Fastly, Inc. (FSLY) Earnings Call Transcript & Summary

Great. Thank you very much, Tammy. My name is Alex Henderson. I'm the security analyst at Needham. It's a real pleasure to have Fastly here. We're going to do a fireside chat. We have both Josh Bixby, the CEO; and Adriel, the CFO. And if you want to ask questions, I would love to have you as interactive as possible, and the less questions coming from me, the better. I've got 26 of these things, so I'm happy to take a back seat. But there's 2 ways you can do that. You can e-mail me at [email protected]. I'll keep my e-mail partially visible, which is not that very easy to do on these computers, but I'll try to, as well as use the question box down below. And with that, welcome, guys. Thanks for joining us.

Let me start off with the obvious question, which is stocks had a much larger hit than the average high-flying tech stock. I think some of that has to do with people's perceptions of you being a beneficiary of last year's closed market and maybe not being as much of a benefit going forward. I currently think -- to me, it looks like a big opportunity to the extent that you're obviously very well tied into the broader community for coding and microservice-based delivery of your technology. And when you get a sell-off like this, it's is usually a great opportunity. So with that, can you give us a little bit of color on what's happened? Why the stock has reacted that way? And how temporary do you think the current pressures are?

Alex, it's a pleasure to be here. Thank you. I think if you look at our core business, which is where about your experience, everywhere around the world, being fast and secure, like you need to track down to that iron level, I think what you see is everyone wants the world to fit into these 2 camps: you either benefited from COVID or you didn't benefit from COVID. And therefore, your future trajectory is impacted by 1 of those 2 things. Fastly is in this incredibly unique position. We benefited from COVID because your experiences everywhere around the world, fast and secure, became really important, but what we're also seeing is coming out of COVID, we're seeing that, that same underlying trend is accelerating for a number of industries. And so we are in a strange position, I think, which is taking the market some time to comprehend, which is not only are we benefiting -- that we benefit from COVID. We're benefiting on the way out. And if you look at what we said by increasing annual guidance and talking about the acceleration we're seeing in Q3 and Q4, that is very different than a COVID stock. I mean, a lot of stocks that are focused on selling, whether it's mask or plexiglass, I mean, it's going to be a tough mask in plexiglass environment, we believe, in Q3 and Q4, we don't believe it's going to be a difficult environment based on what we see today for companies that are leading the trend on this new architecture, how the Internet is coming together. So I believe there's just a little bit of a misunderstanding perhaps about how you could fit into both camps, and we certainly feel extremely optimistic about our horizon because we didn't build this thing to grow for a pandemic. We didn't build this thing to grow for a year. We have now recently hit our 10-year anniversary. And every year, as we have been a public company, we've accelerated our growth. We believe that there are many, many years of high growth here. Just to put that into context, Fastly has just under 400 enterprise customers. If you look at our largest competitor, just directly in sort of a small view of our space, that -- there's 10x more. If you start expanding the view and saying who needs fast and secure web experiences, that is every organization in the world, any meaning of scale. And so 400 compared to hundreds of thousands of potential customers, but us, our perspective is we're in the early innings. So we're investing, and we see tremendous growth opportunities, and that gives us a lot of confidence.

One of the things that I think that people don't really comprehend very much is just how important your relationship into the coding community is and the differential between that 120,000 people -- or companies out there that are greater than 2,000 employees that you could potentially target. You guys actually have a fairly differentiated customer base, where you tend to be more to the born in the cloud companies, you tend to be tied into companies that are driving microservice-based, Kubernetes-deployed clouding, multi-tenant type architectures in their businesses. Can you talk a little bit about how important that part of your business is because I think that's really where you guys shine?

