 Hi, I'm Natalie Erlich and welcome to the AWS startup showcase presented by theCUBE. We have an amazing lineup of great guests who will share their insights on the latest innovations and solutions in leveraging cloud scale data in AI, security and life sciences. And now we're joined by the co-founders and co-CEOs of theCUBE, Dave Vellante and John Furrier. Thank you gentlemen for joining me. Hey Natalie, how you doing? Hey John. Well, I'd love to get your insights here. Let's kick it off and what are you looking forward to? Dave, I think one of the things that we've been doing on theCUBE for 11 years is looking at the signal and in the marketplace, I wanted to focus on this because AI and it's cutting across all industries. So we're seeing that with cybersecurity and life sciences. It's the first time we've had a life sciences track in the showcase, which is amazing because it shows that growth of the cloud scale. So I'm super excited by that. And I think that's going to showcase some new business models. And of course the keynotes, Aligachi, who's the CEO of Databricks, pushing a billion dollars in revenue, clear validation that startups can go from zero to a billion dollars in revenue. So that should be really interesting. And of course the top venture capitalists coming in to talk about what the enterprise dynamics are all about. And what about you, Dave? You know, I thought it was an interesting mix and choice of startups. And you think about AI security and healthcare. And I've been thinking about that. Healthcare is the perfect industry that is ripe for disruption. If you think about healthcare, we all complain how expensive it is, it's not transparent. There's a lot of discussion about, can everybody have equal access? That certainly with COVID, the staff is burned out. There's a real divergence and diversity of the quality of healthcare. And it all results in patients not being happy. I mean, if you had to do an NPS score on the patients in healthcare, it would be pretty low John. So when I think about AI and security in the context of healthcare and cloud, I ask questions like, when are machines going to be able to make better diagnoses than doctors? And that's starting. I mean, it's really an assistance in a play today. But I think when you think about cheaper and more accurate image analysis, when you think about the overall patient experience and trust and personalized medicine, self-service, remote medicine that we've seen during the COVID pandemic, disease tracking, language translation. I mean, there's so many things where the cloud and data and then it can help. And then at the end of it, it's all about, okay, how do I authenticate? How do I deal with privacy and personal information and tamper resistance? And that's where the security play comes in. So it's a very interesting mix of startups that I'm really looking forward to hearing from you. You know, Natalie, one of the things we talked about some of these companies, Dave, we've talked to a lot of these companies. And to me, the business model innovations that are coming out of two factors, the pandemic is kind of coming to an end. So that accelerated and really showed who had the right stuff in my opinion. So you were either on the wrong side or right side of history when it comes to the pandemic. And as we look back, as we come out of it with clear growth in certain companies and certain companies that adopt and say cloud. And the other one is cloud scale. So the focus of these startup showcases is really to focus on how startups can align with the enterprise buyers and create the new kind of refactoring business models to go from, you know, a repivot or refactoring to more value. And the other thing that's interesting is that the business model isn't just for the good guys. If you look at say ransomware, for instance, the business model of hackers has gone completely amazing too. They're kicking butt in terms of revenue. They have their own, they're well funded machines on how to extort cash from companies. So there's a lot of security issues around the business model as well. So to me, the business model innovation with cloud scale tech with the pandemic forcing function, you're seeing a lot of new kind of decision-making in enterprises, you're seeing how enterprise buyers are changing their decision criteria. And frankly, they're existing suppliers. So if you're an old guard supplier, you're going to be potentially out because if you didn't deliver during the pandemic, this is the issue that everyone's talking about. And it's kind of not publicizing the press very much, but this is actually happening. Well, thank you both very much for joining me to kick off our AWS startup showcase. Now we're going to go to our very special guest, Ali Goldsy, and John Furrier will sit with him for a fireside chat. And Dave and I will see you on the other side. Okay, Ali, great to see you. Thanks for coming on our AWS startup showcase, our second edition, second batch season two, whatever you want to call it. It's our second version of this new series where we feature the hottest startups coming out of the AWS ecosystem. And you're one of them have been there, but you're not a startup anymore. You're pushing serious success on the revenue side and company congratulations and great to see you. Likewise, thank you so much. Good to see you again. You know, I remember the first time we chatted on theCUBE, you weren't really doing much software revenue. You were really talking about the new revolution in data and you were all in on cloud. And I will say that from day one, you were always adamant that it was cloud, cloud scale before anyone was really talking about it. And at that time it was on premises with Hadoop and those kinds of things. You saw that early, I remember that conversation. Boy, that bet paid out great. So congratulations. Thank you so much. So I got to ask you to jump right in. Enterprises are making decisions differently now and you are an example of that company that has gone from literally zero software sales to pushing a billion dollars as it's being reported. Certainly the success of Databricks has been written about. But what's not written about is the success of how you guys align with the changing criteria for the enterprise customer. Take us through that. And these companies here are aligning the same thing. And enterprises want to change. They want to be in the right side of history. What's the success formula? Yeah, I mean, basically what we always did was look a few years out, see how can we help these enterprises future proof what they're trying to achieve, right? They have 30 years of legacy software and baggage and they have compliance and regulations. How do we help them move to the future? So we try to identify those kind of secular trends that we think are going to, maybe you see them a little bit right now. Cloud was one of them, but it gets more and more and more. So we identified those and there were sort of three, four of those that we kind of latched onto. And then every year that passes, we're a little bit more right because it's a secular trend in the market. And then eventually it becomes a force that you can't kind of fight anymore. Yeah, and I just want to put a plug for your clubhouse talks with Andreessen Horowitz. You're always on clubhouse talking about, you know, the, I won't say the killer instinct, but being a CEO in a time where there's so much change going on, you're constantly under pressure. It's a lonely job at the top. I know that, but you've made some good calls. What was some of the key moments that you can point to where you were like, okay, the wave is coming in now. We better get on it. What was some of those key decisions? Cause a lot of these startups want to be in your position and a lot of buyers want to take advantage of the technology that's coming. They got to figure it out. What was some of those key inflection points for you? Yeah, so, you know, so if you're just listening to what everybody's saying, you're going to miss those trends. Okay, so then you're just going with the stream. So one, you mentioned it's cloud. Cloud was a thing at