Solving the Trillion-Dollar Cloud Paradox with Scott Sellers, Co-Founder & CEO of Azul

(Joel Beasley at 00:00:02) Hello, my friends. Today, we're talking to Scott, co-founder and CEO of Azul. And we discuss how there's currently a trillion dollars of market cap being weighed down by cloud costs, different tactics for mitigating cloud costs, and how Azul's extremely efficient Java runtime can help you reduce the number of instances your company needs to run an application. All of this right here, right now, on the Modern CTO podcast.

(Scott Sellers at 00:00:36) Here we go.

(Joel Beasley at 00:00:37) This is the Modern CTO podcast.

(Joel Beasley at 00:00:48) So tell me a little bit about your backstory. How did your career get started and end up where you're at today?

(Scott Sellers at 00:00:55) Well, as you mentioned in your background, I've been a technology fan since as long as I can remember, dating back to the days of the original RadioShack TRS-80. My school bought a couple of these when I was in second grade, I think, one of the first true personal computers, and I started messing around with those using what was called the BASIC programming language and really started to get interested in computers. That, of course, evolved into the various Apple computers, Apple I, Apple II. You mentioned your background in music. Early on, I was a piano player also, and one of my initial forays into marrying computers and audio in general was in the early days of Apple. You couldn't even generate a beep without programming that.

(Scott Sellers at 00:01:46) And so being able to basically figure out a way to play a basic Mozart song by literally loops and changing frequencies, literally coding every single note. That was kind of the way that it was back in the day. So I've just continued to progress. I went to Princeton and got my undergraduate education there with a bachelor of science in electrical engineering, which was a wonderful educational background. Princeton in general is a very foundational educational background, and so I was exposed to a variety of different things within the engineering disciplines and ultimately really seemed to gravitate towards a combination of hardware engineering as well as a lot of computer science.

(Scott Sellers at 00:02:37) And so my first job out of school was at Silicon Graphics or SGI. And this is in the early nineties, which at the time in Silicon Valley where I moved to, there was no greater company to work for. It was a who's who of technical talent, fantastic as a young engineer to really learn how to do things right. And we were building some really kick-ass hardware at the time. And if you remember back in the day, may date you a little bit.

(Scott Sellers at 00:03:09) But like, for example, the original Jurassic Park, which just wowed people, and Terminator and movies like that that people had never seen those types of special effects before. And that was all done using SGI hardware and graphics technology. And that was a very, very exciting place to work. So that began my career. I then left to join a smaller company that was acquired. We started doing various graphics boards for more consumer PCs. And then I started my first company in 1995. That was a company called 3dfx. I founded that with two other people.

(Scott Sellers at 00:03:44) And we were really part of the PC 3D revolution, what you take for granted today when you bring up, on your PC or Mac or whatever, and you can do all these amazing 3D games and other applications for 3D. That didn't exist in the mid-nineties. And so 3dfx was really a pioneer in terms of making very high-quality 3D graphics consumable at PC price points. And so we took 3dfx public within a couple of years, and then eventually we sold that to NVIDIA, who I think people would recognize still as far and away the leader in graphics these days. So that was an exciting time. And after selling 3dfx to NVIDIA, I was looking for kind of the next big thing.

(Scott Sellers at 00:04:33) And I was introduced to a couple of other people, Gil Tene and Shampall Lamari. And we really had this classic Silicon Valley meeting in a coffee shop kind of thing. And we picked the coffee shop that happened to have Wi-Fi connections. So again, 2000 era, not every coffee shop had wireless. And so we picked the one that did and met there and brainstormed there. And ultimately got very excited about the revolution that was coming in this language called Java. And we felt that changes in terms of how developers program computers and applications, these don't come around very often, and we felt that Java was headed down a path to really revolutionize how applications would be developed across really every spectrum, everything from embedded to mobile to desktops to servers. And what we narrowed in on was really the opportunity in terms of making a better Java platform for the server side of things. So big enterprises running big data centers, mission-critical, revenue-generating applications.

(Scott Sellers at 00:05:44) And we thought that was a very exciting market to go after. So we started Azul in 2002 to go after that market. The original Azul business model was based on very innovative products, designing our own custom microprocessors to run Java. And we raised a lot of venture capital money from some of the top, best-known VCs in Silicon Valley and had good success building those products. Really did a number of pioneering things and acquired a great number of customers.