Yes, and you make a great point. I mean, this whole -- what was traditionally the developer and operator or DevOps movement, which is now actually DevOps. That idea that all these teams are coming together, and they are consuming products similar in some ways to how we consume consumer products. I mean, you don't have this huge heavy lift when you go sign up to try an app. It's easy. You can do it online. You don't have to talk to a salesperson. You can upgrade it if you want to get to the next level. You can see everything from pricing to documentation. Everything is transparent. So there's an entire community of people that work within large enterprises who obviously have a larger, more [ employees ] in procurement and security organizations. But we have, within that, this energy that get product skill. And if you look at our history, we have an incredible history. You look at our dollar-based net expansion, our last 12 months of NRR, you see that once we're able to get into organizations, which we do, as you see through these fluid channels, which I think we have a tremendous advantage, we're able to take that and build upon it extremely strong enterprise stickiness and growth. So as a business, we have less than 1% revenue churn on an annual basis. As we look at our past year, you have these incredible cohort growth. And you're right, the secret to that has always been that when I walk into a decision-making meeting, for example, with another C-level executive, I have an entire population who had tried my -- our products, who has used it and who's loved it. And that's not only the case on the delivery side of our business, but even more importantly, on the security side because there are a lot of people out there, as you know, selling magic beans in the security business. And to be able to try something before you commit to huge appliances that you have no way to get out for 5 or 10 years, I mean, it's really a different way to think of the market. And we've been able to gain real strength, and I see that accelerating. We just hired a new Chief Revenue Officer, and I've seen that accelerating in our business. The net number of new additions this quarter was really encouraging.

So if I were to think about that positioning, I kind of draw a parallel between what Okta did with the Auth0 acquisition. Okta had a huge position in reaching the CIO, CTO-type selling a particular product. But increasingly, the world on -- particularly on the consumer side of it, we're shifting more to a DevOps-driven model, and they didn't have that skill set to reach the consumer application websites and consumer type of transactions, where identity needed to be feathered into it. And that area is actually much higher growth than the enterprise side. It seems to me that you are much more like the Auth0 side than the Okta side in that context. Is that a fair way to think about it?

Yes. I think it's an interesting comparison. We have a really strong partnership with Okta and just the...

Great company.

Yes, I deeply admire their business and how they're going about it. I think one of the unique qualities of Fastly is we have that bottoms-up energy, but because we are the enterprise -- the modern enterprise company, we also have a little bit of the enterprise capability as well. So because the decisions that our customers make are so important when they choose to go on our platform, I mean, when -- during the IPO, I was asked questions and still am by media like if Fastly goes down, what will we hear about it. And the answer is you'll hear about it when Fastly is back up because the media that you consume, the websites that tell you this information have come down with us. We are so central in functioning of the Internet that these decisions are not made. Even though we have this ground up energy, they're not decisions that are exclusively made by any individual in here. Often, we have to go to the C-level. So I think what's interesting about us is we're sort of a little bit of a combination of both in how I see us grow. Now as you know, our approach to the SMB business is really different. We believe deeply in partners. So tomorrow, you are out to build a website, you had a hobby, you want to start-up a little e-commerce store. We believe, today and in the future, it is much smarter for you to go get a Magento, a Shopify, a Wix or one of these platforms that allow you at a click of a button, with an easy button, have all of the integrated security delivery capabilities that you need to deliver your store. So Fastly is making tremendous inroads into that community. We just do it by getting one customer who then has 1 million as opposed to going and getting them individually. That's exactly the same approach on the developer, who -- many developers want to have a place where they can host their code and talk to the world. Instead of doing that one-off, we go -- we partner with deep relationships with Microsoft, for example, Github, and we have a relationship there. So some of that velocity that we're seeing, we move customers, those millions of customers, you just don't see that in our customer number, but it's actually how it played out in the field.

Right. That's an excellent point. So when I think about the adoption of Kubernetes, which I use as a generic term for all of the various flavors of Kubernetes all the way out to serverless, it's pretty clear that, that is sharply accelerating. The statistics we like to cite is that there's roughly 700 million to 1 billion applications globally. They're growing at about a 30% annual pace, which is a lot of new applications. And the penetration of Kubernetes, which I think was 15% or less in '19, is now crossing into the 25% plus range on its way to in excess of 50% probably in '23-'24 time frame. Now when I put those 2 numbers together, that implies the growth rate of applications combined with the penetration rate of modern architectures, microservice-based technologies that you're looking at a triple-digit end market growth rate there, which you guys tap into in a unique fashion. Can you talk about the uptake of your programming skills and your programming languages and how people are looking at that?

