Cloudflare, Inc. (NET) Earnings Call Transcript

And once you have that...

And then, so I didn't know really what to do. So I started teaching law at a school called John Marshall in downtown Chicago. And as a final exam question, I was really interested in this question of jurisdiction online like, the Internet created for the first time the ability to be sitting here in San Francisco and commit a crime in Saudi Arabia, simply by sending an e-mail. And so I was looking at a lot of sort of the regulations around mail and postal mail. And I find a question, how would you create a jurisdictional hook for something that is like e-mail where you don't -- when you send an e-mail, you don't know where the actual recipient is? And through kind of grading those exam answers and some of those ideas, this idea of, "could you associate sort of a jurisdiction instead of rules and laws with something as femoral as an e-mail address" emerged. And so Unspam didn't really started that company. It started as an entry in a blog where I wrote about how you could build what would be like a do-not-call list but for e-mail, and describe how you do that technologically. And I got a call -- we got a call almost immediately from Senator Schumer's office in D.C., and they said, "Will you come consult on this, this anti-spam legislation being called CAN-SPAM. And there was a piece of that, that was supposed to create a federal -- an e-mail registry. And I was basically -- it was me sort of advising on here's technically how you would do it. And it never got implemented on the federal level, but a bunch of states took it up. And again, I was sort of called in to consult, and then the next thing that you knew, we got to sort of the end of the process, and they said, "Well, now you should bid on the contracts to provide it." And so I've sort of assembled a monthly crew of people I've gone to high school with actually. And one of the worst ways to pick your cofounders is who you have walkers next to in junior high. But the 3 of us, we were all some version of nerdy, wonky geeks. Started sort of this coming called on [indiscernible] we immediately had these 3 -- 2 state contracts to provide to client services.

You guys were how old?

How old?

Yes, you guys were what?

In our 20s. And going up, I remember bidding against EDS and Oracle and all these big things and trying to figure out how government procurement works. We have no idea what we were doing, but it was pretty amazing. We literally wrote the laws that we've then implemented the technology for. And the way that it worked was, it was basically the Superman 3 or office space business model where we collected a fraction of $0.01 every time you send e-mail, if you were in one of the regulated categories. And we didn't care, I mean, we could have been selling dishwashers for all we care. The states though really cared about sort of sin products, so alcohol, tobacco, pornography and gambling. The alcohol, tobacco and gambling folks saw it as a rounding error. The pornography folks thought as a violation of their civil rights. And so...

But they had to pay the $0.01?

They had to pay the fraction of a penny. And so one day, the process server showed up, and one of the classes, I guess, I missed in law school was that if you're providing technical services to the government and your technology is being used to infringe on someone's civil rights, you are jointly and severely liable for that with the government. And so the next thing we knew we had a major federal civil rights, first amendment, tenth amendment, dormant commerce clause case being brought by the pornography industry against all of us. And we had a pretty strong legal case. At the end of the day, the judge that decided it actually cited a law review paper that I had written, which was a little circular, but here we go. And the company kind of lives on, but I took a sabbatical because we were just trying to cut costs and survive and ended up going to business school because I was trying to figure out what -- how I could spend 2 years while the lawsuit worked its way through the system.

All right. So then you go to Harvard. So now we're up to '07, right?

Yes.

Yes. And I heard on the earnings call, I heard you had a nice acknowledgment, sort of shoutout to Clay Christensen. You said he's one of your professors. Tell us a little bit about that, he's one of the innovators to one of...

Yes, he taught a course called BSSE which I took in the second year, right around the same time that Michelle was...

B-S-S-E.

