through our service.

I think it's telling that we had about five times as many users as they peaked out at, despite basically launching almost a year after they did in India.

And I think that's indicative of the fact that the population at large don't just want Facebook.

The population at large, whether you're in North America or whether you're working in rural Tanzania, they want access to the full Internet.

And with our model, we can provide them that. So when we were talking a little bit earlier in the green room, we were talking about your two businesses, and one thing that we kind of brought up was that you both rely, and this may be an obvious point, on the carriers.

You're using free market principles. You're not appealing only to altruism, or mostly not appealing to altruism.

But you have to rely on these guys.

It'd be interesting to talk a little bit about how you interact with them, and the kind of similarities and differences.

Evan, I'll start with you.

Yeah, so for us, our big message to the carriers is that we are a lead generation channel for them, right?

So there's a lot of business out there with schools that they're not doing today.

I mean, schools are connected, but they don't have nearly enough broadband, and they don't all have the fiber optic connections that they have.

So depending on the carrier, if it's a traditional incumbent LEC or cable company, often the discussion is, hey, today you're selling them 100 megabits, they need a gigabit, we're going to help convince them that they need that gigabit, and then we need you to step up and, you know, give them a good deal to get to that gigabit.

And it turns out sort of the more broadband for your budget message is one that works with both the schools, because they don't have huge amounts more money they can spend, but it also works with the carriers because they understand their economics of providing additional, on a wired network at least, providing additional bandwidth.

It's pretty easy for them to do that. So for us, we're first and foremost a lead gen channel for them.

But in addition, you know, I have to say, as I have traveled across America dealing with the biggest carriers and the smallest ones, they want to do the right thing by kids.

And so if you can set them up in a way where they can be successful financially and they can also do the right thing for kids, it's a winning proposition.

It's a win-win. That's right, Nathan, do you have the same experience?

Regrettably, our mobile operators are less inclined to at least take any kind of sacrifice on behalf of kids.

But I mean, and I think actually this is kind of where perhaps our businesses fundamentally differ.

We have to be carrier agnostic. We can't actually go out and just do a tie in with one operator in a country because at the end of the day, you know, P&G or Unilever or Coca-Cola or Twitter or Amazon, you know, they don't want to engage with just Tata DoCoMo subs.

They want to engage with every consumer in the country.

And so what that means is that we have to partner and integrate our technology into the back end billing systems of every single mobile carrier.

And in doing so, what that means is that we can't play favorites and we have to be kind of at arm's length with the mobile operators.

And so the way we've done that is simply by going out and buying in bulk Internet at whatever the price that they're willing to sell it to us.

And we do that across every mobile operator.

And to date, you know, our tech has been integrated now into the back end billing systems.

We just integrated with our 311th mobile carrier. And that brings us our total up to 4.56 billion prepaid subs that we can essentially put money into their accounts, which represents more than half of humanity.

Yeah. And just to be clear, I mean, we're helping the carriers do the right thing.

But the reason they're doing the right thing is because we're creating competition.

Right. So a big part of what we're doing is we're saying we're a lead gen channel who wants the business, but they're bidding on it and creating competition is what's motivating them to do the right thing.

Yeah. But you both have had incredible success in your organizations.

Evan, you've created an enormous amount of awareness for this.

You've been able to garner federal support, billions of dollars to help sort of stay at the state and local level as public schools upgrade.

What are some of the biggest challenges you have?

I know you have a mission by I think it's by 2020 to have every K through 12 public school upgraded.

What are what's your biggest challenge as you as you look ahead to the next two and a half?

Yes. As we look ahead, our biggest challenge, frankly, is marketing.

Right. So we have got these great programs.

We've got 40 governors across the country that have committed to upgrading their schools.

We have four billion dollars a year of funding from the federal government to make those upgrades happen.

And our challenge is it turns out it's hard to sell free to schools.

Right. So our biggest challenge is how do we get those schools to engage with us, to engage with the programs that we're running with the governors so that we can hit our goal of being out of business in 2020?

And that's really been our plan from the start. By 2020, upgrade every school to the minimum standards, 100 kilobits per student, fiber to every school, Wi-Fi in every classroom and then go out of business.

And that has been actually one of the great lessons, unexpected lessons from the work we've done, which is, you know, by having an end target, we have been able to run our organization very differently than most nonprofits.

In fact, we run our organization pretty much the same way that any startup runs, you know, has an end target of an IPO or a sale or whatever it is.

And it allows us to be milestone driven.

It allows us to raise our philanthropy in rounds. It allows us to attract the same kind of people who want to work at, you know, the next startup here in Silicon Valley because they know they'll accomplish something in that period of time.

