Cloudflare, Inc. (NET) Earnings Call Transcript & Summary

Hello, everyone, and welcome to the Fourth Annual Wells Fargo TMT Summit. My name is Phil Winslow. I'm a software analyst here at Wells Fargo. I'm very excited to have what's been one of our favorite companies and favorite stocks, Cloudflare, Matthew Prince and Jayson Noland. Thank you for joining us today.

Thanks for having us. You're right -- I was just -- I'm here in Park City, Utah. You guys -- I thought you were all going to be here. I don't know...

It's exciting. The conference has been there. Where is everyone? Amazing. Look, it's -- or maybe soon we'll be back all together on stage. That is the goal in 2021.

We hope so.

Now obviously, Matthew, I mean we're 10 months plus now into the COVID-19 pandemic. And we believe CIOs have been forced to reevaluate their strategies, whereby no cloud has officially become a no-go going forward, or at least that's the one-liner I've been using. Can you talk about how the prominent shift towards digital and the cloud has been -- has manifested itself in Cloudflare's favor?

I think -- I keep thinking about how much worse off we'd all be had this pandemic happened 10 years ago. And if you think of the things that have happened in the last 10 years that have enabled us to be able to do this conference virtually, been able -- for many of us who are fortunate enough to work in industries that have been able to virtualize, to be able to continue to do our jobs, at Cloudflare ourselves that we've been able to continue to be productive, a lot of that has relied on a better Internet and cloud services that have allowed functions like Zoom to scale and to go on. And so I think that we're proud of the role that we've played over the last 10 years in helping bring the Internet to the place that it can survive like this. I don't think that anything has changed the general trajectory that enterprises were on, of headed towards more cloud services, headed towards more scalable services, moving away from on-premise solutions. But I do think that COVID has accelerated that. I think that Q2 was very much a quarter where everyone was just in panic mode, and we saw most of our business coming from selling more to our existing customers. Q3, we had a very strong Q3. What happened there was not only were we selling more to our existing customers, but you started to see kind of the earliest of the enterprises say, "Okay. This isn't just a temporary thing, but this is a seismic shift, and this is going to accelerate our digital transformation plans." And I think that the real wave of that still lies ahead of us, and it's something that I think is going to be one of the real lasting effects of COVID. It's that people are going to continue to just say cloud-first and really cloud-only going forward.

Yes, yes. I think that was maybe -- I understand when you're saying you had a good Q3. That's a pretty phenomenal Q3. So the -- I kind of want to start this conversation almost from the beginning and at the bottom and work up. One of the things that stood out to us really from the get-go when we started to first research Cloudflare was how different your architecture was at a basic level from the ground up from day 1. So first question here, could you share with us sort of the unique serverless architecture that you deployed? What are the benefits that have supported the rapid growth and product rollout that we've seen at Cloudflare?

First of all, we started as a security company. And the thesis was how could you take all of the functionality that you had in the firewall and deploy it as a cloud service. And in order to do that, we needed to make sure that we had infrastructure that was running that had substantial CPU resources, had substantial memory resources. And we deployed it across the world in order to service that initial use case. We made a determination early on. I remember it was a fight back in July of 2010 about whether we should have different boxes for all of the different functions that Cloudflare had or whether we would have sort of one generic server that we would buy. And then we would use software in order to figure out what ran on that. And that was a really faithful decision that we made, choosing to pick one unified architecture and deploy it everywhere in the world. Because we were a security-focused company, we really invested in making sure that these were beefy machines that had substantial CPU that could run tasks. And then we had to do the hard work to be able to spread load across those machines in a way that either we're spreading it within a rack, within a data center or even across multiple, different cities. And what that has given us is the flexibility to be able to then deploy what are today much more complicated use cases and, even in the cases of our Workers platform, allow us to run our customers' code on those same machines. And we have the ability to move it around, make sure it's optimized for either being fast or inexpensive or anything else. And it's allowed us to serve not only the largest enterprise customers, which today make up a majority of our revenue, but also been able to serve individual developers where millions of developers rely on Cloudflare and our network. And they are our greatest evangelists. They're the ones that are sitting in the Wells Fargo CTO office, pounding the table saying, "We've got to be using Cloudflare," and I think that's been a real secret to our success. But it all starts with that architecture, which gives us the flexibility to be able to service customers from the very smallest to the very largest.

Yes. I'm going to come back to that CPU-heavy comment that you made in a few minutes. But staying with the idea of how the network is different, one of the things you've talked about is sort of like flipping the script, changing what were variable costs into fixed costs. And I'd say, hey, look, the pandemic has sort of been the case study of, is Matthew telling the truth, is this real. Maybe describe that to folks that may be kind of still struggle with that concept of sort of how you do what you do is different.