the time. We thought it's going to be the thing that takes over everything. Today it's actually multi-cloud. So multi-cloud is a thing. It's more and more people are thinking, wow, I'm paying a lot to the cloud vendors. Do I want to buy more from them or do I want to have some optionality? So that's one, two, open. They're worried about lock-in. You know, lock-in has happened for many, many decades. So they want open architectures, open source, open standards. So that's the second one that we've been on. The third one, which, you know, initially wasn't sort of super obvious was AI and machine learning. Now it's super obvious. Everybody's talking about it. But when we started, it was kind of called artificial intelligence referred to robotics. Machine learning wasn't the term that people really knew about. Today it's sort of, everybody's doing machine learning and AI. So betting on those future trends, those secular trends, as we call them, is super critical. Yeah, and one of the things that I want to get your thoughts on is this idea of replatforming versus refactoring. You see it a lot being talked about in some of these. What does that even mean? People are trying to figure that out. Replatforming, I get the cloud scale. But as you look at the cloud benefits, what do you say to customers out there in enterprises that are trying to use the benefits of the cloud, say data, for instance, that you're in the middle of, how could they be thinking about refactoring and how can they make a better selection on suppliers? I mean, how do you know, it used to be RFP, do you deliver these speeds and feeds and you get selected? Now I think there's a little bit different science and methodology behind it. What's your thoughts on this refactoring as a buyer? What do I got to do? Well, I mean, let's start with use of RFP and so on. Times have changed. Back in the day, you had to kind of sign up for something and then much later you're going to get it. So then you have to go through this arduous process. In the cloud would pay us to go model elasticity and so on. You can kind of try your way to it. You can try before you buy and you can use more and more and gradually, you don't need to go in all in and say, we commit to 50 million and six months later, find out that, wow, this stuff is just shelf work, doesn't work. So that's one thing that has changed that's beneficial. But the second thing is, don't just mimic what you had on-prem in the cloud. So that's what this refactoring is about. If you had a Hadoop data lake, now you're just going to have an S3 data lake. If you had an on-prem data warehouse, now you're just going to have a cloud data warehouse. You're just repeating what you did on-prem in the cloud, architected for the future. And for us, the most important thing that we say is that this lake house paradigm is a cloud native way of organizing your data that's different from how you would do things on-premises. So think through what's the right way of doing it in the cloud. Don't just try to copy paste what you had on-premises in the cloud. It's interesting. One of the things that we're observing and I'd love to get your reaction to this, Dave Vellante and I've been reporting on it is, two personas in the enterprise are changing their organization. One is, I call it IT Ops or there's an SRE role developing. And the data teams are being dismantled and being kind of sprinkled through into other teams is this notion of data pipelining, being part of workflows, not just the department. Are you seeing organizational shifts and how people are organizing their resources, their human resources to take advantage of, say that the data problems that are need to be solved with machine learning and whatnot on cloud scale? Yeah, absolutely. So you're right, SRE became a thing, lots of DevOps people. It was because when the cloud vendors launched their infrastructure as a service to stitch all these things together and get it all working, you needed a lot of DevOps people. But now things are maturing. So with vendors like Databricks and other multi-cloud vendors, you can actually get much higher level services where you don't need to necessarily have lots of lots of DevOps people that are themselves trying to stitch together lots of services to make this work. So that's one trend. But secondly, you're seeing more data teams being sort of completely ubiquitous in these organizations. Before it used to be you have one data team and then we'll have data and AI and we'll be done. It's a one and done. But that's not how it works. That's not how Google, Facebook, Twitter did it. They had data throughout the organization. Every BU was empowered. It's sales, it's marketing, it's finance, it's engineering. So how do you embed all those data teams and make them actually run fast? And there's this concept of a data mesh which is super important where you can actually decentralize and enable all these teams to focus on their domains and run super fast. And that's really enabled by this late house paradigm in the cloud that we're talking about where you're open, you're basing it on open standards. You have flexibility in the data types and how they're going to store their data. So you kind of provide a lot of that flexibility but at the same time, you have sort of centralized governance for it. So absolutely things are changing in the market. Well, Ali, you're just a professor. You, the masterclass right here is amazing. Thanks for sharing that insight. You're always got a lot of data. And that's why we have you on here. You're an amazing, great resource for the community. Ransomware is a huge problem. It's now the government's focus. You know, we're being attacked. And, you know, we don't know where it's coming from. There's business models around cyber that's expanding rapidly. There's real revenue behind it. There's a data problem. It's not just a security problem. So one of the themes in all these startup showcases is data is ubiquitous in the value propositions. One of them is ransomware. What's your thoughts on ransomware? Is it a data problem? Does cloud help? Some are saying that clouds got better security with ransomware than say on-premise. What's your vision of how you see this ransomware problem being addressed besides the government taking over? Yeah, that's a great question. Let me start by saying, you know, we're a data company, right? And if you say you're a data company, you might as well just said we're a privacy company. Right? It's like some of us say, well, what do you think about privacy? Do you guys even do privacy? We're a data company. So yeah, we're a privacy company as well. Like you can't talk about data without talking about privacy with every customer, with every enterprise. So that's obviously top of mind for us. I do think that in the cloud, security is much better because, you know, vendors like us, we're investing so much resources into security and making sure that we harden the infrastructure. And, you know, by actually having all of this infrastructure, we can monitor it and detect if something is, you know, an attack is happening and we can immediately sort of stop it. So that's different from when it's on-prem. You have kind of like these separated duties where the software render, which would have been us, doesn't really see what's happening in the data center. So, you know, this IT team that didn't develop the software is responsible for the security. So I think things are much better now. I think we're much better set up. But of course, things like cryptocurrencies and so on are making it easier for people to sort of hide, right? They're decentralized networks. So, you know, the attackers are getting more and more sophisticated as well. So this is definitely something that's super important. It's super top of mind. We're all investing heavily into security and privacy because, you know, that's going to be super critical going forward. Yeah, we got to move that red line, figure that out and get more intelligence. Decentralized trends not going away. It's going to be more of that less of the centralized. But centralized does come into play with data. It's a