Yes, building sustaining-something enterprises (sic) [ Building & Sustaining a Successful Enterprise ] it's like, I don't remember all this, sort of known by its acronym. And it was sort of exactly the same time that Michelle and I were talking about starting Cloudflare. Michelle was a fellow classmate. She was not in that class but we were in the same cohort, our first year. And Clay -- I think Clay always thought that I was Mormon because I grew up in Utah and Clay was LDS. And I didn't disabuse him of that. And so we got to know each other and we became -- and I really just enjoyed that. And it resonated a lot with what I've seen in businesses that have been really successful, where real innovation comes not from the top-down but from the bottom up. And when we looked out, and Michelle and I spent time, we're looking at companies that we really admired. At the times you looked at like Salesforce which at that time was about -- was 10 years old. It's about as old as it's cloud for us today. And you just watch how they went to market, going after the long tail that were underserved by the Oracles of the world and then steadily just kept learning from that and being able to march up market as a result. I think that both what we're sort of learning and I was wondering in Clay's class along with what -- as we looked at companies that we really admired, Michelle and I were sort of seeing, as we were thinking about what the business model for Cloudflare should be, and those things really lined up. And I think if you understand sort of that approach, that, I think, is a very reliable way for SaaS companies to go to market, and I think some of the ones that are the most successful today have followed that path. And I think that's the path that work.

Yes, the path means starting out at the bottom and gradually working your way up.

Yes, and it helps you in ways that are hard to appreciate. So one of -- oftentimes, when people look at our business and they say, well, you have all these free customers, how -- what are you doing to convert those into paying customers? And we spend some time on that, but there's so many other benefits from it. I think the one that is the biggest, but it's the hardest to appreciate is that if you're trying to be innovative, and your customers are all spending -- and you have 10 customers spending $1 million a year with you, it's really tough because you pitch one of those customers off, and/or you break whatever it is that they're doing, it's hard. So the development cycles become really, really slow. Whereas -- and that's why if you look at traditional software companies, they have 2-year development cycles. If you look at traditional hardware companies, they have 2-year development cycle. Intel used to be Tick-tock, but it was -- if you couldn't go out, it was so costly to make a mistake that you couldn't afford to. Even cloud companies, it seems Salesforce is really proud of the fact that they have 3 releases a year, whereas if you look at a Facebook or a Google and they're doing multiple releases a day. And I think the difference is that if you've got a set of customers where you can run experiments and you can do that in a very low cost way, that allows your rate of innovation to be substantially faster. And I think that's why we've been able to stack innovative new products on top of innovative new products and get them to market as efficiently as we have. And again, I think that, that comes back very much to a lot of what Clay wrote about.

It's super cool, it's super cool. It's a little bit like that with stocks, too.

With stocks?

Yes, yes. If you own just a little bit of the stock and it's inconsequential to you, you tend to be a really good decision-maker about when to sell it or when to buy it. If you own way too much, right, then you tend to overestimate the upside and underestimate the downside.

What we are now have [indiscernible].

At a cost there. Okay, so you and Michelle were at Harvard, and how do you get this idea that you're going to start the company? Was it for competition?

We -- so it wasn't supposed to be a company. It was -- so I -- while we were Unspam, while we were trying to kill time and keep our engineers engaged as we were working our way through this crazy lawsuit, the -- we were coming up with just interesting projects to work on. And one of them -- and we did all kinds of things. We did a thing where we predicted the winners of the Sundance Film Festival in advance using basic statistical analysis, which sounds crazy at some level but if you think about it, it's the same technology that you use to tell what's spam and not spam. And if you're a sensible programmer, you know what movies are good or bad. You know what movie got into the festival because it was Robert Redford's childhood friend versus what's going to be the next really great film. And it turns out that as you write those descriptions, little clues come through. The same thing is true with real estate listings. And the general heuristic is -- and the system had identified this, but if nouns are good and adjectives are bad, so like if you -- you mentioned with the real estate listing, if it's like it's spacious or cozy or whatever, that's -- those are all terrible indications for us. Hardwood floors, granite countertops, those are things that are very, very functional, and it turns out that you can track that real estate listings, where they sell versus listing prices and the more nouns they have, the better and the more adjective they have, the worst. So we did that. We had a thing that allowed you to swap your Google cookie with other people, kind of lost in a crowd to make it hard for Google to be able to track you. And then we had this other project that was called Project Honey Pot, and it was a system to track bad guys and how they operated online. And tens of thousands of people have signed up for this thing. And I remember I was sort of sick of business school by the second semester or second year, and I was trying to convince Michelle, who I had not had my locker next to in junior high, but instead, I've watched her over the course of the year, and she just -- all the things that I was really bad at, she was really good at. And we were just really very complementary in that way. And so I was like, "We should start something." And she was like, "I don't know, I don't know, I don't know." And I finally was describing Project Honey Pot to her and I think she was so puzzled why would anyone sign up for this weird wonky thing that she said, "Okay, let's do that." And the original plan was -- I mean, the only name that's worse than Cloudflare because "clo" and "flo" are really hard to say next to each other. And -- but it was what the original name of it was which was Project Web Wall which was Barbara Walters' nightmare. And the idea was we were going to build a box and we're going to take the data from Project Honey Pot and we're going to load it into this box. And that was -- and we were going to go sell these boxes. And as we did it, we started to look at the -- just the economics around...