So the primary components of COGS for us are bandwidth, which traditionally has been the largest component. Second is actually the colocation facilities where we would put servers and power those servers up. And then the third is the actual depreciated cost of those servers themselves. And those -- when we first started, it was that bandwidth was about twice as much as colo, which is about twice as much as the depreciated cost of hardware. And what we needed to do was make sure that as we added customers in this, we grew, that those costs wouldn't grow with us. And so that meant taking largely the bandwidth and colocation costs and converting those from either variable costs into fixed costs or, in the ideal case, converting those from costs into things we don't have to pay for at all. And the key to that, in all of those cases, is to figure out how you can not pay for bandwidth and not pay for where you locate the servers. And the answer, in one case, in some companies, the way that they have built their networks out, they go and they go to like an Equinix and they pay them for a bunch of the space for the servers. And they go to an NTT or an AT&T or a Verizon. They buy the connectivity. And the challenge of that is that as you grow, your costs just grow with you, and we wanted to do something different. And so we've worked through a number of different strategies to make sure that we can put equipment as close to users as possible. And most of the time that we're deploying equipment now, we're deploying it directly into an ISP in a particular region. And so right now, we're working on turning up a data center in Kazakhstan. Why does that matter? I mean we don't have a ton of customers in Kazakhstan. They don't pay us a lot of money in Kazakhstan. The reality is that Kazakhstan has a lot of people in it, and they're accessing our customers. And if we don't service them locally in Kazakhstan, we have to service them for wherever Kazakh telecom connects to us, which might be Frankfurt, where we might have to pay for space and power and bandwidth. If we can do a partnership with the Kazakhstani ISP, then we can actually put our equipment there. We still have to pay for the equipment, but they provide us the space, they provide us the power and they cover the cost of the bandwidth. And what's in it for them is that their customers then get a more reliable, faster experience on the Internet. And not just for our big enterprise customers but for some of those long-tail customers, there might not be a big deal, like a Kazakhstani sports team that may not pay Cloudflare anything. But the people who run a Kazakhstani telecom really, really, really care about it, and so they're willing to do the deal like that with us. And so counterintuitively, as we build our network out, our costs don't go up. They actually go down. The CapEx costs for the servers stay about the same, but the bandwidth and colocation costs drop precipitously. And that's allowed us to build a network which is much broader and much more expansive. It means that in -- we also are faster than everyone else around the world. We're able to connect locally around the world. And I think increasingly, that's going to help us as you have increasingly complicated regulatory challenges where you need infrastructure in literally every country on earth.

Yes. Now let's put that together for a moment because I've heard you talk about -- I agree that hey, maybe in the future, there might be 5 global networks that matter. It's Azure, it's AWS, it's Google and it's Akamai and Cloudflare. When I think about the -- those last 2, you and Akamai, you say, "Okay, Akamai's strategy 20-plus years ago was go after the big accounts," lots of bandwidth demands to get these relationships you've talked about. When I think about your model -- and this has to do with going back to your architecture and how Cloudflare was building, we're able to go after the long tail which, in aggregate, to your point, is a lot of traffic. How do you think of -- a, do you still agree? I'm assuming you do with that sort of that 5 global networks comments. But from a competitive moat perspective, how does it sort of keep others potentially from entering into those, call it, the big 5?

No. I think that you -- so first of all, you can be bullish on us and bullish on Akamai at the same time. And I think that both of us have the opportunity to distinguish ourselves from the other 3 by being independent. And the independent networks, I think, are going to have some real advantage over the hyperscale public clouds. But I do think that the strategy that Akamai's had, where they are in more places all around the world, is the right long-term strategy. And it's not simply for performance. It's actually for costs. And what we believe fundamentally is whoever has the lowest cost to service customers over the long term is going to win. And so we've always tried to work just relentlessly to crush those costs and to think again, as you put it, of how do I convert variable costs into fixed costs or, even in the extreme case, things that aren't costs at all. And so I think that Akamai has had that right strategy from early on. I think the way that they approached it was very different than the way that we did. We're seeing them in the market. We couldn't follow their exact, same game plan. So we went about it a different way. I think it's been very effective in building a similar network today. It's a more modern network that we have, and that's allowed us to have more flexibility, I think, over time. But Akamai is a smart company and they're working hard to do that. I think that though it is very, very difficult if, on the other hand, your entire business has been we're just going to try and chew away Akamai's CDN customers, which company after company after company, whether that was, back in the days, IDERA or Limelight or today Fastly, they're just sort of taking that as their approach on how they go forward. And the challenge of that is it becomes really difficult to figure out how do you get that relationship with Kazakhstani telecom or Pakistan telecom or Egypt -- Telecom Egypt. If you don't have that, then you're not going to be able to drive your costs down as much, and that's just going to make you always at a disadvantage when you're trying to compete with us.