mix that's not mutually exclusive. And I'll get your thoughts on this architectural question with, you know, 5G and the edge coming. Amazon's got that outpost, trends into wavelength. You're seeing Mobile World Congress coming up in this month. The focus on processing data at the edge is a huge issue. And enterprises are now going to be commercial part of that. So architecture decisions are being made in enterprises right now. And this is a big issue. So you mentioned multi-cloud. So tools versus platforms. Now I'm an enterprise buyer and there's no more RFPs. I got all this new choices for startups and growing companies to choose from that are cloud native. I got all kinds of new challenges and opportunities. How do I build my architecture so I don't foreclose a future opportunity? Yeah, as I said, look, you're actually right. Cloud is becoming even more and more something that everybody's adopting. But at the same time, there is this thing that the edge is also more and more important. And the connectivity between those two and making sure that you can really do that efficiently. My ask from enterprises, and I think this is top of mind for all the enterprise architects is choose open because that way you can avoid locking yourself in. So that's one thing that's really, really important. In the past, you know, all these vendors that locked you in and then you try to move off of them, they were highly innovative back in the day. In the 80s, in the 90s, they were the best companies. You gave them all your data and it was fantastic. But then because you were locked in, they didn't need to innovate anymore and, you know, they focused on margins instead. And then over time, the innovation stopped and now you were kind of locked in. So I think openness is really important. I think preserving optionality with multi-cloud because we see the different clouds have different strength and weaknesses and it changes over time. Early on AWS was the only game. The Azure showed up with much better security, active directory and so on. Now Google with AI capabilities, which one's going to win? Which one's going to be better? Actually probably all three are going to be around. So having that optionality that you can pick between the three and then artificial intelligence. I think that's going to be the key to the future. You know, you asked about security earlier. That's how people, you know, detect zero-day attacks, right? You ask about the edge, same thing there. That's where the predictions are going to happen. So make sure that you invest in AI and artificial intelligence very early on because it's not something you can just bolt on later on and, you know, have a little data team somewhere then now you have AI and it's one and done. Great, great insight, Ollie. I got to ask you, the folks may or may not know but you're a professor at Berkeley as well, done a lot of great work that's where you kind of came out of when Databricks was formed and Berkeley basically was invented distributed computing back in the 80s. I remember I was breaking in when Unix was proprietary when software wasn't open. You actually had to deal it under the table to get code. Now it's all open. Isn't the internet now with distributed computing and how interconnects are happening? I mean, the internet didn't break during the pandemic which proves the benefit of the internet and that's a positive. But as you start seeing edge, it's essentially distributed computing. So I got to ask you from a computer science standpoint, what do you see as the key learnings or connect the dots for how this distributed model will work? Obviously hybrid cloud is clearly the operating model but if you take it to the next level of distributed computing, what are some of the key things that you look for in the next five years as this starts to be completely interoperable. Obviously software is going to drive a lot of it. What's your vision on that? Yeah, I mean, so Berkeley, you're right for the geeks. There was a now project 20, 30 years ago that basically is how we do things. There was a project on how you search indices very early on, we think to me, that became how Google and everybody else do search today. So Berkeley was super, super early, sometimes way too early. And that was actually the mistake was that they were so early that people said that stuff doesn't work. And then 20 years later, you reinvented. So I think 2009 Berkeley published this above the clouds saying the cloud is the future. At that time, most industry leaders said that's just, that doesn't work. Today, recently they published a research paper called Sky Computing. So Sky Computing is what you get above the clouds, right? So you have the cloud as the future. The next level after that is the sky. That's one on top of them. That's what multicloud is. So that's a lot of the research at Berkeley and the distributed systems labs is about this. And we're excited about that. Then we're one of the sky computing vendors out there, right? So I think you're going to see much more innovation happening at the sky level than at the compute level where you needed all those DevOps and SRE people to like, you know, build everything manually themselves. I can just see the memes now coming, Ollie. Sky Net, Star Trek, you got space too, by the way. Space is another frontier. You're seeing a lot of action going on because now the surface area of data with satellites is huge. So again, I know you guys are doing a lot of business with folks in that vertical where you're starting to see real time data acquisition coming from these satellites. What's your take on the whole space as the, not the final frontier, but certainly as a new congested and contested space area for data? Well, I mean, as a data vendor, we see a lot of, you know, alternative data sources coming in and people are using machine learning AI to eke out signal out of the, you know, massive amounts of imagery that's coming out of these satellites. So that's actually pretty common in FinTech, which is vertical for us and also sort of in the public sector. Lots and lots of satellites imagery data that's coming and these are massive volumes. I mean, it's like huge datasets and it's, you know, a super, super exciting what they can do. Like, you know, extracting signal from these satellite imageries and, you know, being able to handle that amount of data, it's a challenge for all the companies that we work with. So we're excited about that too. I mean, definitely that's a trend that's going to continue. Ollie, I'm super excited for you and thanks for coming on theCUBE here for our keynote. I got to ask you a final question. As you think about the future, obviously your company has achieved great success in a very short time. And again, you guys done the work. I've been following your company. As you know, we've been breaking that Databricks story for a long time. Been excited by it. But now what's changed? You got to start thinking about the next 20 mile stare when you look at, you know, the sky computing, you're thinking about these new architectures. As the CEO, your job is to one, not run out of money, which you don't have to worry about that anymore. So hiring and then you got to figure out that next 20 mile stare as a company. What's that going on in your mind? Take us through your mindset of what's next and what do you see out in that landscape? Yeah, so what I mentioned around sky computing optionality around multi-cloud, you're going to see a lot of capabilities around that, right? Like how do you get multi-cloud disaster recovery? How do you leverage the best of all the clouds while, you know, at the same time, not having to just pick one. So there's a lot of innovation there that, you know, we haven't announced yet, but you're going to see a lot of it over the next many years. Things that you can do when you have that optionality across the different clouds. And the second thing that's really exciting for us is bringing AI to the masses, democratizing data and AI. So how can you actually apply machine learning to machine learning? How can you automate machine learning? Today, machine learning is still quite