So in a way, our biggest challenge is marketing. But our other biggest challenge is the same set of challenges that any high growth organization has in Silicon Valley.

Thinking about the same question for you, Nathan, thinking about connecting four and a half billion people and getting them to interact with the even the mobile Web.


That sounds challenging. I mean, I think we kind of use a similar principle about having a deadline or perhaps at least a stretch goal of by 2020 getting to a billion people.

And we're about a 20th of the way there now. So we're making slow and steady progress.

But I think it's a tractable goal. And in terms of like the biggest challenge that I think prohibits us potentially from hitting getting giving a billion people free free Internet access, it comes down to advertiser demand and it comes down essentially to where is the money going to come from?

Right. To give a billion people free free connectivity.

You know, you're talking about tens of billions of dollars that you need to that you need to come up with.

And what that ultimately means is that we've got to figure out how do you how do you take what's being spent right now in the developing world on advertising, which is about 200 billion dollars.

And how do we get about 10 percent, 20 percent of that capital and redirect it away from the people who own the billboards or the radio stations or the television channels and and and put redirect that capital directly into the pockets of the very consumers that these large advertisers are trying to reach.

And that's a challenging value proposition. But that's that's the that is the big challenge.

We have a marketplace, but on one side we have people who want free Internet and the other side we have advertisers who want to start engaging with this audience.

We're never, you know, in the foreseeable future ever going to run out of people who want free Internet.

So we don't have to advertise our products.

We you know, we have basically it seems like an unlimited supply of people who are constantly downloading our app, wanting to basically engage more with advertisements so that they can get free connectivity.

And our challenge is like we've we've got to figure out how to source the demand that can start enabling these next billion people to to get online.

One other question, then we'll open it up to the audience.

I think in your case, it's sort of obvious someone who's not connected becomes connected.

Now, all of a sudden they can text message or see sort of basic content and communicate on a phone.

Evan, I think maybe maybe in your case, at least for me, it's sort of less obvious.

I mean, what's what's an example of something that a kid in school today can't do that he or she will be able to do as a result of the kind of upgrade that you're you're working on?

So, I mean, there's an incredible number of things that are happening in classrooms because of technology.

But the one that I always go to, there are two that I probably point to first.

The first is what people talk about as the course equity issue.

Right. So there's an amazing statistic that came out not too long ago. Forty percent of America's high schools don't offer physics.

Fifty four percent don't offer calculus.

Seventy five percent don't offer computer science. That one was probably a little more expected, but 40 percent and the numbers for chemistry are high and so on and so on.

So the first thing that schools are using this for is to give kids access to those courses.

And that's how we're leveling the playing field.

Right. Technology in many ways and the Internet in education is about leveling the playing field, making sure every kid has access to the same kinds of educational opportunities.

So so that's one. The second one, which which I think is ultimately where we'll head, which is personalization of learning.

But the starting point for that is empowering teachers and it's giving teachers the data that they need to know who is getting it and who's not.

Where do I need to focus my attention?

And my favorite example of this was in an eighth grade math class not too long ago.

Every kid in the class had an iPad and they came in and the teacher said, OK, your go problem is on your is on your iPad.

Do it and submit it to me over the wireless network.

And after about a minute and a half, the teacher gets up from the front of the classroom and he starts walking up to some of the kids.

And I'm sitting there thinking, well, what's he doing?

And I realized that what's going on is he's walking up to the kids who haven't submitted the problem yet.

So he knows instantly which kids understood last night's homework and yesterday's lessons and which kids didn't.

And more than that, because he's having a chance to talk to them, he's understanding what they didn't understand.

So now he in real time is adapting what he's going to teach that day.

And he starts with a review of the concepts that kids are not getting.

And then he can move on to the next set of material, which if the kid didn't understand yesterday's material, they're certainly not going to understand today's material.

So those are two examples. But the thing that is so exciting about this is it's it's there's a thousand flowers or ten thousand flowers or maybe one hundred thousand flowers blooming out there where teachers are figuring out how they can take technology to make their classrooms more effective.

That's great. I know we're running a little bit short on time, but we have time for for one or two questions from the audience.

Grab this and if it grabs.


So this question is for Evan in regards to providing broadband connectivity to schools.

Are there any initiatives that you're doing to also provide hardware, which is also a major cost for them?

Yeah. Thanks. Yeah, so there have been some initiatives of companies like Sprint, for instance, has donated a lot of hardware to some schools.

But the reality is, I believe that the market is going to actually take care of this problem.

So the market is driving down the cost of devices.

We're sub two hundred dollars on Chromebooks and things like that already.