Wait a minute, Project Honey Pot was tracking the bad guys' e-mail addresses.

Well, it was tracking how bad -- it was anyone who had a website, we would give them a little piece of code that they can insert and then it created effectively a trap on that website. And if anything fell into that trap, then we would get data back. So an example would be, we would put a login form on a page, and then we would look at people who are just randomly inserting credentials in it. Or when you would access the page, we would generate an e-mail address. It looked like a perfectly normal e-mail address but it was actually unique to that moment in time and that allowed us to track this through. And what -- I mean, this is a hobby. We think a number of times we almost shut it down, it's crazy. And then I remember when I was in business school, that I got a call from -- someone from the Department of Homeland Security, and they said, "Do you have any idea how valuable the data that you have is?" And I was like, "No." And they're like, "How much do you charge to license the data?" And I kind of added up all the costs in my head and multiply for over 3 years that we built it then multiply it by 10, and I was like, it will be $20,000 a year. And that check was FedEx-ed so fast. And I told the story to Michelle, and Michelle was like, "Okay, maybe we should start a company around this, but what's for sure the case is you're not in charge of pricing."

Yes, so what's Michelle good at and that you're bad at and vice versa? High level? Just -- she's not here, right?

Yes. I think so. I think so. Michelle might be on the webcast. Michelle is -- she also has a Six Sigma black belt. Michelle is the operational gene. Michelle is a person who focuses internally and make sure that we have process behind how we hire people, how we develop products. We do those sorts of things. She's -- but right now, really focused on how do we think about our go-to-market strategy and make sure that we have that in a way that's going to continue to scale going forward. So Michelle is the real classic COO in a COO role to the extent that when we started the company, she was like, honestly, I don't think I've earned the title COO, so what if I just don't have a title at first. And I mean, I think the humility of that is really pretty amazing. Michelle has sort of ego in the Freudian sense of egos, that there's a right and wrong in the world and there's a direction, we should aspire to it. She has 0 vanity. And so if you just imagine kind of the opposite of that, that's what I'm good at.

That's great. Okay, so how does this thing really get going? So there's something at Highland Capital in '09, right? Was that...

So we want -- we ended up -- Project Web Wall, we pitched to, again, in Tom Eisenmann. He's our entrepreneurship faculty. And I basically wanted to get out in class in the second year of business school. And Michelle wanted to learn how to write a business plan and pitch to investors and do all those things, which, again, describes the difference between our personalities. And Tom said, listen, I'll take you on, I've got way too many projects, but I'll take you guys on, but you have to enter the business plan competition. So we did. We entered the HBS business plan competition. We ended up winning it. It was a pretty -- the class of '09 out of HBS is pretty amazing group. You've got Jen Hyman, who started Rent the Runway; Chris and James who started thredUP. Maybe the most successful company actually was Will Dean, who started Tough Mudder, which -- it just generates so much cash. I mean, it's an incredible business. And then after -- and there are a whole bunch of folks. And all of these...

These are all in the competition?

All were in the competition. And we ended up winning it. And I remember that...

You beat Rent the Runway?