complicated and it's pretty advanced. It's not going to be that way 10 years from now. It's going to be very simple. Everybody's going to have it at their fingertips. So how do we apply machine learning to machine learning? It's called auto ML, automatic, you know, machine learning. So that's an area and that's not something you can be done with, right? But the goal is to, you know, eventually be able to automate away the whole machine learning engineer, the machine learning data scientists all together. You know, it's really fun in talking with you is that, you know, for years, we've been talking about this inside the ropes inside the industry around the future. Now people are starting to get some visibility. The pandemics forced that. You're seeing the bad projects being exposed. It's like the tide pulled out and you see all the scabs and bad projects that were justified, old guard technologies. Well, if you get it right, you're on a good wave. And this is clearly what we're seeing and you guys are an example of that. So as enterprises realize this, that they're going to have to look double down on the right projects and probably trash the bad projects. New criteria. How should people be thinking about buying? Because again, we talked about the RFP before, I want to kind of circle back because this is something that people are trying to figure out. You're seeing, you know, organic, you come in freemium models. As cloud scale becomes the advantage and the lock in frankly seems to be the value proposition, the more value you provide, the more lock you get which sounds like that's the way it should be versus proprietary, you know, protocols. The protocol is value. How should enterprises organize their teams? Is it end to end workflows? Is it, and how should they evaluate the criteria for these technologies that they want to buy? Yeah, that's a great question. So, you know, it's very simple. Try to future proof your decision-making. Make sure that whatever you're doing is not locking you're in. So, whatever decision you're making, what if the world changes in five years? Make sure that if you're making a mistake now, that's not going to bite you, you know, in about five years later. So, how do you do that? Well, open source is great. If you're leveraging open source, you can try it out already. You don't even need to talk to any vendor. Your teams can already download it and try it out and get some value out of it. If you're in the cloud, these pay as you go models, you don't have to do a big RFP and commit big. You can try it, pay the vendor, pay as you go $10, $15. It doesn't need to be a million-dollar contract and slowly grow as you're providing value. And then make sure that you're not just locking yourself in to one cloud or, you know, one particular vendor as much as possible, preserve your optionality because then that's not a one-way door. If it turns out later you want to do something else, you can, you know, pick other things as well. You're not locked in. So, that's what I would say. Just keep that top of mind that you're not locking yourself in to a particular decision that you made today that you might regret in five years. Well, I really appreciate you coming on and sharing your insight with our community in theCUBE. And as always, great to see you. I really enjoy your clubhouse talks and I really appreciate how you give back to the community. And I want to thank you for coming on and taking the time with us today. Thanks John, always appreciate you talking to you. Okay, Ali Ghatzi, CEO of Databricks, a success story that proves the validation of cloud scale, open and creating value. Values the new lock-in. So, Natalie, back to you for continuing coverage. That was a terrific interview, John, but I'd love to get Dave's insights first. What were your takeaways, Dave? Well, if we have more time, I'll tell you how Databricks got to where they are today, but I'll say this. The most important thing to me that Ali said was he conveyed a very clear understanding of what data companies got right and are getting right. He talked about four things. There's not one data team. There's many data teams. He talked about data is decentralized and data has to have context and that context lives in the business. He said, look, think about it. The way that the data companies would get it right, they got data in teams and sales and marketing and finance and engineering. They all have their own data and data teams. And he referred to that as a data mesh. That's a term that Jamak Degani coined. And the warehouse or the data lake, it's merely a node in that global mesh. It's discoverable. He talked about federated governance. And Databricks, they're breaking the model of shoving everything into a single repository and trying to make that the so-called single version of the truth. Rather what they're doing, which is right on, is putting data in the hands of the business owners. And that's how true data companies do it. And the last thing you talked about was sky computing, which I loved. It's that future layer. He talked about multi-cloud a lot that abstracts the underlying complexity of the technical details of the cloud and creates additional value on top. I always say that the cloud players like Amazon have given the gift to the world of $100 billion a year they spend in CapEx. Thank you. Now we're going to innovate on top of it. Yeah, and I think the refactoring, too, Dave. How about you, John? Well, that was great insight, and I totally agree. The refactoring piece, too, was key. He brought that home. But to me, I think Databricks, that Olly, share there. And he's been open in sharing a lot of his insights in the community. But what he's not saying, because he's humble and polite, is they cracked the code on the enterprise, Dave. And to Dave's point, it's exactly reason why they did it. They saw an opportunity to make it easier. At that time, Hadoop was the rage, and they just made it easier. They were smart. They made good bets. They had a good formula, and they cracked the code with the enterprise. They brought it in, and they brought value. And see, that's the key to the cloud, as Dave pointed out. You get re-platform with the cloud, then you refactor. And I think he pointed out the multi-cloud, and that really kind of teases out the whole future and landscape, which is essentially distributed computing. And I think companies are starting to figure that out with hybrid and this on-premises, and now SuperEdge, I call it, with 5G coming. So it's just pretty incredible. Yeah, you know, Databricks IPO is coming, and people should know. I mean, what everybody, they created Spark, as you know, John, and everybody thought they were going to do was mimic Red Hat, and sell subscriptions and support. They didn't. They developed a managed service, and they embedded AI tools to simplify data science. So to your point, enterprises could buy instead of build. We know this. Enterprises will spend money to make things simpler. They don't have the resources. And so what they got right was really embedding that, making and building a managed service, not mimicking the kind of the Red Hat model, but actually creating a new value layer there. And that's a big part of their success. If I could just add one thing, Natalie, to that, Dave's saying is really right on. And as an enterprise buyer, if you go to the other side of the equation, it used to be that you had to be a known company, get PR, you fill out RFPs, you have meet all the speeds, if he's like going to the airport and get a swab test and get a COVID test and all kinds of mechanisms to like block you and filter you. Most of the biggest success stories that have created the most value for enterprises have been the companies that nobody's understood. And Andy Jassy's famous quote of, being misunderstood is actually a good thing. Databricks was very misunderstood at the beginning and no one kind of knew who they were, but they did it right. And so the enterprise buyers out there, don't be afraid to test the startups because you know, the next Databricks is out there. And I think that's where I see the psychology changing from the