Got it. Now let's go a layer up. We spent a lot of time talking about the architecture of the network itself, but let's talk about the products and solutions. I mean I'm consistently amazed. I follow you on LinkedIn. I follow the company on LinkedIn and the blog. I mean it feels like every month or so, it's a new product, new service, a new add-on, et cetera. Talk about just the product development cycles at Cloudflare and the release cycles. I mean they seem fast, high quality and potentially even accelerating. So maybe walk us through your -- just your product development strategy and cadence.

So we want to be one of the iconic technology companies. And if we're going to get to that sort of $100 billion market cap and up, when we look at the companies that have achieved that, they all have -- with a notable couple of exceptions, they all followed what is a platform strategy with multiple products that are bundled together in order to create complete solutions. And so that's the path that you're on. You can do that by going out and acquiring a bunch of companies. That's what Salesforce did, and obviously, it was very successful. That's what Adobe's done. That's the usual path. The challenge of that is then you have to integrate them in various ways, and it can be really expensive and you can make mistakes along the way. And so we really optimized Cloudflare, first and foremost, to how could we make it as fast and easy for our team to innovate themselves incredibly quickly. And I think that the real 2 secrets behind that are, first, we created a development platform for ourselves that our own team uses. And so our Workers platform, which has a lot of excitement about in the market -- but I really think that the biggest opportunity in at least the short to medium term is it's allowing us to build products incredibly quickly and get them out to market. But that was really built for our own team first. And if you look at Cloudflare for Teams, if you look at Magic Transit, you look at some of these really innovative products that we're releasing really quickly, the reason we're able to do that is we have this Workers platform that allows us to very quickly develop and create code and deploy it out there. The second piece though, which is the really differentiated piece, is that we looked at what are other companies that have very fast release cycles. And Facebook and Google are 2 companies that we admire, obviously fundamentally different businesses than we have, but they are incredibly fast at developing new products. And we say, "Why are they able to do that?" And the answer was because they had a certain set of users that didn't pay them anything, that they could use as basically guinea pigs to try new products out and quickly innovate. And so Google uses Australian users as their guinea pigs. Facebook uses New Zealand. And they are constantly rolling out new features to it. And if something breaks, so yes, who cared? It was as a free service, and so it broke." But they are able to then quickly iterate and have the best QA team in the world because their QA team are actual users that are out there. And so when we were building Cloudflare, we said, "How can we get that same thing?" If we have to rely on our largest customers to test new features, then our development cycle is going to be incredibly slow. If you make a mistake in traditional software, it's a real pain. So you have these average of 2-year -- in Silicon, it's now up to 3-year development cycles to get out there. In cloud companies like a Salesforce, they're really proud that they have 3 releases a year. That would drive us crazy because we want to be significantly faster than that. And the key is because we have all of these long-tail users, even if they don't pay us anything, whenever we're like, "Hey, we're building a new product for -- at one of our big clients. Who wants to test it?", we'll have tens of thousands of users raise their hand to volunteer to effectively be our virtual QA team. And as a result of that, our rate of innovation can be significantly faster. So you have to have an efficient development platform. That's what Workers has given us. And then you have to have real traffic from real users that are going to be willing to allow you to experiment. Of those 2 things, the one that I think is the hardest for any of our competitors to replicate is that millions of low-end customers that are volunteering to test out our new features, and that's really the secret to our ability to innovate as fast as we have.

Got it. I laughed to myself. I've always joked to people, I don't know why ANZ is everybody's sort of guinea pig region, but it's...

For us, it's always been Toronto in part because Michelle, my co-founder, is Canadian. They speak English and they're very nice. So when we -- they -- whenever we break something in Toronto there, they're always like, "Oh, it must be our fault." And I'm like, "No, no. We put some new codes." So it's been [ a new code ], nice people of Toronto.

The second time, it's like the -- let's take on this product innovation aside here because WARP, Magic Transit, Network Interconnect, Access, Gateway, the Magic firewall, browser -- okay, I'm forgetting what else, but I can keep going. And bam then boom, Cloudflare One was born, piecing all these things together in sort of a cohesive offering. That was a real sort of aha moment for a lot of investors. A couple of questions for you. First off, how has Cloudflare One or sort of just the -- I call it the positioning of all these services been received by customers? And how do you think about pricing and packaging?