Yes. It's a great question. I think deep insight into this claim, which as you say is people are being -- people are moving and being more safe on customer demand to really focus on the primitives of the future. How will these applications be delivered? And I think there's some really interesting spreads in those statistics. The first is people want portability, right? They want the ability to move workloads to benefit their business. They also are really eager to find ways to not have lock-in, right? I mean, if you really look at what Kubernetes does in some ways, it allows you to avoid lock-in, which is something our customers are talking a lot about.

Multi-cloud, yes.

Yes, right, and have a multi-cloud environment. I think one of the other things is by going to a microservices-based approach, you can apply -- you can have the right workload -- right work at the right level. And in traditionally, as you know, we used to have a myopic view that all the code we wrote had to live on a web application or database server housed in a central location. I mean, for 25 years, that's the way we thought about workloads. When you look at computed edge and the work that we've done for 10 years at the edge in bringing compute there, we have helped really galvanize this energy to hold on folks. There are just web application, database servers and the end device, right, what we now see is so powerful. There's actually this additional layer, and that layer can work in harmony. This is a layer where our developers that use our product can trust that these environments are compliant, that they meet their requirements, that they have the resources they need. And so we're really looking at architectures, which span from the device all the way back to the database server with very important workloads being run at the edge, the workloads that need to scale and that have the security mentality because we know that by having security at that perimeter, we can make a significant difference. So I guess, I would say all of this is pushing us toward really this true essence of serverless, which is a word that is very confusing because serverless had servers. What it means is our developers don't have to worry in -- our companies don't have to worry in the way they used to about making sure those resources were in the right place and scaled in the right way. That's our duty. So we worry about the servers. We have them in the right place, and we ensure they're ready to scale. And when you can gain that trust, what we have with the largest organizations in the world, you unlock so much creativity because you're not -- I need to provision 1,000 servers in Tokyo. Okay, that's a 6-month process. Like, you don't worry about any of that. We've got that covered. We have a huge network, we've got. And you can now -- with the advent of computed edge, you can now run those workloads wherever you would like. It's a very exciting time.

Well, so let me throw a phrasing in front of it, but I've got a question from the audience that ties into computed edge. But I think your nearest largest -- large competitor, Brand A, if we will, went out with a comment recently that they expected that today, less than 10% of applications are at the edge. And I think the number was -- in 5 years, they thought it would be 50%-plus at the edge. So can you talk a little bit about the cases you've seen so far for computed edge? And just define what computed edge is for those who might not know it.

Yes. So in general, when you look at an edge cloud, one of the capabilities of an edge cloud is the ability to take code written by developers and run it at edges. There are many edges. For us, we think of ourselves as the developers edge, the lap plates in this line, where we are close to users, and it can be trusted by the engineering organization in the enterprise in the world. So when you look at the use cases, they really bucket into a few categories. One is really easy for all of us to understand. We know that, traditionally, e-commerce has gone between this balancing act of I want to make things fast and I want to make them personal. And the reason that they actually have the dilemmas because personalization happens where the logic lives, which is down in the central location, and performance, we know its just a way faster if I can deliver it right next to you. That should be your business. And so what we've seen is a drive for sort of the perfect solution, where I can get personalization to happen at the edge, and that really what compute allows these organizations to do. So when you go to a personalized web page, why does that page have to transact all the way back to the central server in order to determine what you have access to, what you should see, what's recommended. That's not necessary. We had a type of computing. So that's one example. I think the other example is also proliferate. We see examples that actually is [ adherent ] and the compliant, right? Either these are the rules in certain countries. They don't want them to get an access to content that's only available in different country or I don't want user-sensitive data or I want the data out only in a specific location. This world of multiple geographically distributed Internet that are country-based, it's something that is not new, but were really accelerating. These are really hard problems for organizations to tackle on the data and compliance. So that another example. We're certainly seeing some really interesting use cases.