old IT buyers, Dave is like, okay, let's test this company. And there's plenty of ways to do that. He illuminated those premium, small pilots, you don't need to go on these big things. So I think that is going to be a shift in how companies going to evaluate startups. Yeah, I think you think about it this way. Why should the large banks and insurance companies big manufacturers and pharma companies, governments, why should they burn resources, managing containers and figuring out data science tools if they can just tap into solutions like Databricks, which is an AI platform in the cloud and let the experts manage all that stuff. Think about how much money and time that saves enterprises. Yeah, we've got 15 companies here. We're showcasing this batch and this season if you call it that episode or what we're going to call it. They're awesome, right? And the next 15 will be the same. And these companies could be the next billion dollar revenue generated because the cloud enables that, Dave. I think that's the exciting part. Well, thank you both so much for these insights. Really appreciate it. AWI startup showcase highlights the innovation that helps startups succeed. And no one knows that better than our very next guest, Jeff Barr. Welcome to the show. And I will send this interview now to Dave and John and see you just in a bit. Okay, hey, Jeff, great to see you. Thanks for coming on again. Great to be back. So this is a regular community segment with Jeff Barr who's a legend in the industry. Everyone knows your name. Everyone knows that. Congratulations on your recent blog post where I've been reading a ton of news. I want to get your update because 5G has been all over the news. Mobile World Congress is right around the corner. I know Bill Vass has a keynote out there, virtual keynote. There's a lot of Amazon discussion around the edge with wavelength specifically. This is the outpost piece. And I know there's news I want to get to but the top of mind is there's massive Amazon expansion. The cloud is going to the edge. It's here. What's up with wavelength? Take us through the, I call it the power edge, the super edge. Well, I'm really excited about this. Mostly because it gives a lot more choice and flexibility and options to our customers. This idea that with wavelength, we announced quite some time ago, at least quite some time ago if we think in cloud years, we announced that we would be working with 5G providers all over the world to basically put AWS in the telecom providers, data centers or telecom centers so that as their customers build apps, that those apps would take advantage of the low latency, the high bandwidth, the reliability of 5G, be able to get to some compute and storage services that are incredibly close geographically and latency wise to the compute and storage that is just going to give customers this new power and say, well, what are the cool things we can build? Do you see any correlation between wavelength and some of the early Amazon services? Because to me, my gut feels like there's so much headroom there. I mean, I was just riffing on the notion of low latency packets. I mean, just think about the applications gaming and VR and metaverse kind of cool stuff like that where I mean, having the edge be that, how much power there? It just feels like a new, it feels like a new AWS. So I mean, what's your take? You've seen the evolutions and the growth of a lot of the key services like EC2 and S3. So welcome to my life. And so to me, the way I always think about this is it's like when I go to a home improvement store and I wander through the aisles and I often wander through with no particular thing that I actually need, but I just go there and say, well, they've got this and they've got this and they've got this other interesting thing. And I just let my creativity run wild. And instead of trying to solve a problem, I'm saying, well, if I had these different parts, well, what could I actually build with them? And I really think that this breadth of different services and locations and options and communication technologies, I suspect a lot of our customers and customers to be are in the same mode where they're saying, I've got all this awesomeness at my fingertips. What might I be able to do with it? Here, it reminds me when Fry's was around in Palo Alto, that store's no longer here, but it used to be back on the day when it was good. It was, you go in and just kind of spend hours and then next year you built a computer. Like what that, I didn't come in here. I had to get some cables. No, I got a motherboard. I still remember Fry's. And before that, there was the weird stuff warehouse, it was another really cool place to hang out, if you remember that. I do, I do. I wonder if I can jump in, you guys talking about the edge. And Jeff, I wanted to ask you about something that is, I think people are starting to really understand and appreciate what you did with the entrepreneur acquisition, what you're doing with Nitro and Graviton and really driving costs down, driving performance up. I mean, there's like a compute renaissance. And I wonder if you could talk about the importance of that at the edge because it's got to be low power. It has to be low cost. You're going to be doing processing at the edge. What's your take on how that's evolving? Certainly, so you're totally right that we started working with and then ultimately acquired Annapurna Labs in Israel a couple of years ago. I've worked directly with those folks and it's really awesome to see what they've been able to do just really saying, let's look at all of these different aspects of building the cloud that were once effectively kind of somewhat software intensive and say, where does it make sense to actually design, build, fabricate, deploy custom silicon? So from putting up the system to doing all kinds of additional kinds of security checks to running local IO devices, running the NVME as fast as possible to supporting EBS. Each of those things has been a contributing factor to not just the power of the hardware itself, but what I'm seeing and have seen for the last probably two or three years at this point is the pace of innovation on instance types just continues to get faster and faster. And it's not just cranking out new instance types because we can. It's because our awesomely diverse base of customers keeps coming to us and saying, well, we're happy with what we have so far but here's this really interesting use case and we need a different ratio of memory to CPU or we need more cores based on the amount of memory or we need a lot of IO bandwidth. And having that Nitro as the base lets us really, I don't wanna say plug and play because I haven't actually built this myself but it seems like they can actually put the different elements together very, very quickly and then come up with new instance types that just our customers say, yep, that's exactly what I asked for from and be able to just do this entire range of from like micro and nano sized all the way up to incredibly large with incredible just to me like when we talk about terabytes of memory that are just like actual just RAM memory, it's like that's just an inconceivably large number by the standards of where I started out in my career. So it's all about putting this power in customer hands. You use the term plug and play but it does give you that Nitro gives you that optionality. And then the other thing that to me is really exciting is the way in which ISVs are writing to whatever's underneath. So you're making that transparent to the users and so I can choose as a customer the best price performance for my workload and that's just gonna grow that ISV portfolio. I think it's really important to be accurate and detailed and as thorough as possible as we launch each one of these new instance types with like what kind of processor is in there and what clock speed does it run at? What kind of, how much memory do we have? What are the just the ins and outs and is it Intel or ARM or AMD based? It's such an interesting to me contrast. I can still remember back in the very, very early days of EC2 back, going back almost 