Yes. So Cloudflare One was really born out of conversations with customers. We had a set of products that we think of as protecting infrastructure, so like our web application firewall, our DDoS mitigation services, our load balance. So those are all kind of in one bucket. And then we had a newer set of products that were all around protecting individuals within companies, which are our Cloudflare for Teams products. And we had organized the world where there was sort of almost a moat between those things. But when our salespeople would go out and talk to customers, they would be like, "What makes total sense for us to adopt the Teams products because we believe in zero trust networking and everything else and this helps facilitate it, but for exposing Teams to the Internet, then we have a bigger DDoS mitigation risk. We have a bigger web application firewall risk." And as a result, those pieces go together. And so while we've looked at other companies that we admire, like Zscaler, and really thought about Teams as those 2 pieces going together, when we actually talk to customers, they said that, that moat that we have created was really artificial and that the overall bundle was actually all of what Cloudflare was putting together. And so Cloudflare One started out as us actually just responding to what we were hearing from customers. And so as a result of that, I think the response has been incredibly beyond what we expected. We saw an almost doubling of inbound leads when we announced Cloudflare One because people said, "Aha, now I get it, and this is finally the network of the future that I need for my business." I think that, that was a really positive surprise for us in how quickly the market caught on. And I think it's caused us to think about -- we understand kind of what the package is but now how do we price it? And so Cloudflare One is really an idea and an architecture more than a SKU. But we're certainly thinking about, is there a way that we could bundle up all of those different Cloudflare services and sell them just basically on a per-seat basis. So you get DDoS mitigation, you get WAF, you get all of those different pieces plus the Teams pieces, and you just pay based on how many employees are within the company. And that's something that we're starting to test how that will work. But we think that, that is, first of all, really powerful to kind of our roots where we can give the best enterprise functions that sometimes we've reserved for people who spend a lot with us. We can actually bring those down market really effectively well. And then the second is that we can find a way to very naturally then grow with our customers and give them a complete solution that doesn't require them to have to stitch together a number of different providers to get that one network. So I think right now, Cloudflare One is much more of an architecture than a SKU, but watch this space because it's definitely something that we're thinking through carefully.

Got it. Now historically, kind of building on that with Cloudflare One and just your broadening product portfolio, you've talked about sort of the magic number, 4. In other words, once a customer adopts about 4 products, competition sort of largely like dissipates, churn rates go down, upsell goes up. How do you think about sort of your conversations with customers, particularly larger customers, where you've been super successful? You guys just mentioned just last quarter the breakout of revenue. In terms of sort of getting invited to that, you described as from the kid's table to the adult's table. I mean are all these things connected? Or how do you think about that? Does the product and the portfolio reach your critical mass and you have sort of enough customers that, that beget customers? Kind of help us walk through that.

Yes. Again, I think we look at the great companies that have come before. So we look at a Salesforce. We look at an Adobe. We look at those companies that have made that migration from selling individual products to selling suites of solutions for companies. And for us, I think that we sort of stumbled into the fact that once you sort of put 4 different products to Cloudflare together -- and for customers, it's often a Mastermind game, right? Customer A might have -- or customer 1 might have products A, L, Q and Z, whereas customer 2 might have C, D, E and R, right? And so it's kind of mixed and matched together. I think as we look forward, what we really think -- again, the companies that make it to sort of the next stage, the adult's table or $100 billion market cap and up, that they are selling what is almost a site license of all of their features as one bundle and that they're finding ways to expose all of those features to customers so that they can adopt it, and that becomes incredibly sticky. If you think about how a Microsoft or an SAP or an Oracle or increasingly a Salesforce or an Adobe, how they're able to really sell a much more complete solution, that, I think, is what's really, really the interesting opportunity for us. And where we think we fit is in providing those network services as a feature. And as I look at sort of big budget dollars that are getting spent that seem ripe for disruption, if you look at all of the Fortune 500, almost all of them are spending north of $10 million on their sort of telecommunication services, whether that's connect -- basic connectivity to the Internet from their offices, branch-to-branch connectivity over MPLS lines, other ways of having those network services. And they're largely undifferentiated. They're not products that people are particularly loyal to or wed to. And there are relatively dumb pipes that are out there. You can really see how we are now building together what is that network of the future and if we could provide that to those sorts of companies. We're already seeing how kind of the early adopters of Cloudflare One are essentially getting that kind of functionality. But I think that's how you can see a substantial portion of the Fortune 500 spending multi-digit millions of dollars with us on an annual basis.