And the real big opportunity is that America spends eight billion dollars a year on textbooks.

OK, and so by transitioning that eight billion dollar a year budget to things like hardware applications, content and teacher professional development effectively, I think there's plenty of funding to actually address those kinds of needs.


Hi. So I have a couple of questions. One is based on your work internationally, both of you, and how you are addressing issues of culture and how you've had success in ensuring and how you're thinking about gender issues with respect to culture internationally.

And then the other one is domestically, how you are working within indigenous communities and on reservations and what you're doing to make sure that they're not forgotten in terms of being connected.

You want to split those one for one?

We do no business in North America with, you know, with anyone.

And we do no business in Western Europe either. So we're exclusively in emerging markets.

And in terms of localization, I mean, it's it's challenging.

And we have operations in over 90 countries. And we operate this company out of Boston.

And it's difficult to have bring in in-house experts in Filipino culture and in Nigerian culture and people who are experts.

And like, how do you engage with the South African consumer?

What's striking is across cultures, across continents, there does seem to be kind of one underlying feature that that seems to unify at least all of these these groups of our users globally.

And that's the desire to get free Internet access.

So so that's kind of that's our first, you know, we throw lots of stuff against the wall.

Some things stick, some things don't.

But there is kind of this, you know, undeniable demand for for free Internet access.

And so that's kind of what we're trying to do. And we're trying to do it in a way that resonates for the product that we have.

And then it's just a question of like, well, how do we localize it in a way that really resonates?

What we found works in a market like Pakistan categorically fails across the board when you try to launch that type of feature in a place like Manila.

So we do have to do more localization than I thought we would have to.

And and ultimately what in many instances that means is putting people on the ground and getting exposure and doing these kind of face to face user interviews.

And and what's what I find gratifying is that to your gender question, it is striking how, you know, getting getting access to connectivity without having to incur cost really does resonate with basically underserved demographics in some instances, including including women where they don't need to take money to be able to communicate with their sister living up country or or whatever else that they would like to get the access to the Internet for.

And what's particularly gratifying from our business model is that and for the case of women, those are exactly the types of, you know, the demographic that are some of the largest advertisers in the world are trying to reach.

You know, whether we're working with the Hindustan Unilever or PNG, it's it's the women that are in many instances making these decisions about what kind of soap to buy for their household.

And so we're enabling our advertisers to go out and target the demographic that they want.

And at the same time, I think ultimately adding a lot of value on the user side.

Yeah, in terms of your question on tribal communities and things like that, so they're 100 percent part of our target in terms of getting them upgraded.

And we're working with the Bureau of Indian Education as well as tribal leaders across the country to make sure that they're included in these opportunities as well.

But in some ways, similar problem that I was talking before, marketing, right?

How do we make sure that they respond when given these opportunities to actually do something?

Well, I think we're we're out of time.

Thank you both for being here and for for a great discussion.

Thanks. Hi, we're Cloudflare.

We're building one of the world's largest global cloud networks to help make the Internet faster, more secure and more reliable.

Meet our customer BookMyShow.

They've become India's largest ticketing platform thanks to its commitment to the customer experience and technological innovation.

We are primarily a ticketing company.

The numbers are really big. We have more than 60 million customers who are registered with us.

We're on five billion screen views every month, 200 million tickets over the year.

We think about what is the best for the customer.

If we do not handle customers experience well, then they are not going to come back again.

And, you know, BookMyShow is all about providing that experience.

As BookMyShow grew, so did the security threats it faced. That's when it turned to Cloudflare.

From security point of view, we use more or less all the products and features that Cloudflare has.

Cloudflare today plays the first level of defense for us.

One of the most interesting and aha moments was when we actually got a DDoS and we were seeing traffic bursts up to 50 gigabits per second, 50 GB per second.

Usually we would go into panic mode and get downtime, but then all we got was an alert and then we just checked it out and then we didn't have to do anything.

We just sat there, looked at the traffic peak and then being controlled.

It just took less than a minute for Cloudflare to kind of start blocking that traffic.

Without Cloudflare, we wouldn't have been able to easily manage this because even our data center level, that's the kind of pipe, you know, is not easily available.

We started for Cloudflare for security and I think that was the aha moment.

We actually get more sleep now because a lot of a lot of the operational overhead is reduced.

With the attack safely mitigated, BookMyShow found more ways to harness Cloudflare for better security, performance and operational efficiency.

Once we came on board on the platform, we started seeing the advantage of the other functionalities and features.

It was really, really easy to implement HTTP2 when we decided to move towards that.