(Scott Sellers at 00:06:18) What we found though is that being in the hardware business and certainly building your own microprocessor is a very, very challenging thing because you're always competing against truly the big boys. And ultimately, what we heard from our customers is, even though they felt our products were phenomenal and really were embracing them, that when we looked at where the cloud was heading, commodity computers, et cetera, that really we wanted to pivot our product offerings to pure software. And so we began that journey around 2010. And that transition from hardware-based products to software-based products has really catapulted Azul to where it is today. We're a very high-growth company.

(Scott Sellers at 00:06:58) We serve predominantly Global 5000 types of customers that basically are able to deliver a better Java platform for their business-critical applications.

(Joel Beasley at 00:07:10) That is super cool. So I do want to just highlight that the 3dfx is really cool to me because, I mean, I really enjoyed video games growing up. And also, when I was talking to your colleague, Alex, ahead of this interview, he mentioned to me that it was a really cool experience for him getting to work with you because Voodoo cards changed the way he was able to play games. And I didn't realize though that—

(Scott Sellers at 00:07:39) It was a great company. Yeah. I still get emails, couple a month probably, of people that are fans of 3dfx and still have their Voodoo boards and ask me for drivers for their Voodoo boards and stuff like that. So it was a lot of fun to be part of that revolution in graphics for the PC.

(Joel Beasley at 00:08:01) Yeah. I mean, different components of computers, particularly gaming computers, definitely get their cult followings for that community of building your own gear to game on. So that's really cool then. Undoubtedly. So I want to get into the cloud paradox because, as I mentioned earlier, I was reading over that article you wrote in Forbes on it. And so the software solution that Azul is now running as its flagship product is a Java runtime. Is that correct?

(Scott Sellers at 00:08:39) That's exactly right. Yep.

(Joel Beasley at 00:08:41) So tell me a little bit about the cloud paradox just in general so we can set that context for the audience. And then I want to hear about how your Java runtime plays into that after.

(Scott Sellers at 00:08:55) So the cloud paradox, in general, what that is all about is basically this balancing act between being able to embrace public cloud and all the advantages that come along with public cloud. And by public cloud, I'm talking about enterprises that use AWS from Amazon, use Azure from Microsoft, use the Google Cloud. And I think for most listeners, you probably have an appreciation for the advantages of cloud computing in general. You get much greater scalability, allowed for much more efficient and rapid deployment, being able to use cloud capacity as needed. In other words, when your applications need more compute, storage, or bandwidth, you just add more and it just is there because it's in the cloud. It's like an infinite amount of resource. So all those things are fantastic. And really the industry more and more is not only embracing cloud, but really thinking about cloud as their first type of deployment. So we've moved away from moving applications from traditional on-prem data centers or hosted data centers. And now developers and architects are really thinking about cloud first in terms of as they think about new applications and develop new applications.

(Scott Sellers at 00:10:20) So cloud is certainly a true foundational change in terms of how enterprise IT works. The paradox aspect of the cloud paradox concept is really, all the advantages of cloud are wonderful, but one of the challenges associated with the cloud is cost. And one of the sort of unknown secret, if you will, of cloud is that once enterprises start to embrace cloud more and more, more and more applications, more and more use of it, when you actually take a step back and look at the underlying costs that they're paying Amazon or Microsoft or Google, it actually is significantly more expensive in terms of, again, paying for the compute, the storage, the network, significantly more expensive than if they would have stayed on-prem or in a more traditional hosted ISP type of infrastructure. And that creates a challenge because at the end of the day, there are significant advantages from agility and flexibility and reliability and those types of things associated with the cloud. But how much is that worth paying for?

(Scott Sellers at 00:11:36) And if you're talking about 5 to 10%, maybe 15%, that's a reasonable cloud tax because of the added flexibility. But when you're talking more profoundly 2x, 3x, 4x, and we see and hear from our customers all the time that that is the cost of cloud for them, apples to apples comparing that infrastructure from an on-prem type of deployment to the cloud deployment, that is out of balance. Right? And so the cloud paradox talks to the fact that especially with so many customers, with so many enterprises now where their expenses directly impact, their cloud expenses directly impact the bottom line, that you can literally map the cloud costs to the market capitalization of a public company.