Yes. At Wells Fargo, the initiative is called our virtual front door, sort of like -- sort of the one way and how it was sort of the network or the front door to Wells Fargo of the future look like and the -- so stay tuned tomorrow for that session with our CISO. It's -- but one follow-up on that, too. I think this is a key point. It's not just simply you have a broader product portfolio and more to sell, but you have a unique -- in that it's not just network and it's not just performance, not just security versus the point vendors because to your point, there's not sort of a single path that a customer takes or a unique path, let's say, any customer takes across that Chinese menu of yours. How do you think about just sort of the competitive differentiation of having so much under one roof?

Again, that -- the way -- most of the people that are just building one thing well become features over the long term. And they -- you build a great company that does that. But from the beginning, we really set out to build Cloudflare as one of the stand-alone, iconic technology companies that can stand alone. And so when you're looking out at what are the areas that are just ripe for innovation, providing network connectivity for enterprises that has the intelligence and security and reliability built into it, that's a huge opportunity. If you could spend with us and know you don't have to worry about a significant portion of the security challenges you had before, you don't have to worry about being knocked off-line by some hacker kid, you don't have to worry about a lot of the identity and access management challenges in the past and that comes bundled into one piece and costs about the same as what you would have spent with your ISP before, that, I think, is a very compelling opportunity for us. And those are not budget dollars that needed to be created out of thin air. Those are budget dollars that are ripe for the taking. And I think that that's where we're spending a lot of time thinking about as we go to market with larger and larger companies. And we would love to be Wells Fargo's front door.

It's like -- well, I'll talk to my guys. I've been pitching you. So I hope [ you'll be entirely supportive ] when it does happen for me. I want to get presents throughout this year. No, the -- but the -- it's -- but last but not least then, Jayson, everyone go with this Workers. I mean I think I'm the original Workers love. I've got it tattooed on my shoulder here. But the -- but give us an update on Workers. Obviously, we've seen a ton of potential there on edge, your third-party code running on your network. I guess you have 2 questions. One, just an update. And two, what has surprised you about some of the workloads or use cases that you've seen in Workers?

I think this is one of the most exciting and misunderstood areas that's out there. I personally hate the term edge because Cisco describes themselves as an edge company, and I mean everyone describes themselves as an edge company. If everyone is an edge company, then edge doesn't mean anything anymore. And so I'm not quite sure how to describe what this is. I also kind of hate the word computing because actually, the computing part is the least interesting part of the entire thing. And most of what's been focused on in this is performance. And it turns out there aren't that many applications that are so sensitive to performance, that shaving a few milliseconds off makes that big of a difference. There are some but it's maybe $100 million opportunity. It's certainly not a $10 billion opportunity. We have an internal drinking game wherever -- whenever one of our competitors or an analyst mentions driverless cars as like the future of edge computing, we all take a shot. It's -- that's not the future. I mean maybe it is, but you certainly can't forecast a TAM around it. The piece that I'm the most excited about in this is so incredibly boring but incredibly lucrative, which is what your CTO and CIO and General Counsel at Wells Fargo, what anyone that's running a global business is terrified right now of is increasing regulatory burden and compliance around data sovereignty, data locality and what you're seeing coming out of countries like Germany, India, Singapore and Brazil. If, all of a sudden, we wake up where every country says, "My users' data has to stay local in the country and has to be processed in the country," there really aren't any good public computing platforms that allow you to do that. And where we're seeing the most interesting use cases around edge computing has a little bit to do with computing, but it has a lot to do with edge storage. And so there was an announcement of a product called Durable Objects that we made during our Birthday Week, which was the week of September 27. That's enabling a set of ways that companies can better comply with an increasingly complicated regulatory environment. And watch this space. We like to announce a series of products over what we call Cloudflare Weeks. We've said we've got at least one left before the end of the year and there aren't that many weeks left before the end of the year, but it's going to be focused on privacy and compliance. And that's the $10 billion edge computing opportunity. It has very little to do with driverless cars. It's very little to do with IoT. It doesn't have very much to do with 5G at the end of the day. But something as boring as regulatory compliance is something that every global company cares about, and Cloudflare is uniquely positioned because we're already in 100 different countries around the world and can help you comply with those rules.

I love it. So I think we will leave it there. We'll stay tuned for the next -- for the coming weeks. All right. Well, we're excited. And Matthew, thank you for joining us. Sorry we missed you in Park City, but everybody's coming eventually.

Come on out. The ski resorts are opening this week, and the weather is great and look forward to it. And thanks, Phil. It's always a great -- it's awesome.

Fantastic.

Thanks, Phil.

We could for another half hour. Thank you, Jayson, for making it all happen. You're the man. So thank you.

Bye.