Cloudflare Workers, which is the, you know, computing at the edge, we can move that business logic that we have written custom for our applications at the Cloudflare edge level.

One of the most interesting things we liked about Cloudflare was everything can be done by the API, which makes almost zero manual work.

That helps my team a lot because they don't really have to worry about what they're running because they can see, they can run the test and then they know they're not going to break anything.

Our teams have been able to manage Cloudflare on their own for more or less anything and everything.

Cloudflare also empowers BookMyShow to manage its traffic across a complex, highly performant global infrastructure.

We are running on not only hybrid, we are running on hybrid and multi -cloud strategy.

Cloudflare is the entry point for our customers, whether it is a cloud in the backend or it is our own data center in the backend, Cloudflare is always the first point of contact.

We do load balancing as well as we have multiple data centers running.

Data center selection happens on Cloudflare.

It also gives us fine-grained control on how much traffic we can push to each data center depending upon what is happening in that data center and what is the capacity of the data center.

We believe that our applications and our data centers should be closest to the customers.

Cloudflare just provides us the right tools to do that.

With Cloudflare, BookMyShow has been able to improve its security, performance, reliability and operational efficiency.

With customers like BookMyShow and over 20 million other domains that trust Cloudflare with their security and performance, we're making the Internet fast, secure and reliable for everyone.

Cloudflare, helping build a better Internet.

Bots have become increasingly challenging for us.

We've had everything from attempts at brute force account takeovers to competitors scraping user information to mass spamming.

Many of our customers, whether they're small individuals or large enterprise organizations, are struggling with large amounts of bot-generated traffic that's impacting their site and their business.

One of my customers came to me and said, I have no idea what's happening on my login page.

And we turned on bot management and we saw 270 million credential staffing attacks per day.

That was crazy.

We didn't expect that. Our station enables artists and designers to showcase their work in a slick way so that they can find work and now they can even make money on the platform by selling their digital goods and their prints on our marketplace.

As a platform, we have to be seen as a safe and high quality service for our users who are depending on it for their livelihoods.

The way that they were hitting us was a whole new level of sophistication because we already did have things like rate limiting in place.

We knew that we had to go to the next level.

We had to bring in some next-gen tools in order to stop the bots.

Our mission at Cloudflare is simple but bold, and that is our mission is to help build a better Internet.

With bots, we're focusing on ways that we can leverage and build an intelligent and powerful security solution, but make it simple and easy to use for anyone on our service.

As soon as we turned on Cloudflare bot mitigation, we could see that there were hundreds of thousands of requests coming from a variety of IP addresses and countries.

There was even requests and bot activity that we weren't even aware of.

Once we turned it on, Cloudflare stopped the bots dead in their tracks.

Here is how Cloudflare bot management works.

When a website gets attacked, that website is not alone.

In fact, it's part of a global network of 13 million Internet properties. What we do is we build a machine learning algorithm that learns from that data and it computes a likelihood, a score, is this request from a human or a bot.

So when a human comes to visit your website, we analyze that score and let the human request go through.

But if a bot comes to your website, we stop that. That is the power of machine learning.

Every new attack we see makes our protection smarter.

It's all about the inputs. We service over 13 million Internet properties in over 150 countries.

And so we get incredible breadth and depth of data.

The other thing that differentiates Cloudflare's bot management solution is its ease of use.

With simply one click of a button, you're good to go. Of course, we have all of the robust capabilities to more fine grained control and manage the solution.

But it's really important to us that if a customer is struggling, they can come to Cloudflare and with a matter of minutes start securing their site from the bots that they don't want.

Cloudflare bot mitigation has given us peace of mind. They've got our backs.

We have seen malicious foreign actors attempt to subvert democracy.

What we saw was a sophisticated attack on our electoral system.

The Athenian project is our little contribution as a company to say, how can we help ensure that the political process has integrity, that people can trust it and that people can rely on it?

It's like a small family or community here.

And I think elections around the nation is the same way.

We're not a big agency. We don't have thousands of employees.

We have tens of employees that we have less than 100 here in North Carolina.

So what's on my mind when I get up and go to work every morning is what's next?

What did we not think of and what are the bad actors thinking of? The Athenian project, we use that to protect our voter information center site and allow it to be securely accessed by the citizens of Rhode Island.

It's extremely important to protect that and to be able to keep it available.

There are many bad actors out there that are trying to bring that down and others trying to penetrate our perimeter defenses from the Internet to access our voter registration and or tabulation data.

So it's very important to have a elections website that is safe, secure and foremost accurate.

The Athenian project for anyone who is trying to run election anywhere in the United States is provided by us for free.

We think of it as a community service.