We did beat Rent the Runway. So -- and thredUP at the time...

What did you get, by the way, when you win? You got to get something.

We got -- I think you we got some money. And I think that we got some free services for various things. I think I still have -- there's some incubator in Cambridge that if I want to just go crack it, I've got a year of free space. But what we really did is we got phone calls from every BC in the world. And Michelle at the time had offers to go work at Google and lululemon and LinkedIn. She was actually being recruited very heavily by LinkedIn. And then -- and it wasn't clear that -- again, this was a school project, so she -- Highland Capital came and said, "Hey, we'll give you space and a little bit of money and you guys can incubate the idea." And so we moved out. My mom and I loaded a U-Haul with all my stuff and all of Michelle's stuff. Michelle was dating a guy who she got married to who's based in Vancouver. And they flew back to Vancouver. My mom and I drove the U-Haul from Boston to San Francisco and we showed up and start working on the idea out of Highlands office down in Palo Alto in December of '09.

Wow. All right, good. So we're going to have to skip 2009 to 2019. There's a story. All right. So let's talk about what's going on today. We'll start off at the top, which is how's business?

Business is great. I mean, I think that we identified back in 2009, 2010, that there are a whole bunch of boxes that solve real problems that people have, that the way of the world -- at the time in 2009, we saw -- I think we were smart enough to see that the world was going to shift to SaaS. I don't think that we were smart enough to see just how incredible the shift to public cloud was going to be as well, but those things lined up. And once that happened, the problems of security and reliability and performance, those don't go away. But the solutions fundamentally have to change. And so I think that, that trend has happened. I think it happened faster than we thought. And that one of the things that we realized early on was fundamentally Cloudflare's business was how can we process a byte of information cheaper, more efficiently than anyone else. And so we have relentlessly, over the last 10 years, built what is today one of the largest networks in the world where we're in more than 200 cities, we're in most of those locations, we're in multiple different ISPs, multiple different data centers directly connected with the vast majority of the world's networks. And we've built this incredibly flexible platform which started out servicing the long tail, but today services more than 10% of the Fortune 1000. And we just see every single day boxes being retired, and those same problem that those boxes used to solve shifting to Cloudflare in order to be a service that can actually scale to what the kind of modern IT stack looks like.

Yes. I like the way your growth is trending, right, because a lot of the companies I cover, you -- it gaps up. Those 40, 50, 60, 70, 80 like quarter-to-quarter. And then the problem with that is you know that it can't accelerate forever. And then -- but you guys have got to serve this 48, 49, 50 and this last -- I mean, that's not exactly in that order, but in this last quarter, 51, right? So what's causing this sort of nice, steady slow acceleration?

Yes, I had dinner at my house last night with a whole bunch of sort of tech CEO folks. And Eric from Xoom was seated next to me, and I didn't...

You host the dinner? Was Eric's phone ringing off the hook?