(Scott Sellers at 00:12:26) And many of the public companies now, for example, if they're SaaS companies, their net income, their earnings, the more that they spend on cloud, the less they're going to make on their earnings. And so the cloud paradox was originally coined from a blog post from a couple people at the venture capital firm, Andreessen Horowitz. And their analysis suggested that there's almost a trillion dollars of market cap that is being impacted by cost of cloud. And, whether it's a trillion or 500 billion or some other very large number, the precise number is not relevant. The point of it being that there is literally a significant anchor that is weighing down market capitalization of these public software firms.

(Scott Sellers at 00:13:24) Conceptually, it's simple. It all comes down to cloud cost, but what do you do about that? And so part of our, the tools and the capabilities that we offer our customers with our products and expertise is helping them manage their cloud costs. And there is no one simple answer. It's multifaceted, and there's a lot of different things that enterprises need to do to keep their cloud costs in line.

(Scott Sellers at 00:13:51) However, our products and our expertise are one aspect that can be quite compelling in terms of reining in this spiraling cloud cost problem.

(Joel Beasley at 00:14:00) So I saw on your article you broke it down into four stages of cloud adoption. Laggards and early adopters being two of those stages that I think most people are familiar with. But then the other is being cloud paradox victims and value-optimized leaders. And the cloud paradox victims is a category that's rapidly growing, but why is it still rapidly growing if this is a well-known problem? Like, why aren't people taking steps to avoid falling into the trap?

(Scott Sellers at 00:14:35) Oh, I think they absolutely are taking steps. But the reality is the problems are challenging. And that's why you get into that phase of the overall cloud journey. It's a complicated problem because, again, there's so many benefits of being in cloud that oftentimes what we see is enterprises become a victim of their own success. And before they know it, they are dealing with an infrastructure that is not particularly manageable or visible. They're dealing with runaway instances that they didn't even know actually were running. But by the end of the week or the month when they get their Amazon bill, now they see it. And so a lot of these problems are unique to the cloud and new to the cloud. And it's not surprising that so many enterprises fall into it because they are new challenges and new processes and new techniques and new solutions have to be brought to bear in terms of being able to optimize their cloud cost infrastructure. So there's a lot of different things.

(Scott Sellers at 00:15:43) And if you think about the major tenets of where cloud cost comes from, compute, storage, networking, and overall being able to monitor and manage all those different resources is really critical. And there's different techniques that enterprises use in terms of reducing their costs associated with each of those major categories. Azul specifically addresses the compute side of things. And we can get into that in a minute. But being able to reduce the number of instances that a given enterprise has to deploy to be able to service a given application, given its volume needs, its number of user needs, its SLA, which means how quick does it need to respond or how much bandwidth or how much traffic does it need to be able to process. All of those things impact the amount of instances that need to be purchased from a cloud provider, and you pay for every one of those. Every single instance that you use, you pay for, whether that's on an annual basis if you buy ahead and you get some benefits of buying ahead, or many enterprises end up just paying on demand. And every time they use an instance, they're going to be charged for that. And so anything that you can do to reduce the number of instances needed has a direct correlation to a cost that an enterprise will pay to the cloud provider.

(Joel Beasley at 00:17:21) We've actually done a couple interviews recently with a company called Spot. They're owned by NetApp, and they actually help you identify Spot instances is what they call them, that are just lower cost where less usage is happening there. So they're able to purchase the instances at a much lower rate than usual with the on-demand purchasing. But from what they show, save a ton of money by doing that, which is really cool. But what are some of the other techniques?

(Scott Sellers at 00:17:57) Yeah. That's why I describe that being able to really have an optimized cloud cost is a multifaceted problem. Right? As you mentioned, for example, just in that realm of compute, that needs to be optimized in a variety of different ways. For example, you can buy big instances or you can buy small instances.

(Scott Sellers at 00:18:18) You can buy reserved instances or you can buy spot instances. And all of those are going to have different costs ramifications to them. And so, you know, various companies look at those different scenarios. And it all comes down to, you know, what is the use case for a given application? And this is again why it becomes very complicated very quickly because a given enterprise might have a hundred different applications.