I stay optimistic by reminding myself there's a light at the end of the tunnel.

It's not a train. Having this protection gives us some peace of mind that we know if if for some reason we were to come under attack, we wouldn't have to scramble or worry about trying to keep our site up that Cloudflare has it back.

It's very hard to follow a hacker session with bottles of rum and you know, stick to water and then.

But I feel like we're up for the task. I have the honor to spend the next 30 minutes with Jayshree Ull.

She is the CEO of Arista and Jayshree is just an amazing, amazing person.

She was part of Cisco's first acquisition while she was at Crescendo back in 1993 and ended up staying at Cisco for 15 years and built a 10 billion dollar business line for Cisco.

That's a billion in revenues. And so that's pretty incredible.

And then after 15 years of doing that, she left and in 2008 joined a 30 person startup called Arista.

And now eight years later, you're a public company.

You are on the well on the path to a billion in revenue on your own revenues.

And you've built a team to twelve hundred people. And so please join me in welcoming Jayshree.

Thank you. It's great to be here. We're definitely going to get to the technology and what this means for the future of the Internet.

But I just you you've had such an incredible career as an operator.

I want to spend a little bit of time talking about your career because there's a lot of people here who are working at tech companies who are entrepreneurs, who are investors trying to invest in the next Arista.

And so tell us what it was like when you were at Crescendo having a lot of fun and Cisco, this big company came and said, hey, we want to acquire you.

Yeah, no. So first of all, I think every entrepreneur has to pursue their dream.

And one of the first goals in any company is to when you're asked, what are you doing in your company and what's your exit strategy?

Remember, there is no such thing as an exit strategy.

If anybody ever tells you that, only think in terms of entrance strategies, not exits, because the minute you contemplate an exit, you really haven't built a company.

You're thinking too much about how to exit. Right. So at Crescendo, our entrance strategy, if I may use those words, was to really drive at that time what we thought was unbelievable, amazing bandwidth, which was all of 100 megabits.

Today we talk about 100 gigabits and terabits. But back in 1993, when probably most of you were toddlers, the world was still at one megabit and 10 megabit and pushing 100 was amazing.

And on top of that, we were going to push amazing 100 megabits over not optics, but category five twisted pair, which was defying every law of physics in terms of ground noise.

And the capability to run that over 100 meters was just simply unheard of.

We went on to do that. But what we quickly also did at that time was pioneer a new class of switching.

And we went to Cisco and said, look, we're doing all these great things.

You guys are building great routers.

We're building great switches. We think we should partner.

And they said, yeah, we like that. And Cisco, you have to realize, was a very young company.

They were smaller than even Arista is today. And what ended up being something we wanted to do from a partnering point of view ended up being an acquisition.

One none of us wanted. So we my CEO and myself, we all wept the day we got acquired because there's a great feeling of loss, loss of dependence, loss of strategy.

But perhaps one of the greatest things Cisco did was it was their very first acquisition.

And they committed me to stay two years. I ended up staying 15, which was easily eight times longer than I had planned on, but always did so in two years at a time increments.

I never thought I'd be there 15. And probably the greatest thing Cisco ever did was leave us alone, leave us autonomous and allow us to continue to cherish our entrepreneurial goals.

And and Crescendo went on to become a third of Cisco's revenue and over 10 billion.

That probably is one of the most successful tech acquisitions.

I think so. And a lot of it had to do with the fact that it was great technology.

But more importantly, the team was great. Every one of us stayed and worked together and and basically shaped the direction of switching for Cisco.

So those 15 years that you're at Cisco, there are probably a lot of high points and low points.

Definitely. When you look back on your career there, what were some of the what are some of the aha moments or defining moments where you're like, these are the two that really stand out?

Well, as you said, there were highs and lows.

I'll talk about both because nobody should ever think that 10 billion was perfect and it was filled with nothing but goodness.


Probably one of the greatest high points that actually started out as a low point was when we went to see a major financial and we were trying to show them the architecture as the world was moving to the World Wide Web.

This is how far back we are talking.

And everybody's financial environments were mainframe applications and client server.

And we said, look, the world of shared hubs is gone.

Need to migrate to switching. Here's the software architecture we put together. And it must have been by far one of the worst customer interactions I've had.

How many of you here from New York?

Then you'd understand the person was frank. They told me how terrible my product was in stronger language and how we didn't know what we were doing.

And it was just one hour of constant berating on everything we've done wrong.

And I walked out of there saying, we're never going to make it and we're never going to win this customer over.

And this must be one of my worst interactions. You know, I kept my cool through it all.

I mean, the customer is always right. But I can tell you the customer was always wrong on this one.