Yes, it was dinner. No, I can talk to you about it [indiscernible] I wasn't supposed to bring it up but Eric is one of the nicest and just most humble and just a great guy. And I think we both admire each other's businesses for different reasons. I admire -- I mean, I guess, fundamentally, the growth rate that they have, but really, it's the very straightforward go-to-market promotion they have which is go by a list of people who use buybacks, call them up and say, I have a product which is better, and I'll charge you less for it. Here's the way to try it. And that's amazing, and that's very, very simple and I can tell that story. I think the thing which is challenging about Cloudflare is there's not a clear list of people that we just buy a list and say, "Hey, rip this out and replace it with us." Instead, our go-to-market motion tends to be an organization has a gap somewhere in what they're doing. So they have -- they're a major CPG firm that got hundreds of brands that sit behind them. One of those brands has a CMS that it's using for its web presence, that has a vulnerability. Their existing hardware-based firewalls don't cover that vulnerability, so they need to put something in front of it. And so we will go -- they will find us or we will go to them, get in front of that, solve that one little problem. And then what ends up happening is we win the war of attrition because at some point, their F5 load balancer falls over and die. And then the F5 salesperson comes in and goes, "Hey, do you want to buy another one of these?" And then, "We could do that or we can just click this button in Cloudflare and get load balancing." And then their VPN falls -- their Cisco VPN falls over and they say, well, will you go and replace that or it looks like this new Cloudflare for Teams and Access product replaces that. And so we tend to do that. And I think that when you look at our business, what it looks like is we have products that are a series of S-curves. And what we try to do is stack those S-curves up one after another after another. And so various products are going to be at different stages of that. But I think what Eric would say, and I'm putting words in his mouth, but what he admires about our business is it's really clear what our second, third, fourth, fifth, sixth, seventh, eighth active would be. And, yes, we're at different development cycles in that, but when I look at what products are -- how quickly -- as we take a new product in, you can just watch as they all kind of follow that same cycle. And what's powerful was we built a network and we have this diversified set of customers that allows us to rapidly innovate and put new products in market. And our strategy is, what I tell our team, if we come out and we're top right in a Gartner Magic Quadrant the first year that our product is in market, we waited too long to launch it. Whereas what we want to do is we want to get to market quickly, and then we want to move up into the right every year faster than anyone else. And within that 2.5 years, we want to be in that upper right quadrant, but we want to start somewhere in the lower left. And where we think the world is going is that there are going to be probably 5 networks that matter in the world. And in the past, you would pick one solution provider. So if you needed load balancing, F5 was the leader, and yet Cisco sold it and other people sold it, but F5 was the category leader and you'd pick best-of-breed for that one solution. The challenge is -- and the reason that worked in the hardware space is if you could rack and stack a heterogeneous set of products next to each other, it's why RSA -- if you just never want to be depressed for humanity, get a badge, walk around the floor at RSA, and it's just box after box after box that does some slight variation on some different thing. And that works because as long as you could be a category that Gartner would define, you could stack these boxes next to each other. And there was no technical penalty because you go from one box to another, and it's microseconds, there was no support penalty. That was -- that was Cisco's sales pitch, but it turns out it didn't -- it was not that much harder to support 6 different vendors than just 1 vendor. And there was no cost penalty because the boxes all basically cost the same. As the world shifts to the cloud, that changes. And so you can't have a routing-as-a-service company that is different than your load balancing-as-a-service company, that is different than your firewall-as-a-service company, that is different than your VPN-as-a-service company. You can't have these point cloud solutions stacked next to each other because going from one cloud to another is 100 milliseconds. Each of those -- I mean, support costs, you call your routing of service company, you say something broke, they're going to be, "It's not us, it's a load balancing of the [indiscernible] company. And then the -- every one of them is going to have overhead that they have to cover with expensive top of rack router and relationships with ISPs and all of those different things. And so what we saw was, fundamentally, you're not going to pick who's the best load balancer, you're going to pick who is the network that has good enough load balancing and then good enough firewall and good enough these things, and then the data from all of those services allows them to get better and better and better, so that over time, they'd become best-of-breed. And so I think you're going to have a very -- a relatively -- where the world was in the past, thousands of different box vendors, that's going to consolidate down to a world of relatively few network providers that have to provide this functionality. It's the old Jim Barksdale quote is that there are only 2 ways to make money in tech, bundling and unbundling. What you're seeing in sort of the network security space is bundling, and that's what we're doing. I think that's what Akamai is doing. I think it's what you'll see the major public cloud providers do. What you're seeing in the sort of SaaS software space is unbundling. You're taking the suite of SAP, Oracle and Microsoft and unbundling them into their component parts, where it's CRM and ERP and collaboration, but then even verticalizing it further where you can be a $20 billion company doing CRM for drug reps. And that makes total sense. But where the cloud has caused there to be many winners in the software space, I think there are going to be much -- many fewer winners in the space that is the network that sits in front of all that software.

SaaS, unbundling, network bundling. We'll start after that. Okay, let's pause and see if we have any questions from our audience. Yes, in the back.

[indiscernible] Can you talk a little bit about some of the stuff that are really catching your eye on Cloudflare Workers both here and the next year? And where do you see that offering going?