(Scott Sellers at 00:18:48) It could have a thousand different applications. And each of those different applications is going to have different characteristics in terms of, well, what's the right approach for this application? It could be that that application is largely dormant, meaning it's not doing a whole lot on a regular basis. But then, you know, maybe it's the end of a trading day or maybe it's during an end of Friday sale, then you get this huge spike in demand. Right?

(Scott Sellers at 00:19:14) And something like that, a spot instance strategy can be fantastic because you don't want to pay all the time for something that you only use very infrequently. Alternatively, there's other applications, for example, like a website. Right? A website is constantly on. And, you know, even though website traffic varies, it is fairly predictable in terms of how much compute storage bandwidth that you need to run a given corporate website.

(Scott Sellers at 00:19:44) And in that case, what you really want is buying ahead. I'm willing to pay Amazon, you know, a full year or maybe a full three years worth of compute capacity because I have very good predictability in terms of how much I need. And that's always going to be the cheapest. If you're willing to commit for, you know, a certain number of instances for a longer period of time, that's always going to be the most cost effective way to go. So, and again, if you have to make this decision across hundreds of thousands of different applications, you can gain appreciation for why this problem actually is quite complicated.

(Scott Sellers at 00:20:20) So there's techniques like that. In addition to that, there's techniques like just, you know, making your application more efficient so that you don't even need as many compute instances period regardless of which compute instance. Like, how do I just instead of a thousand compute instances, how do I get that to 500? Right? Just the most basic math.

(Scott Sellers at 00:20:42) How do you do that? And there are software solutions like what Azul provides that allows you to, you know, literally just cut in half the number of compute instances needed because you're running a more efficient software platform.

(Joel Beasley at 00:20:57) So, yeah, tell me a little bit about what Azul does and how that plays into this.

(Scott Sellers at 00:21:03) Yeah. So Azul provides Java platforms for enterprise customers. And the Java programming language, which was invented back in the mid-nineties by Sun, now, of course, owned by Oracle, Java represents the grand majority and most popular programming language for enterprise applications. And while there are other choices other than Java, Java remains the number one in terms of, you know, mission critical, business critical applications that need high throughput, high stability, high security, you know, all those types of things. Java is and remains far and away the language of choice for that. The way that a Java program works is that a developer and team of developers at an enterprise will create an application, write it in Java, or write it in another language that uses Java. And there is a piece of software that is installed on a server called a Java Virtual Machine or a JVM is the parlance or JDK. They all kind of mean the same thing. And at the end of the day, that JVM software is installed and is responsible for running the application.

(Scott Sellers at 00:22:22) You can think about it a little bit like an operating system. Right? A Windows or Linux operating system that is also required to be installed on the server to be able to run an application. Well, the JVM is another layer of software that sits on top of the operating system and is responsible for running a Java-based application. And so anything that is developed in Java requires a JVM to run.

(Scott Sellers at 00:22:48) It's one of the reasons that there are 50 billion instances of Java running in any given point in time. It's a massively deployed, very, very popular, truly everywhere. It's almost impossible to look around without being able to, you know, see an example of Java being used, you know, really everywhere. And so, as a result of that, the efficiency of the JVM running the Java application is, you know, fundamental to how much server infrastructure you need, how fast an application will be, how many users that application can support, how much data and throughput can that application support. So in a very simple manner, the more efficient your JVM, the more efficient your application.

(Scott Sellers at 00:23:38) The more efficient your JVM, the less number of computer servers, cloud instances that you need. And so all JVMs are not created equal, you know. And so what we have in the form of our flagship product, which is called Azul Platform Prime, is in essence, a better JVM, faster, more performant, more scalable, more deterministic that allows for applications to run faster. So a Java application running on something like the Oracle JVM will not run as well in terms of throughput, speed, consistency of performance, those types of things, will not run as well compared to when that same application runs on the Azul Platform Prime JVM. And the benefit really can be taken advantage of in multiple different ways.

(Scott Sellers at 00:24:28) It can be better customer experience. It can be, like, say, for example, in the case of a trading application, it can mean that you're trading faster. You're actually, you know, being able to be more competitive relative to your competition. So you can use speed simply for benefits of speed, which I think are fairly obvious. But you can also use the speed and performance benefits of Platform Prime to actually reduce the number of compute instances that are needed in the cloud.