And I kept and I kept giving counterpoints and no, no, no.

You need to look at it from this way when you don't share.

And he said, no, if you if you share, if you have switching, then you lose visibility.

And I said, yes, you do. But you can add visibility tools. This went on for a while.

And at the end of it, my sales guy said to me, that was one of the best meetings.

I said, you must be kidding. That was one of the worst meetings I've been in.

He said, you don't understand. He was not apathetic. He argued every point.

And because he argued every point, it's going to go well. It went on to become our largest customer the next year.

I could have never believed it. But through that, you learn that even in an adverse situation, what most human beings want is the facts and the interaction.

They don't want to be sold to.

They want to be discussed with. And, you know, I what turned out to be one of one of one of the biggest low points became one of the high points, Cisco.

And and at that time, the switching division went on to become a huge presence in the financial markets.

And we were and the biggest one was this one customer. In terms of low points, there were there were many, too, because Cisco did many acquisitions.

Some of them were good and some of them are not so good. And as you and I were discussing earlier, whether you adopt a child or you organically have a child, you still have to raise it and for a long time raising it.

Yeah, you can't give it back to the hospital.

And it's the same way with companies. When you acquire them, you kind of go, what did I get?

It looked good for a day, but the next day when you actually have to integrate it into your culture and redefine the technology or whatever.

So we had many moments like that where we had to change directions.

We had to change the culture. And at the same time, we weren't getting at Cisco the respect of the company because we were at zero and the company was at a billion and they go, oh, you don't know what you're doing.

And we said, no, we do.

We have to build it this way and raise it this way. So one of my low points, and I'm still if you ever talk to John or anybody in Cisco, they might have many good things to say about me.

There's one bad thing they'll all say, which is I'm a lousy forecaster.

And if any of you can teach me how to do that better, I'd love to learn because there's no science behind that.

I think it's mostly guesswork.

So I said, I think we will do 200 million. So they said, all right. And they went off and built manufacturing capacity.

And you can relate to this since we have some issues ourselves in that department.

Now, you know why the so then they built something for 200 million and I ended up doing 400 million that year.

And in fact, it was three sixty five million, precisely one million a day.

So they said, you know, so I was getting yelled at for basically doubling our numbers because they hadn't built the manufacturing capacity when nobody was going to at that point look at the bright side of it, which is we had contributed almost two X more than we had.

We had promised to. This became a pattern. So the next year I said, I'm going to get really smart and I am signing up for 600 billion.

Guess what we did?

Seven fifty eight a billion. Oh, wow. So so it got to a point where anything I said, they all started adding buffers and it's just as well because that continues to be my hallmark.

And that was a low point because, well, you know, it's it's always right if you can match supply and demand.

And if you miss if you miss that vector, you can lose a lot of opportunities.

OK, so you're at Cisco. You built this 10 billion dollar business unit.

What I mean, a 10 billion dollar business line is a lot.

There aren't that many companies doing 10 billion dollars in revenue.

What was it like? What was the first five billion and the last five billion like?

Yeah, no, that's a really good question. The first five billion, I felt I was running the business.

The second five billion, I felt it was running me.

And and here's what the difference really is. In the first five billion, you're in a new market.

You're addressing trends. You're building technology. You're disrupting things.

You're dealing directly with the customers. And all of a sudden it gets so big that you have layers and layers of decision makers, aka management.

And, you know, management is too important to ignore.

But it's also important to shape correctly.

And nobody should in this room think that management is just overhead.

I don't believe it should be in a small, medium or large company. And we in Arista very much believe in what we call the coach player model, where we don't believe in just professional managers managing.

Why is that? Because I think unless you understand the pain points and actually do it yourself, it's very hard to lead a set of people.

And also people have to be led, not just managed.

You don't need to. It's not like you're in kindergarten and you need to tell them exactly what to do.

Good engineers, good, good account managers, good salespeople are naturally self -empowered and self-motivated.

So in the first five billion, there was a lot of that.

And the second, there was a lot of process and things would just happen.

And many of the times it's natural. This is the sign of evolution of a company.

And many of the times you're putting out a fire after it's happened rather than actually deciding the direction.

So you're responding to things more than you're shaping things.

It's probably one way to look at it.

Got it. OK, so back in 2008, you joined Arista. At the time, 30 people, now 1,200 people approaching a billion dollars in revenue.

And you've been selling a lot of hardware boxes the last 20 years.

You know, what has changed the last eight years since you started Arista?

What what what what has changed in this industry?

No, I think, first of all, there is no doubt that just the nature of software and networking stacks and how they need to be built, where you are fundamentally dealing with a much more monolithic stack with old principles of computer science has moved very rapidly in the last decade.