Would you mind just repeating it for the webcast?

Yes, so the question is what's really catching our eye in terms of Cloudflare Workers and where do we see that product going. So Workers is Cloudflare's edge computing platform. Traditionally, there have been 2 places you could write code. You could write code that ran on a client device. So -- and the reason you would do that is if the latency between the user and the application really mattered. So gaming is a classic example where you have a game that runs on your local device, and that makes sense. Or you could write code, and it would be in some centralized data center that would be somewhere. And when you would choose, that would be of latency between the application and the database mattered the most. And the classic example of that, that would be Search, like there's no way to take Google Search and run it on your laptop, right? It has to be kind of in a centralized place. And what we realized was there's an opportunity for a third place to write code, which is actually build code and deploy it in the network itself. I think that Workers is out. We're seeing 20% attach rate among new customers that are using it. And I think that it has a short-, medium- and long-term opportunity. The short-term opportunity is that it makes our platform the most programmable and extensible in the world. So you want the most programmable load balance in the world, Cloudflare offers that. The most programmable firewall in the world, Cloudflare offers that. Because what we see is that companies are able to write how our network functions in a really powerful way. So New Relic is an example of this customer, signed up for Cloudflare in Q3, used us for some basic security and performance services and then start playing with Workers. And they said, "Wow, we're this application monitoring program. We need to be online no matter what, so we can't go all in on AWS or Google or Azure because if we do and one of our customers is using them too and AWS falls over, we need to be able to tell you that that's the case." And so we offer our load balancing product that allows you to be multi-cloud in various ways. And he said, yes, yes, that's great, but we want to write really sophisticated logic that says something like if this customer is monitoring AWS, then don't monitor it from AWS, monitor it from Azure. And that's not something that we would ever build into our own application, but it was something that really mattered for New Relic. And so they were able to sit down, write a bunch of code, deploy it into our network, build this incredibly powerful load balancing function and increase their spend with us from Q3 to Q4 dramatically. I mean by almost an order of magnitude. And that's sort of, I think, the short-term opportunity. And I think that's following in some classic examples. The F5 won load balancing space because it was the most programmable load balancer that is out there. Palo Alto Networks has one that's sort of next-generation firewall space because they have the most programmable platform to be able to extend the boxes. And so I think as we do that, that's going to be the near-term largest use case for our customers. And that's the short-term opportunity. I think the medium-term opportunity is actually why we built Workers in the first place, which is we built it for ourselves. And as I look out at development platforms that have been successful, I can't think of an example of one that didn't start by being built by the developers of the company that built it for themselves first. So that's how Microsoft built Windows. That's how Salesforce built salesforce.com. That's how Amazon built AWS. You build it for yourself first. And then the key, I think, is to implement one rule and hold it as absolutely sacred, which is any tool that we built for ourselves or any API that we built for ourselves, we also expose to our customers. And so that's what we're doing. And so if you look at new products that have come out, whether it's Cloudflare for Teams, which is our -- I mean, it's our Zscaler competitor, whether it's the S2 acquisition that we just did, all of these ride on top of the existing Workers platform, and it's allowed us to have this innovation engine where developers can develop and deploy code incredibly quickly. I think the long-term opportunity is the one that has the hardest time sort of assigning a TAM for. And there are plenty of people out there who will do it. Gartner will tell you that it's been like by 2025 or 2030 that 25% of workloads are going to move to the edge. I don't know. If that's the case, that's really exciting for us because we are in market with a live production edge development platform that more people write for on a daily basis than anyone else that's out there that's trying to do this. And so I think that's super exciting. But I don't know. I think it's going to be interesting to see what use cases really drive that. I can tell you some that are -- that I think are interesting. We have -- there's a large payment processing company that processes a lot of credit card transactions. And it takes on average 4 seconds for a credit card transaction to go through. They think that if they move that out to the edge you could get that down to hundreds of milliseconds. That would be extremely valuable to them. That, I think, is an interesting use case for this. But over time, I think I don't know how many of those workloads that are running on EC2 today are going to eventually run here. I think it's going to be some. It might be a lot, but I think it's early to sort of forecast what that long-term opportunity is. But if it's just those first 2 things, honestly, it's just that first opportunity that's giving us the ability to win larger and larger deals. And what we also see is that once someone writes code and deploys it on our network, then it becomes very, very difficult -- the stickiness of our solution becomes super high.