15 years at this point and effectively everybody said, well, not everybody. A few people looked and said, yep, we kind of get the value here. Some people said, this just sounds like a bunch of generic hardware, just kind of generic hardware in Iraq. And even back then it was something that we were very careful with to design and optimize for use cases. But this idea that is generic is so, so, so incredibly inaccurate that I think people are now getting this that it's okay, it's fine tune, not just for the cloud, but for very specific kinds of workloads and use cases in the cloud. And you guys have announced obviously the performance improvements on lambdas getting faster. You got the per billing, second billings on windows and SQL server on EC2. So I mean, obviously everyone kind of gets that. That's been your DNA. Keep making it faster, cheaper, but easier to use. But the other area I want to get your thoughts on because this is also more on the footprint side is the regions and local regions. So you got more region news. Take us through the update on the expansion on the footprint of AWS because a startup can come in and these 15 companies that are here, they're global with AWS, right? So this is a major benefit for customers around the world. And Ali from Databricks mentioned privacy. Everyone's a privacy company now. So huge issue. Take us through the news on the region. Sure. So the two most recent regions that we announced are in the UAE and in Israel. And we generally like to pre-announce these anywhere from six months to two years ahead of time because we do know that the customers want to start making longer-term plans to where they can start thinking about where they can do their computing, where they can store their data. I think at this point, we now have seven regions under construction. And again, it's all about customer price. Sometimes it's because they have very specific reasons where based on local laws, based on national laws, that they must compute and or store within a particular geographic area. Other times they say, well, a lot of our customers are in this part of the world. Why don't we pick a region that is as close to that part of the world as possible? And the one really important thing that I always like to remind our customers of in my audience is anything that you choose to put in a region stays in that region unless you very explicitly take an action that says I'd like to replicate it somewhere else. So if someone says I want to store data in the US or I want to store it in Frankfurt or I want to store it in Sao Paulo or I want to store it in Tokyo or Osaka, they get to make that very specific choice. We give them a lot of tools to help copy and replicate and do cross-region operations of various sorts. But at the heart, the customer gets to choose those locations. And in the early days, I think there was this weird sense that you would, you'd put things in the cloud that would just mysteriously just kind of propagate all over the world. That's never been true. And we're very, very clear on that. And just, I just always like to reinforce that point. It's great stuff. Jeff, great to have you on again as a regular update here, just for the folks watching. Don't know Jeff, he's been blogging and sharing. He's been the one man media band for Amazon at the early days. Now he's got departments. He's got people's on doing videos. It's a media franchise in and of itself. But without your updates, we wouldn't have gotten all the great news. We subscribe to it. We watch all the blog posts. Essentially the flow coming out of AWS, which is just a tsunami of a new announcements. Always great to read, must read. Jeff, thanks for coming on. Really appreciate it. It's great. Thank you, John. Great to catch up as always. Jeff Barr with AWS. Again, follow his stuff. He's got a great audience and community. They talk back. They collaborate and they're highly engaged. So check out Jeff's blog and his social presence. All right, Natalie, back to you for more coverage. Perfect. Well, did you guys know Jeff took a three week AWS road trip across 15 cities in America to meet with cloud computing enthusiasts? 5,500 miles. He drove really incredible. I didn't realize that. Let's unpack that interview though. What stood out to you, John? I think Jeff Barr is an example of what I call direct audience, a business model. He's been doing it from the beginning and I've been following his career. I remember back in the day when Amazon was starting, he was always building stuff. He's a builder. He's classic. And he's been there from the beginning. But at the beginning, he was just the blog and became a huge audience. It's now morphed into, he was power blogging so hard. He has now support and he still does it now. It's basically the conduit for information coming out of Amazon. It's, I think Jeff has single-handedly made Amazon so successful at the community developer level. And that's the startup action happened and that got them going. And I think he deserves a lot of the success for AWS. And Dave, how about you? What is your reaction? I think everybody knows about the cloud and CapEx, the OpEx and agility and eliminating the undifferentiated heavy lifting and all that stuff. But one of the things that's often overlooked which is why I'm excited to be part of this program is the innovation. And innovation comes from startups and startups start in the cloud. And so I think that's part of the flywheel effect. You just don't see a lot of startups these days saying, okay, I'm going to do something that's outside of the cloud. There are some, but for the most part, in software, you're starting in the cloud. It's so capital efficient. I think that's one thing. I've throughout my career, I've been obsessed with every part of the stack for whether it's the close to the business process with the applications. And right now I'm really obsessed with the plumbing, which is why I was excited to talk about the Antipurna acquisition. Amazon bought Antipurna for $350 million, it's reported, maybe a little bit more, but that is an amazing acquisition. And the reason why that's so important is because Amazon is continuing to drive costs down, drive performance up. And in my opinion, leaving a lot of the traditional players in their dust, especially when it comes to power and cooling, you know, often overlook things. And so, and the other piece of the interview was that Amazon is actually getting ISVs to write to these new platforms so that you don't have to worry about, does the software run on this chip or that chip or X86 or ARM or whatever it is, it's just, it runs. And so I can choose the best price performance. And that's where people don't, they misunderstand, you always say it, John, and you just said that people are misunderstood. I think they misunderstand, they confuse, you know, the price of the cloud with the cost of the cloud, they ignore all the labor costs that are associated with that. And so, you know, there's a lot of discussion now about the cloud tax. I just think the pace is accelerating. It's not, the gap is not closing, it's widening. If you look at the one question I asked him about wavelength and I had a follow up there when I said, you know, what do we riff on it? And you see, he lit up like he was beaming because he said something interesting. It's not that there's a problem to solve at this opportunity. And he conveyed it to like, I said welcome through fries, but like, you're going to a store, he's a builder. So he sees opportunity. And this comes back down to the Martin Casada paradox post he wrote about, do you optimize for CAPEX or future revenue? And I think the tell sign is that the wavelength edge piece is going to be so creative. And that's going to open up massive opportunities. I think that's the place to watch, that's the place I'm watching. And I think startups are going to come out of the woodwork because that's where the action will be. And that's just Amazon at the edge. I mean, that's just cloud at the edge. I think that is going to be very effective. And that's a little tell sign. He kind of revealed a little bit there, a lot there with