Go ahead. I'm afraid your audio went wonky for a second.

Is that better?

Now it's great.

We're also hearing some really interesting use cases as well around the application of machine learning at the edge. So we know that if you take some of these sort of trophy machine learning example, like I'm going to go train this machine probably [ using ] pictures as an example. So I take a picture, I take like a bunch of these pictures and run it through a machine learning app. Then I can give it a new picture. They hit their hot [indiscernible] these pictures. Well, the reality is all of that training is in front is computed edges, but the influencing from the model, you actually can do in at the edge for performance. Another area that we're really excited is actually security use cases. So imagine today, security is very much a centralized structure where we've got -- we're quite coming in. They had 1 central location. We made this decision that I can [ make ] in the agility way, where I think better at the edge to be making some of those decisions. And so when we have decision-making engines, we have human security. In our most recent announcement, we talked about [ security ] awareness. We talked about that [indiscernible] incredible because the largest organizations in the world are making huge advocacy. We know that. It's a massive issue. How do you get around it? Well, what's the best place to mitigate that, I should say, is [ on the edge ]. So we're seeing lots of different use cases. I believe, over time, the specifics that we're seeing in the industry, which is we're adding very early days right now, and we are moving to a mass adoption phase, that's certainly what we're seeing across.

So another question came in on computed edge. I have a couple of questions teeing up, but let's just stay on that one subject first. The question is, and I think it's a good one, how are -- how is your offering at computed edge different than, say, Cloudflare Workers?

Sure. So mainly, there are 2 sort of spectrums of requirements, and I don't think -- I don't know if I should say better or worse. I think everyone is building for their customer, right? And so I don't think this is a better or worse context. I would say, for a certain customer that doesn't have a huge scale, that doesn't have the security requirements, that come of a lesser larger enterprise customer, you have a certain kind of solution. It's lightweight. You've baked all these requests together. It doesn't need to start up that quickly, but that's okay because the needs are [indiscernible]. You have a completely different mandatory from customers who are the largest in the world, and they have different requirements. I can sort of put it into the perspective of large enterprise departments, which is different than non-large enterprise departments. So if you look at the 2 camps, and it's not just adds and pluses. There's an emerging energy here. You can see that, for example, with organization of [indiscernible]. There's a [indiscernible]. We are getting people in one orbit, which is WebAssembly, the orbit of WebAssembly and very much geared to these incredibly high-scale, high -- highly burdened -- highly security burdened workload where security really matters. And you have a more lightweight example. And so if you look at it, I'd say there's a couple really important elements. Performance is different, how important this is. So exactly, we can't go back to our customers and say, "It's only going to be performing if you do a few requests." Well, we're going to need performance because these are the large workloads in the world. So performance and the consistency in that performance. You'll see, in one camp, the world's highest performance, you'll see another camp with a lot of descriptions. You can only do a little bit of this. You can't do a lot of that. So that's performance. You also have scale, which goes hand in hand with performance. So again, I can't give you many of the competitors had [ time ] limits. You're going to need 1,000 of these. You're going to do a 1,000 of these of locations. And then really importantly for our enterprise customers to be [indiscernible]. Well, we believe each of these request or responses need to be in their own sandbox environment. Others believe it's okay to mix them together. The danger of mixing them together, of course, if you mix them together, you have one code base that can help intercept and foresee what's happening in another area. And you have the challenge of specter and meltdown type of tasks being really prolific in those kind of co-mingled environments. But I would come back to better or worse. I think certain workloads require a certain level of scale, security and performance. And certainly, others don't, and that I think it's more of a workload question. We conceptualize in those that really need a high scale, high performance and high security.

So a question came in. So everything in the telecom world has ultimately proven to be commodity because someone will be able to figure out how to do it equally as well or better, which pushes pricing down. And I think that's certainly the perception around the CDN industry as pricing of video content being delivered at the edge has gone down as fast as the volume has grown, resulting in a fairly flattish market. So the question is, what is it that is your secret sauce that allow them -- allow Fastly to gain share at a faster rate than their competition and is more durable?