Hardware tends to move every 18 months.

You have new chips, you have, you know, new nanometer designs. You have, you know, poor density increases, power decreases.

Software in our industry hasn't even changed every decade or two.

So clearly the software needed to move at a faster pace to keep up with the fact that the last 30 years it had been stagnant.

That's first. But even hardware has changed. Back in my days when I was running Cisco, you could not rely on merchant silicon.

Merchant silicon is basically getting independent switch chips just like you get CPU processors for servers from separate companies like Broadcom or Cavium or Marvell or you name it.

That phenomena didn't exist. And because it didn't exist, you had to build all the chips yourself.

And you never worry about building the first chips. You have to worry about building three generations of chips.

Three generations of chips is a lot of silicon, a lot of fab manufacturers you have to deal with.

And each chip can take over 150 million dollars.

So you have to be able to invest a billion dollars to say, I want to be in ASICs or application specific integrated circuits.

The advantage when Arista came in is there was a massive disruption going on in hardware where I could take advantage of merchant silicon and I could shape their merchant silicon and therefore put most of my investments not on the billion dollars of hardware, but on billion dollars of software.

And that's what we did.

Arista went to literally a clean sheet of paper. And I remember talking to a number of VCs.

I hope none of you are in the room who said, are you crazy?

You know, this game's over. Cisco's won. They've built a networking stack. Why are you building another one?

And I think this comes to the third point, which is there was hardware disruption.

There was definitely software disruption where we were bringing a lot of Unix and modern state principles to our software stack.

But there was also a cloud disruption and the word cloud, which meant nothing back then, has much more depth and meaning today.

Always on, very reliable, massive scale, auto provisioning, real time, a new class of apps.

This wasn't as understood when when Arista was founded in 2004.

The other big thing is because we were not venture funded and we were funded by rich people who who belonged in the company, Andy Bechtolsheim, who founded Sun and many companies thereafter invested in Google was not just a funder.

He was a founder. He was very active. So we took the time to build this product over five years.

Unheard of typically in the entrepreneurial.

But that's what it took. If you're going to go make a disruptive play in infrastructure, you can't do this quickly.

So we built an architecture that was programmable at the control plane, at the data plane, at the management plane, you know, with very rich in APIs, what later became known as software defined networking.

We were not smart enough to come up with a marketing term like that. But I can today safely say while others were defining it, we were driving it.

So I like to think we're doing software driven networking, not software defined networking.

But this was key that that confluence of bringing disruptive hardware, disruptive software and having a disruptive customer base that we call the cloud titans.

And these cloud titans are the top eight companies that power the public world that have completely changed the way service providers now build networks.

Right. Whether it's Google or Amazon or Facebook, Yahoo, eBay, you name it.

These guys are now all over the world are shaping it. So I think the discussion is less.

Are you hardware or software? The discussion is more, how do you build a programmable software stack so that you can adapt to various types of hardware?

It's amazing.

I feel like we're so spoiled being in the Bay Area to have all of this innovators and people who are trying to push the envelope for it all the time.

And it's just very rare and inspiring.

I have a couple more questions and I'm going to move to questions to the audience.

So if you have one for Jayshree, now's the time to start thinking about it.

And so, you know, as we get more and more bandwidth, you know, all this video that's going on and, you know, it's not about it's not about megs anymore.

It's now 100 gigabits per second might not be enough. How does that impact your business?

And what and what do you see for your customers? Like if you're a large business, what do you see the impact to them?

Well, I think video, everybody always said was the killer application.

But from where I'm sitting, video is just part of how we do business today.

We all don't just look at voice and audio.

I mean, the previous was all about real time streaming. And yet if you go look at the infrastructure for video today, it's all analog.

It's all these big fat coax cables and a lot of analog running at megabits.

So there's a huge there's almost two generations video needs to skip to get to the real time workflows and streaming.

And almost every major TV station is going to IP and every major piece of content is having to adapt to that.

So Arista is very well positioned.

And I think this is one of those areas where the the video guys have recognized that it isn't, you know, whether it's Netflix or Comcast or all of the real time broadcast and audio, it's not just about how fat the pipette pipe is.

It's about how real time you can make the experience. And so you got to work both vectors.

You got to work the predictable latency vector as well as the capacity vector.

And I think that's one of the beauties of what Arista is doing. We're trying to not do the end, not just the or, you know, as you when you and your leadership team get together and plan for the next three to five years, what are some of the trends that you're seeing over the horizon that, again, the investors, the entrepreneurs here, the business owners in the room should be thinking about?