that comment. Well, that's a two B continuum conversation. Indeed, I would love to introduce our next guest. We actually have Soma on the line. He's the managing director at Mitrona Venture Group. Thank you, Soma, very much for coming for our keynote program. Thank you, Natalie. You know, great to be here and great to have the opportunity to spend some time with you all. Well, you have a long-tenured history in the enterprise. How would you define the modern enterprise also known as cloud scale? Yeah, so I would say like, first of all, like, you know, we've all heard this now for the last, you know, say, 10 years or so, like, you know, hey, software is eating the world, okay? Put it in another way. We think about like, you know, hey, every enterprise is a software company first and foremost, okay? And companies that truly internalize that, that truly think about that and truly act that way are going to sort of continue running well and things that don't internalize that and don't do that are going to be left behind sooner than later, right? And the last few years, you sort of think and take it to the next level and talk about like every enterprise is now going through a digital transformation, okay? So when you sort of think about the world from that lens, okay? Modern enterprise has to think about like, hey, I'm first and foremost a technology company. I may be in the business of like, you know, making a car or, you know, manufacturing paper or like, you know, manufacturing some healthcare products or what have you kind of thing, but technology and software is what is going to give me a unique differentiated advantage that's gonna let me do what I need to do for my customers in the best possible way kind of thing, right? So that sort of level of focus, level of execution has to be there in a modern enterprise. The other thing is like, every modern enterprise needs to think about like, hey, I'm competing for talent not anymore with my peers in my industry. I'm competing for technology talent and software talent with the top five technology companies in the world whether it is Amazon or Facebook or Microsoft or Google or what have you kind of thing, right? So you really have to have that mindset and then everything flows from that. So I got to ask you on the enterprise side and you've seen many ways of innovation. You've got, you know, been in the industry for many, many years. The old way was enterprises want the best proven product and the startups want that lucrative contract, right? And get that beach and used to be and we addressed this in our earlier keynote with Ali and how it's changing, the buyers are changing because the cloud has enabled this new kind of execution and call it agile, call it what you want. Developers are driving modern applications. So enterprises are still trying, there's no, the playbook's evolving, right? So we see that with the pandemic. People had needs, urgent needs and they tried new stuff and it worked. You know, the parachute opened as they say. So how do you look at this as you look at stars you're investing in and you're coaching them? What's the playbook? What's the secret sauce of how to crack the enterprise code today? And if you're an enterprise buyer, what do I need to do? I want to be more agile. Is there a clear pass or a TSA to let stuff go through faster? I mean, what is the modern playbook for buying and being a supplier? That's a fantastic question, John, because like, and I think that world is that sort of playbook is changing even as we speak here, can I think, okay? Couple of key things to understand. First of all, is like, you know decision-making inside an enterprise is getting more and more decentralized, okay? Particularly decisions around what technology to use and what solutions to use to be able to do what people need to do. That decision-making is no longer sort of, you know all done at like the CIO's office or the CTO's office, can I think, right? Developers are more and more like you, right? You said like, you know, sort of the center of the workflow and the decision-making process, okay? So it'll be who both the enterprises as well as the startups to really understand that. So what does it mean now from a startup perspective? From a startup perspective, it means like, hey in addition to thinking about like, hey, you know, do I go create an enterprise sales post? Do I sell to the enterprise? Like what I might have done in the past is that the best way moving forward or should I be thinking about a product led growth go to market initiative, right? You know, build a product that is easy to use that where self-serve really works, you know get the developers to start using to see the value to fall in love with the product and then you think about like, you know hey, how do I go translate that into a contract with enterprise, right? And more and more what I call particularly, you know startups and technology companies that are focused on the developer audience are thinking about like, you know how do I have a bottom up good market motion? And sometime I may sort of, you know overlap that with the top down enterprise sales motion that we know there has been going on for many, many years or decades kind of thing but really this product led growth bottom up go to market motion is something that we are seeing on the rise. I would say like, you know hey, more than half the startups that we come across today have that in some way shape or form and so the enterprise also needs to understand this. The CIO or the CTO needs to know that like, you know hey, decision making is getting decentralized. I need to empower my engineers and my engineering managers and my engineering leaders to be able to make the right decision and trust them. I'm going to give them some God right so that I don't find myself in a soup, you know sometime down the road but once I give them the God raise I'm going to enable people to make the decision people are closer to the problem to make the right decision. So what are some of the ways that startups can accelerate their enterprise penetration? I think that's a great, that's another good question. First of all, you need to think about like, you know, hey, what are enterprises wanting today? Okay. If you sort of take like two steps back and think about like, you know, hey, what do enterprises really think about like, you know, hey, I'm a software company but I'm really manufacturing paper. What do I do, right? The core thing that most enterprises care about is like, you know, hey, how do I better engage with my customers? How do I better serve my customers? And how do I do it in the most optimal way? At the end of the day, that's what like most enterprises really care about, right? So startups need to understand, hey, what is the problem that the enterprise is trying to solve? What kind of tools and platform technologies and infrastructure support and you know, everything else that they need to be able to do what they need to do and what only they can do in the most optimal way, right? So to the extent you are providing either a tool or a platform or some technology that is going to enable your enterprise to make progress on what they want to do, you're going to get more traction within the enterprise. So start, you know, in other words, stop thinking about technology and start thinking about like, you know, the customer problem that you're going to solve and the more you anchor your company and more you anchor your conversation with the customer around that, the more the enterprise is going to get excited about wanting to work with you. So I got to ask you on the enterprise and developer equation because CSOs and CXOs depending who you talk to, have that same answer. Oh yeah, in the 90s and 2000s, we kind of did, we throttled down, we were using legacy developer tools and cloud came and then we had to rebuild them. We didn't really know what to do. So you're seeing a shift and this hasn't kind of been going on for at least the past five to eight years, a lot more developers being hired in. I mean, Fintech is clearly vertical, they always had developers and everyone had developers but there's a