(Scott Sellers at 00:25:02) And, you know, we find that our customers oftentimes can instead of, again, meaning, say, for example, a hundred compute instances, they're paying Amazon for every single one of those. When they simply instead of using, for example, something like the Oracle JDK, when they move to Platform Prime, instead of those hundred compute instances needed, maybe they only need 80 or 70 or in some cases, 50. So the fact that compute instances in the cloud can so quickly be changed, whether that's adding or reducing, you know, we find customers who deploy Platform Prime, literally, the next day could start saving on their Amazon or Microsoft or Google bill. So it's quite profound.

(Joel Beasley at 00:25:46) That's crazy. That's got to be an easy sell. Just immediate money. So do you have a...

(Scott Sellers at 00:25:53) Yeah.

(Joel Beasley at 00:25:54) Do you have a number of, like, an the efficiency jump that you get when going from, like, the Oracle to the Platform Prime, or does that vary quite a bit depending on the application?

(Scott Sellers at 00:26:06) It does vary on the application, just to be honest. And, you know, there are cases for some reason that maybe the application actually doesn't get a lot of benefit because at the end of the day, you know, it could be limited by other reasons. Could be limited by how quickly storage can be accessed or could be limited by network latency or these types of things. So not every application is going to be limited by how fast the Java app itself is running. But most applications are.

(Scott Sellers at 00:26:39) And, you know, we see it does vary again by application. But if we're able to reduce—if we're able to improve the speed, for example, of an application by two X. Right? So instead of being able to handle 10,000 requests a second, we can handle 20,000 requests a second when running on the Platform Prime. You can use that benefit of performance to cut your infrastructure in half.

(Scott Sellers at 00:27:09) Right? So, you know, using performance advantage to, in essence, do more with less, you know, is sort of the obvious benefit of what you can do with a performance benefit in the cloud. And so, as I mentioned before, our customers typically see somewhere between 20 to 50 percent reduction in the number of compute instances needed in the cloud when they simply change from the Oracle JDK to the Azul Platform Prime JDK across a variety of different applications.

(Joel Beasley at 00:27:47) That is crazy and very significant. How did you guys get better at writing a JDK than Oracle themselves?

(Scott Sellers at 00:27:57) Well, you know, I think like most innovative companies, we have an advantage of being singularly focused. And Oracle's an amazing company. I would love to grow Azul to be anywhere near the size of Oracle. I mean, they're fantastic and, you know, obviously have done phenomenally well. But again, the challenge of companies as they get bigger and bigger is focus.

(Scott Sellers at 00:28:26) And Oracle has a tremendous number of products. They've acquired countless companies, and so they have so much complexity to their business. It's obviously been very beneficial to them. They're, you know, one of the best of the best. However, the bigger you get, it does become harder and harder to be really good at, you know, a certain number of things.

(Scott Sellers at 00:28:48) And so I'd say overall, the reason that we have, you know, such incredible differentiation relative to not just Oracle, but anyone, is we are a hundred percent focused on the Java platform. We are true experts in it. We hire the best of the best. We have for years and years. And I think that, well, not you could even call it maniacal focus. That maniacal focus on doing one thing exceedingly well, obviously, has benefit. And, you know, Oracle has fantastic engineers. IBM, Red Hat, you know, they also provide a JVM. They have fantastic engineers. But those companies are just they're just so big and they have so many other things that they also needed to, you know, to develop and spend their engineering dollars on and, you know, take customer calls from.

(Scott Sellers at 00:29:39) Inevitably, the bigger you get, the harder it is to be really good at a few number of things. And so, you know, that is the advantage of being a smaller company is that singular focus. And it's allowed us to hire, you know, truly, I would argue the best JVM engineers in the world. And certainly, we've been able to produce the world's best JVM as a result.

(Joel Beasley at 00:30:00) That's amazing. So for some context, how big is Azul right now, number of employees?

(Scott Sellers at 00:30:07) We're about 300 employees worldwide.

(Joel Beasley at 00:30:10) Cool. And so I know earlier you mentioned that you're growing really fast. Do you have, like, a hiring goal for the year, like, of engineers and employees in general?

(Scott Sellers at 00:30:24) Yeah. Generally speaking, from an employee growth rate, we're growing at about somewhere between 25 to 30 percent per year. You know, I believe that we will continue to do that for the foreseeable future. Maybe even actually, you know, exceed that, just given how successful our business is.