Well, I think, you know, three to five years for many of us is a lifetime away.

So you put a plan and then you iterate. And there are at least three or four trends we see that are going to shape the next three to five years.

And maybe some of you will agree with me and some of you won't.

And I'll get a little bit networking specific since that's what I know and live and breathe a lot of fear.

Please do.

And the first one is Ethernet everywhere. I don't know how many of you have witnessed, you know, sand technologies with Fiber Channel or InfiniBand with the high performance compute or token ring to go back a few years or FTDI.

But every Ethernet technology I've ever gone through finally meets its end because Ethernet is just capable of scaling, addressing the volume, addressing the automation and operation of networks so well that the more you can use of it, the better scale and also advantage of price you get.

So we are big believers of that.

And for the first time, you're seeing the rapid migration from not just one and 10 gigabit to 40 and 100.

But you're seeing Ethernet not just approach this in increments of 10 and take intermediate points.

You've seen standards for 25 and 50 gig.

And so I'm very, very bullish about that. The second thing is most networks were built as silos.

You had a storage area network called the sand. Then you had a campus network, a LAN.

You had a data center network. Then you had a branch network.

Then you had a core network. And they were called pins, places in the network.

We see that that doesn't need to be all these separate pins. And all it's doing is serving to get you all to buy lots of separate gear and interconnected with optics and make it five times more costly.

Right. So we see what we call the universal cloud network or the pin where the fundamental essence and building block is the same for all of them.

And maybe you need some different features for each of those market segments.

And we think the collapse of the pins is a very powerful convergence on the networking side and a huge advantage on operational savings and costs.

The other big thing you see is that routers and switches were considered two different markets, each of them, you know, almost taking a different path.

And again, we don't see why it has to be that way with with hardware advances.

You can you know, what's a switch? What's a router? And why can't a router be a switch?

And why can't a switch be a router? It's not a technical problem.

It's really been a legacy problem or an old habit problem. So we see switching and routing coming together.

And, you know, and really the land, the when the sand, the men are all coming together.

You know, so all of these men or women, if you will, need to converge.

And finally, the last one, which probably has more implications to you all, is if you have the right network infrastructure, we're also seeing a drastic change in the type of applications.

In the past, you had your consumer apps and you had your enterprise apps and you could sort of distinguish them between database.

And, you know, today, everything's allowed everywhere.

We use box. We use LinkedIn. We we use all of these social applications to run our business.

They're mission critical applications. I don't know what to call an enterprise app versus a cloud app.

And so the key, again, is to build a real time agile infrastructure that can address both.

And this line and that distinction that's been drawn by a mainframe or client server app versus a cloud app is less and less.

A lot of hopefully that exactly hopefully that gives you a sense of how tremendous the opportunity is for every one of us here, both at the compute storage virtualization front, as well as the infrastructure itself.

Great. We'll take one question. If someone has a burning question to ask, Jayshree, now is your opportunity.

All right, Wendy, here goes one quick, quick for the quickly.

This is all about women today. You're so inspiring. Advice for women in tech, please.

Oh, yeah, I have to get the women question, don't I? But just how about anybody in tech?

So it's a tough one to answer, because if I kept walking around thinking, I'm a woman, I'm a woman, I'm a woman, I never get my job done.


So 50 percent of us are one gender and 50 is the other. But I think what I would stress and particularly since I have two daughters and a lot of you are role models to women is make sure that they're not afraid of tech.

It's so easy in middle school and in some of the most influential grades where they think they're not good at math or they can't excel in science.

And men are women. I think there's so much going on in technology that you may not excel in one aspect of engineering.

I, for one, could not see in three dimensions.

So it's not always good at the civil engineering projects.

And, you know, just didn't my mind didn't go there. But there are so many aspects to applied science and math and that I strongly encourage that the biggest issue I see with women in high tech is there's no pipeline.

There's not enough of us.

And if there's not enough of us, that problem starts with school, middle school.

And every one of you can men or women should be a role model for that.

You know, after that, I think the opportunities are what we make of it. And really, in some of these high growth areas, there's so much opportunity that everyone's looking to hire talent.

And they're really not emphasizing whether it's men and women.

And it's particularly easy now for you all than it was back in my time where I would have to physically drop a child.

And, you know, I didn't have all the email and access to Internet tools at night to work.

I'd have to go back to work or I'd have to take meetings late or whatever.

So the world is shrunk and is a much smaller place.

And the flexibility of getting your work done and having impact no matter where you are makes it easier to be a good parent, man or woman, and get your work done.

Right. Great. Well, with that, thank you so much, Jayshree. This was such an honor.

Thank you. Thank you, guys.