That's super. Okay, one last question, which I've asked everyone else today which is, is the coronavirus having any impact on your sales cycles or your leads or anything? And then how are you thinking about it in terms of marketing events and things you would attend or you would host?

I think that -- so dinner last night, it was -- we joked that we should have sort of a contra coronavirus ETF because it was Eric from Xoom and Dan from DocuSign and Jeff from Twilio and Stewart from Slack and Patrick from Stripe and me. And it was -- and I can make a case...

All these people were in your house? That was some dinner party. What was the theme?

Yes. Techs -- it was a bunch of tech leaders with a bunch of very senior U.S. government folks. And I think that as -- I think the tech industry as a whole has really done a disservice by not engaging with government. And I think government has actually done a disservice by not engaging with the tech industry. And so we posted this center for the last 3 years and it's been -- it's a just very casual, good time to get 2 groups together that really actually have very similar common goals. And there are going to be a handful of places where we're going to have knock down, drag out fights, but in a lot of other places, I think that there's a lot of respect. But I think that the short answer is that the places where we're looking at that are -- in our terms of our supply chain for building and delivering servers, we've taken -- we have a diversified supply chain on that, but that's somewhere that I think we're watching carefully. I think that the place, that there might be in the most fragility in the supply chain will be around the supply of flash coming out of Korea. So if you really see the coronavirus take out -- off of Korea, that, that would be a place where you might see supply chain challenges for other companies. We've tried to get ahead of that quite a bit, so moved up. And I think we feel, at this point, very comfortable around that. I think there are definitely changes...

How many of these boxes -- it's a commodity, right? How many do you ship a quarter?

So we don't talk about that enough, I guess, would be the right answer. And I think that we've really optimized what I'm really proud of with our supply chain is that we typically -- we can -- we've set it up to any of the major facilities that we're in. We're in 200 cities around the world. We sort of categorized them in terms of their priority. But any of the ones that we categorize as sort of the type 1, type 2 facilities, we can double the capacity in less than 30 days. So we almost think of it as just-in-time infrastructure. And what that often means is that we can generate revenue from CapEx before we have to pay for the CapEx, which is part of how we have the nimbleness to be able to do this. And it's part of what's allowed us to, as we saw risk from coronavirus, adjust what we're doing. So we don't see a risk to that -- we don't see an impact to that today, but it's something that we're definitely watching carefully. From a marketing perspective, there are definitely trade shows that are being canceled. And so -- but that just shifts around how you make marketing spend. It has an impact to us in a lot of ways, and I think that it's not quite the direct relationship that -- I can make a super bull case for Eric at Xoom or Dan at DocuSign, where, like, I'm not touching someone else's pen to sign a document anymore. I mean, DocuSign makes all the sense in the world. But I think that anything that drives more Internet usage is good for us. And I think anything -- one of the things that we started talking about yesterday and that we're starting to get more inbound interest from is our Cloudflare for Teams product which really makes supporting remote workforces much more effective as you have more people working remote, I think that drive -- it makes it harder to solve problems without Palo Alto Networks box or a Cisco VPN. And so that then forces you to the cloud in an interesting way. That's the long-term bull case. I don't think we're going to see any effect of that in the short term, but I think that these are the sorts of -- when you're having continuity planning as an organization and you're thinking, okay, if we had to say no one come into the office, what would we do? I think that, that's just one more thing, which pushes you away from hardware-based box solution to a cloud security solution like Cloudflare.

Wow. All right. Matthew, thank you.

Thank you for having me.

A lot of good stuff to think about. It's really great having you here.

Absolutely. We really appreciate it. Thank you.