(Joel Beasley at 00:30:45) That's awesome. What are some ways that you're attracting the best of the best in, I mean, today's super competitive hiring market?

(Scott Sellers at 00:30:55) Well, I think at the end of the day, people want to have passion about what they're doing and they want to work with great people. It really all comes down to that. You know, people, of course, make decisions based on compensation and, you know, the hope for upside and, you know, those types of things. But when you really dig in, and of course, we're very competitive on those other aspects as well. But people want to wake up in the morning and they want to have passion about what they're doing.

(Scott Sellers at 00:31:22) And, you know, one of the really interesting aspects about what we do is it's incredibly sophisticated software. It's very, very challenging. It really, I think, speaks to, you know, a very technical minded person who loves technology, loves challenges, loves to compete, loves to build, you know, the world's best products. And so as a result of that, we have very unique jobs that are just hard to find if, you know, you are really into programming and really into, you know, performance-oriented types of workloads and applications and, you know, that sort of thing. We're very unique in that regard.

(Scott Sellers at 00:32:05) I think the other advantage that we've had is that we've always, you know, been very, very distributed in terms of our overall workforce. So COVID for us was really just a blip because we always were very, very distributed. So, you know, we are based in Silicon Valley. We have a fantastic engineering team here, but we also have various engineers throughout the world. And that flexibility in terms of truly hiring the best of the best wherever they are has really allowed us to build a tremendous team.

(Scott Sellers at 00:32:39) And that's our philosophy moving forward as well is, you know, don't be overly focused on only hiring people in Silicon Valley or what have you. You know, we find the best, we hire the best, and we do the best we can to retain the best.

(Joel Beasley at 00:32:54) Yeah. I was just talking to a tech company in New Zealand called Xero that was talking about how the pandemic was actually really beneficial to them in terms of hiring because a lot of New Zealanders in particular were realizing that they can come back and live at home in New Zealand and work for any company that they want, but that as a result, resulted in this specific company getting a lot of market share back with their hiring because it's just easier to work within time zones. But, yeah, that was just feedback I've been getting from a lot of people.

(Scott Sellers at 00:33:37) I think that's very consistent. I think the new normal has been especially beneficial to those outside of the, you know, kind of typical metro areas that, you know, that's usually where the best jobs are going to be. And, you know, for those that are not living in metro areas now realizing that they can get those same jobs working predominantly remotely, it's been a boom. And and so, you know, it's been you could think about it as a leveling effect where you don't have to be in Silicon Valley or New York or Boston or, you know, pick your other favorite tech hotspot. You don't have to be there to be able to still, you know, be able to work for the best companies, command fantastic salaries, and ultimately do, you know, really interesting work.

(Joel Beasley at 00:34:24) Yeah. I'm just waiting for Starlink to get good enough on moving vehicles to be able to really get solid connection in a bus or something, then we'll really be leveled.

(Scott Sellers at 00:34:40) That's right. Well, you marry that with autonomous driving and you can be coding while you're driving too. So that's coming. That's the future.

(Joel Beasley at 00:34:49) Yeah. That would be quite the dream. That would be so fun. But we covered pretty well what, like, Azul actually does. But I want to—I'm curious as to, like, what are the hard problems you're actively solving today? What are some of the stuff that your engineers are working on?

(Scott Sellers at 00:35:07) Well, I think the next major phase that, you know, we're really looking at and beginning to introduce some products on is thinking holistically about the Java programming language and about the Java platform and, you know, really thinking about this in the context of a cloud world. And as I mentioned before, Java was invented in the mid-nineties, and it was really invented in a world that is quite different than the way it is today. That's an obvious statement to say that. At the time, the cloud didn't exist. Java, one of its hallmarks has been, you know, the fact that you can write it once and run it anywhere.