Yes. I read the history differently. I think the most prominent story in the history in absolutely the video market, which has been heavily commoditized. I think the less total but equally important story is to look at what Akamai has done on the non-video business, which is they have -- for many years, had a business which is not commoditized in any meaningful way and continue to have incredible margins in that business. And for 20 years, they've driven and built a market, which is not very competitive because it's hard to do not because -- if anything else, it's just really hard to deliver the largest workloads in the world early in a performing way. And we think the history shows that they're 2 stories. One story is where it's easy, you're going to get commoditization. And traditionally on the video-on-demand side, except for organizations that don't have the desire or can't monetize their contents in a meaningful way, yes, they're sitting there trying to find ways to get the cheaper price. That's not in a space for a lot of workloads in the world. When you have high-value -- when you've got $1,000 pair of jeans, aiding [indiscernible] of any, if it means maybe on sale as many jeans. but you're never going to move down that [indiscernible] of work. And so what we have is the bifurcation of 2 worlds. We've got a commodity world, which is commodity priced. We had a high-value world that has good value just because it's commensurate -- it's the value that's been brought. So the answer is I don't think that, that's the actual way is to be involved in our market. I think the answer on the non-video side, it is extremely difficult to deliver the largest, most valuable workloads in the world. And ultimately, only 2 companies have shown that they have that capability, and that's us and Akamai. Now Akamai is still in the middle of this massive scale. And we are certainly respectful of how hard that is to built. So I think the answer, one, it's hard to do. You need to scale. One of the things to scale that is substantiating this, if we look in a rearview mirror, over the last 10 years, we haven't seen sort of network, peers and delivery of products scale that built on their own networks. And that's because 10 years ago, venture capitalist and even investors were at speed at the pipeline, plus [indiscernible] that in the other way, they preferred people who would say, "I'm going to build it on Amazon," or "I'm going to build it on Google." So when we look in a rearview mirror, adding 130 terabits per second network that is not in alliance with the largest cloud players is incredibly a unique asset. We [ need ] hundreds and hundreds of millions of dollars, if not more, to reproduce [indiscernible]. We have a moat in the network because we don't have investment. We've got a moat in the technology. I think we have a really good example and a lot of learnings over 20 years, that in fact that is really [indiscernible]. I think there's a lot there, and I would sort of disagree with the premise of how we've seen that evolve.

I tend to agree with your position. So the next question that came in was I have a general lack of understanding of how Fastly fits in with hardware and existing platforms, i.e. AWS and Azure. Could you clarify it? And the question is, will Fastly utilize NVIDIA hardware-software stack in their network? And how Fastly computed edge handle AI workloads given the processing requirements around them?

Yes. Well, if you look at our position in the ecosystem, we are, as I said, having a huge network, having tremendous amount of [indiscernible] storage obviously. And that allows us to have really complementary relationships. So look at our relationship with Google, for example. We go the market together. We know that it is great for our customers a huge physical cloud. We use Google Cloud. We want our customers to see that value as well. We know that our customers are demanding a multimode environment. One of the ways that they see themselves getting into that position is like adding a third-party like us, even the load down in the cloud, a place where you can centralize your logging, centralizing your key security workloads. Something that you want to reproduce that's over and over and over today [indiscernible]. So there's a really synergistic relationship with our product relationships. We have a customer relationship with all of the large clouds. And I think we're all in understanding that we all have an important part to play for our customers because what our customers are demanding is lots of choice, and Fastly helps deliver that choice. Now when it comes to AI and compute in general, you'll see us continue to innovate with our vendors and our partners in this space. So I think over time, we'll absolutely see more and more compute capabilities and unique [ customer ] capabilities delivered to the edge. I think that's important, and you should -- you will see more of that from us in the future as well.

Let's shift over to the security side. Can you just talk a little bit about the Signal Science acquisition, how customers are responding to it, how it broadens your security platform? And I think if we look at the broader market for security here, there is an increasing need for security to protect the applications across the broader web. How do you see yourself participating in the broader element of domains and domain traffic protection and optimization and beyond just the edge compute world?