fast ramp up of developers now and the role of open sources change. Just looking at the participation, they're not just consuming open source, open source is part of the business model for mainstream enterprises. How is this, first of all, do you agree? And if so, how has this changed the course of an enterprise's human resource selection, how they organized, what's your vision on that? Yeah, so as I mentioned earlier, John, in my mind the first thing is, and this sort of financial services has always been sort of hiring people kind of thing. And this is like, no, a five year old story. So bear with me, I'll tell you the five year old story and then come to today kind of thing, right? I was trying to, the cloud CIO at Goldman Sachs. Okay, and this is five years ago when people were still like, hey, is this cloud thing real? Is cloud gonna take over the world? Am I really ready to put my data in the clouds? There are a lot of questions and conversation kind of thing. The CIO at Goldman Sachs told me two things that I remember to this day, okay? One is, hey, we've got an internal edict, okay? That we made a decision that in the next five years, everything in Goldman Sachs is gonna be on the public cloud. And I literally jumped out of the chair. I said, like, no, are you gonna get there? And then he laughed and said, like, no, it really doesn't matter whether we get there or not. We want to set the tone, set the direction for the organization that like, no, hey, public cloud is here. Public cloud is there and we need to move as fast as we realistically can and think about all the financial regulations and security and privacy and all these things that we care about deeply, right? But given all of that, the world is moving towards public load and we better be on the leading edge as opposed to the lagging edge. And the second thing he said, we're talking about like, hey, how are you hiring engineers at Goldman Sachs kind of thing? And he said, like, hey, my team goes out to the top 20 schools in the US, okay? And the people we really compete with are, and he was saying this, hey, we don't compete with JP Morgan or Morgan Stanley or pick any of your favorite financial institutions. We really think about like, hey, we want to get the best talent into Goldman Sachs out of these schools and we really compete head to head with Google. We compete head to head with Microsoft. We compete head to head with Facebook. And we know that the caliber of people that we want to get is no different than what these companies want. If you want to continue being a successful leading edge, financial services player, okay? That sort of tells you what's going on. You also talked a little bit about like, hey, open source is here to stay. What does that really mean kind of thing, right? In my mind like now, and you can tell me that like, hey, from given my pedigree at Microsoft, I can't tell you that we were the first embraces of open source in this world. So I'll say that right out of the bat, okay? But having said that like, we did turn around and said like, hey, this open source is real. This open source is going to be great. How can we embrace and how can we participate? And you fast forward to today, like now Microsoft is probably as good at open source as probably any other large company I would say, right? Including like another work that the company has done in terms of acquiring GitHub and letting it stay true to its original promise of open source and community kind of thing, right? I think Microsoft has come a long way kind of thing, okay? But the thing that like in all these enterprises need to think about is you want your developers to have access to the latest and greatest tools. To the latest and greatest that the software world can provide and you really don't want your engineers to be reinventing the wheel all the time, okay? So if there is something available in the open source world, go ahead, please sort of think about whether that makes sense for you to use it. And likewise, if you think there is something you can contribute to the open source world, go ahead and do that. So it's really a two way symbiotic relationship that enterprises need to have and they need to enable their developers to want to have that symbiotic relationship. So much fantastic insights. Thank you so much for joining our keynote program. Thank you, Natalie, and thank you, John. It was always fun to chat with you guys. Thank you. Thank you. John, would love to get your quick insight on that. Well, I think first of all, he's a prolific investor at a great firm, Madrona Venture Partners, which is well known in the tech circles. They're in Seattle, which is in the hub of I call Cloud City. You got Amazon and Microsoft there. He's been at Microsoft and he knows the developer ecosystem. And reason why I like his perspective is that he understands the value of having developers as a core competency in Microsoft. That's their DNA. I mean, you look at Microsoft, their number one thing from day one besides software was developers. That was their army, their thousand centurions that won everything for them. That has shifted. And he brought up open source and .NET and how they've embraced Linux. But Satya Natela, before he became CEO, we interviewed him on theCUBE at an Excel partners event at Stanford. He was open before he was CEO. He was talking about opening up. They opened up a lot of their open source infrastructure projects to the open compute foundation early. So they had already had that going. And at that price, since that time, the stock price of Microsoft has skyrocketed because as Ali said, open always wins. And I think that is what you see here. And as an investor now, he's picking in startups and investing in them. He's got to read the tea leaves. He's got to be on the right side of history. So he brings a great perspective because he sees the old way and he understands the new way. That is the key for success we've seen in the enterprise and with the startups. The people who get the future and can create the value are going to win. Yeah, really excellent point. And just really quickly, what do you think where some of our greatest hits on this hour of programming? Well, first of all, I'm really impressed that Ali took the time to come join us because I know he's super busy. I think they're at a $28 billion valuation now. They're pushing a billion dollars in revenue, gap revenue, and again, just a few short years ago, they had zero software revenue. So of these 15 companies we're showcasing today, there's a next Databricks in there. They're all going to be successful. They already are successful and they're all on this rocket ship trajectory. Ali is smart. He's also got the advantage of being part of that Berkeley community, which they're early on a lot of things. Now, being early means you're wrong a lot, but you're also right and you're right big. So Berkeley and Stanford are obviously big areas here in the Bay Area as research. He is smart, he's got a great team, and he's really open. So having him share his best practices, I thought that was a great highlight. Of course, Jeff Barr highlighting some of the insights that he brings, and obviously having a perspective of the VC. And we're going to have Peter Wagner from Wing VC, who's a classic enterprise investor, super smart. So he'll add some insight. Of course, we're going to have the community session. We're going to have our influencers coming on, Sar beat coming on at the end, as well as Katie Drucker, another Madrona person. Talk about growth hacking, growth strategies. But yeah, I'm psyched for Ali coming on. Terrific. Well, thank you so much for those insights. And thank you to everyone who is watching the first hour of our live coverage of the AWS startup showcase. For myself, Natalie Ehrlich, John Furrier and Dave Vellante. We want to thank you very much for watching and do stay tuned for more amazing content as well as a special live segment that John Furrier is going to be hosting. It takes place at 12 30 PM Pacific time. And it's called cracking the code lessons learned on how enterprise buyers evaluate new startups. Don't go anywhere.