(Scott Sellers at 00:35:47) And that's why JVM exists. If you have a Java application, so long as the JVM is installed on a server or desktop, a phone, a car—so long as the JVM exists, it can run that Java application. And so that was one of the true pioneering aspects of Java when it was developed. But Java was never envisioned to be thought of in a cloud context. And so as, obviously, the world has now moved to cloud, we're doing a lot of thinking and planning about what does Java look like in a cloud world and what does a cloud native JVM need to do? So when an application right from the beginning is cloud first—is only designed to be run in the cloud, has always been architected to be deployed in the cloud—if you remove the shackles of needing to run on-prem data center or another non-cloud environment and you're purely cloud first, what should the Java platform look like in that kind of context? Because that certainly, we believe, represents the majority of new application development moving forward—a cloud first architecture. And turns out there's a lot of different things that you can do within the JVM to make for a yet more efficient deployment capability, knowing that the application has been written with the cloud in mind. And so we are continuing to add capabilities and features and new elements of our product portfolio that are very focused on a cloud native JVM. For example, we recently introduced a part of our platform Prime product that is called the Cloud Native Compiler.

(Scott Sellers at 00:37:44) And what the Cloud Native Compiler is all about is using the infinite resources of the cloud in terms of being able to deliver a much more optimized runtime for the application. And it's an example that without the cloud, we wouldn't be able to offer this sort of level of capability. But with the cloud and the fact that it's part of a cloud native approach, you can now deliver yet much more optimizations. You can be much smarter about—oh, this application was run yesterday and learn from what happened yesterday. Today, when you deploy that application, you should have knowledge through artificial intelligence and other aspects in terms of not forgetting what happened yesterday, but absolutely use all the information that was available during yesterday's run today and make smarter decisions in terms of optimizations and overall improved efficiencies.

(Scott Sellers at 00:38:49) And so the Cloud Native Compiler is our first product element that incorporates a cloud native JVM vision. And that's a major theme for us moving forward in general—really thinking about Java deployments in the cloud and what they really mean and what it really means to be cloud native.

(Joel Beasley at 00:39:11) So is that, following that logic, eventually going to mean an entirely new JVM that you build that is just for the cloud—cloud native?

(Scott Sellers at 00:39:22) Absolutely. And doing it in a way that's fully compatible. Right? So, again, just like when you think about cloud native, it truly means that you don't have to worry about all the old stuff and all the compatibility that would normally go along with needing to run on older traditional servers, on-prem servers, et cetera, et cetera. So absolutely, a product that requires the cloud to run and takes advantage of the cloud to run is really what a cloud native JVM is all about.

(Joel Beasley at 00:39:54) That makes a ton of sense. And I really look forward to following along that journey, because it sounds like you guys are building the back end of the future, which is just really, really cool.

(Scott Sellers at 00:40:09) Well, it's on the one hand, it's exciting. Right? Some might say it's not the sexiest thing, right? Because we're, like I mentioned before, we're kind of like an operating system.

(Scott Sellers at 00:40:20) Right?

(Joel Beasley at 00:40:21) Yeah.

(Scott Sellers at 00:40:21) Is an operating system sexy? Probably not sexy in this day and age. But at the end of the day, operating systems are the backbone of everything that's run on anything. Right? So we may not be in the sexiest space, but it certainly is one that is profoundly important to every individual out there because we all use dozens, hundreds of different applications that impact our lives every single day.

(Scott Sellers at 00:40:48) And so that's what I think gets us excited about the space that we're in—the overall market is really being the mortar, if you will, of the brick walls which form the foundation of the enterprise.

(Joel Beasley at 00:41:04) Yeah. I mean, you're getting your hands into every single industry that affects our society. That is really exciting. With—I mean, as it goes with a lot of tech, it's not something that the general public most of the time is super psyched on. But it can be thankless work. I mean, to go back to how you were talking, comparing to the operating system—yeah, everyone just kind of passively uses their OS, but when it stops working, then they're really mad. But they don't thank you when it works.

(Scott Sellers at 00:41:41) That's exactly right.

(Joel Beasley at 00:41:47) But so before we wrap up, is there anything that we didn't get a chance to touch on today that you want to make sure we get out to the world?

(Scott Sellers at 00:41:54) I think you did a great job of your questions and certainly touched on the topics that are very top of mind for us. Certainly for those that are listening, we encourage you to come by our website and take a look at azul.com and check out the product offerings. Java is not—what it is now is not what it used to be, and there's a lot of new and innovative things that are happening in the Java language and the Java platform. And we encourage your listeners to come and check us out and take us for a test drive.

(Joel Beasley at 00:42:31) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email: [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.