Sure. It's a great question. So we've been incredibly excited in the Signal Science's acquisition. It has exceeded our expectations. We know, because we've been down the security for a number of years, how important it is for our customers. We had [indiscernible] as you saw some really nice cross-selling and enterprise customer engagement on the safe side. We thought [indiscernible] call about their quarter-to-quarter growth, meaning we're really seeing strong numbers there. I think what's really exciting is what you allude to, which is the future. And if we look at the future, it comes back to a few core principles, principles Fastly has always kept core. One, if you want to build a strong [ network ] and then do what we say we're going to do. And I think that seems like a low bar, but it is just amazing how often I will walk in the C-level executives who say to me, I make them so happy that the product and the [ incentive ] is going to do. It's crazy that that's a low bar, but I think in the security space, especially, there are a lot of people selling magic beans, where when you plant them, they don't turn into one of things they're going to be. There's a lot of hype. There's a lot of oversell. And when we've signed Signal Sciences and with Signal Sciences customers that's with them for a long time is how effective the product is. I mean, just to give you an example, remote web application firewalls are not in walking [indiscernible]. It means you're still having an alarm that you don't turn on. You just see the people walking out of your house, the bad people. And you start to the next phase, scroll through a set of log and they say, "Oh, it's going to be so tough for my house." The reason that is the case because most of these are incredibly ineffective. They have way too many false positives, and they don't have capacities. And those remote applications, like even with the most -- the majority of our competitors, web application firewalls are in with log [indiscernible]. They didn't log when the bad people come in. And the next day, you have to go through and sit through and adjust. I mean 95%-plus of the Signal Science's web application firewall is a locking [indiscernible]. Again, for those of you in the audience who are not in the security space, that may seem like, aren't most alarms always on, aren't most alarms connected to the police departments. Like no, they're not. And on the Internet, we're actually doing [indiscernible] plates where you have security for the checkpoints. There's sort of check box and the compliances, the organization is realizing this security had to work. They have to work at scale, and it's going to block the bad players. And that is absolutely critical. In your point, the bad players also leverage hopping from one place to another. And so I think collaboration -- we just announced a really important collaboration with Okta in terms of shared data of Okta's platforms because there are very few of us that are sort of security-oriented, DevOps friendly and secured companies. So I think we are in for an excellent revolution in how these products all out will go in the appliances, out will go the systems that are only there to be a PPI compliant checklist. In will come when we're a team is avalanched, a product that work, a product that are -- you can build upon to iterate and that are friendly for this microservices world that we are in. And it's -- I think we're going to look back on 2020 and part of 2021 as the catalyst for that.

It seems to me that there's 3 major arenas in this broader security element, which are, obviously, the web application firewall. But there's also bot net protection and then the third one being DDoS protection. Can you talk about what portion of those 3 segments you've got good traction in and where you see the growth coming from?

Yes, I mean, all 3 of those are huge areas of traction and investment for Fastly. So let me start with the bot side. That has been a sore area. Largely, we believe, as we do in all of our product lines and ecosystems that we've got an excellent partners and that would leverage our technology. We have a team in securities, we talked about it in our annual -- in or quarterly release, which does bot protection for [indiscernible] and other types of products, the largest [indiscernible] world. Very excited and that's something, looking at our compute capability, looking at our security ability, building on top of us because that allows us to get a global reach that we need in order to defend against the bad. Bot was an area both investment for Fastly, but also a strong partnership. On the web application firewall in the black markets, I mean, with the Signal Sciences acquisition branded us as the best [indiscernible] run from a capabilities perspective, so that is a huge growth area. And it's not only the strong growth area for Fastly. I think one of the areas that we'll continue to see us evolve in is this IT security market in privacy. And some of these other questions are all intermingling as well. And so that's an area we have a lot of insight and interesting opportunities for us down the road. Our [indiscernible] was brought to us to deliver the large workloads [indiscernible] their customers, how else can we hope that [indiscernible] directly, indirectly. And those are the big areas of [indiscernible].

And just going back to bot net for a second just to dive into that a little bit more. I don't think people realize just how big a problem the bot net problem is. There's a massive amount of fraud that is being implemented with these bot nets that then in turn take resources away from companies that are offering things like seats on airplanes or alternatively seats at a movie theater or a play or what have you. How big a problem is that? Can you give us some sense of the scale of it?

Yes. I mean, you hear upwards of 50% of [indiscernible] in some of these cases. And I can tell you that in our experience, sometimes it's more. And I think the interesting thing about bots, let me give you an example. You've got a limited-edition sneaker, and you want to get it out in the world. So from your perspective, every sneaker that's bought has the retail potential of taking a $400 sneaker and selling it on the open market for $20,000. But what you care about is getting those sneakers in the hands of your dedicated customers. But at the end of the day, there's so much arbitrage built in it that's [indiscernible] for years. If they conquer -- and counterfeit can be sold at the same moment as you buy it for 5x for what you bought it for. I mean, any time you see that arbitrage, you have this opportunity where the arbitrage -- where you will get -- you will invest in a partner -- in your offensive capabilities up to the point where it comes too offensive. And then you lose -- you lose all of your rents in the market. So this is hugely challenging. And in some cases, I've seen customers were 90, 95 [indiscernible] when they come live with the sale is traffic they don't want. Now [indiscernible] accelerate those retail customers in the small time frame where they had limited supply of to PS5s, and PS5s were a high commodity. And they knew that they couldn't change the retail price. So you could sort of look at it and say, "Well, they don't care who gets it," but they deeply care because this is a unique benefit they can provide to their customer. So buying demand, off customers and limited supply of assets is it probably [indiscernible] in the world. It's no different. The online experience gives you the ability to [indiscernible] in a way that we've never seen. And so I think the answer is, if issues probably count, I want to under score, most people don't know how big a moment is because you might think the people who are buying from you are legitimate users when they're not. So not only it created this iceberg effect, which is you know the bad users because you block them, but you also know that under this iceberg, it's a huge number of illegitimate ad clicks on ads or volume of a specific product. So yes, it's a huge problem. I think important great challenge of the Internet basis over the next 5 years.

So if we look at a world where, increasingly, applications are pushed out to the edge, and edge compute is highly distributed, and the applications are built on microservices that are continuously being integrated and updated, how do you think about the ramifications of the update of a service mesh on a distributed global basis and how hard that is? I noticed that Google brought CIO back in-house, but I would think that the CICD pipeline in microservices is where the agility and that value associated with the new coding approach really gets distributed. But when I think about potentially 1,000 different locations globally simultaneously being updated every 5 minutes with a microservice, it starts to become a daunting task to envision. Is this something that gives you an additional competitive advantage as that microservice-based mesh deployed across the global edge footprint becomes a daunting task for many people?

I think that's a good question. More people have been seeing this for 10 years and we've been solving that problem for us. We have thousands of servers in thousands of locations. No customers have to worry. That's the value of service. They don't have to think about where those servers are, how those update. That's our business. So we know how to do an SLA ultimately. And as long as -- so when a customer has a flash sale in Tokyo, they don't need to care. When that flash sale changes and now it's in Australia, they don't need to care. They don't need to care about when we bring up a new location, your capital [indiscernible] automatically. So the entire premise [indiscernible] under the pin in a high [indiscernible] place, which means we need to be held to an extremely high standard and -- from both ourselves and our customers because that's a lot of trust. And ultimately, what we're going to see is not only ourselves, but also the developer tools organizations. And that why we're so happy to see Microsoft [indiscernible]. There has to be the ones to help solve this [indiscernible] alone.

That's a great answer. We're running out of time, unfortunately, but I think you can see from this conversation that Fastly is a highly differentiated company with exceptionally good growth prospects. And given the recent decline at a much more attractive stock price, so Joshua, Adriel, I guess, we didn't get to talk about the financial side of it very much, but thanks for joining us. And the people in the audience, and there was a very large audience here, we really appreciate everybody joining and listening in.

Thanks, guys. Appreciate it. Take care.