 Welcome back everyone, it's theCUBE's coverage. We're here on location at AWS re-invent 2023. Here in Las Vegas, I'm John Furrier, your host with Dave Vellante. Dave, been a great 11 year run with AWS. This is our 11th season, 11th episode, 11th year event. And it's the biggest ever for theCUBE with more content pumping ever before. And we have a CUBE alumni, Doug Merritt, CUBE of AVIatrix. Great to see you, great to see you at the show. Well, thank you, great to be here. I cannot believe that you guys have been here really it's 12 years old, right? Amazing, 2013 was our first time, right? And we had a small setup kind of like this. It was like we're back to our roots, kind of like AWS, back to infrastructure. Yeah, my first was 2014 and just comparing, contrasting. Like that's probably 5,000 people. And I can't, there's 5,000 people in the hall. Right, you can't go anywhere, not have 5,000 people. Right, it was just one hall and then now it's just a crush of people. And it's like the old Comdex days, it's incredible. I told my friends, I feel like I'm going in and out of a Japanese subway car all day long. Not a word. Stress is high. Doug, it's great to have you on. And obviously for folks who don't know, but most know you were the CEO of Splunk, grew that company, took it public. And it's a data company by the way too, and all the right reasons. But we're on the Gen AI and you've seen, you've had a front row seat to that wave, okay? And you really made a ton of value creation out of that. Now we're in the mode now, we're with Gen AI. It's for everybody. And so we're at a whole nother level of innovation with the AI. Gen AI would serve the language models and foundation models, but the infrastructure underneath. And one interesting thing that came out of this trip was clearly that networking, what's going around the chips matters, what's around the software matters. So a whole nother level of needs, performance, energy, the constraints moving to power. You can put all the GBs you want on a rack, but if it's out of power. And so you got to kind of think differently around the system. For sure. This is the hot take here. What's your vision of how this happens, how this rolls out, how do you see it? So going back to selfishly the opportunity with the Navy for Aviatrix, what people seem to have forgotten about the cloud is networking is not invisible. It doesn't just happen by magic. You still have to physically move bits around. And without a data center and a compute center and applications having effective, guaranteed, high visibility, resilient network connection, then that never matters. So at that whole point of AI is to be able to do those quick inferences and come back with intelligent answers. So I love being at the foundational layers of what ultimately is necessary to power AI. What I was really impressed with with Adam's keynote and throughout this entire show so far is both the refocusing on those core elements, they went back to their core chip story and which was brilliant. But they really tied everyday use cases and how to practically bring AI to life through Bedrock and SageMaker and their entire stack. But how's it going to affect HR? How's it going to affect marketing? How's it going to affect finance? It's, and when I get excited about AI, there's certainly a lot we can do with networking to make networking more intelligent, more dynamic, more in the moment and effective. But within Aviatrix, that literally the first thing I did my first week was bring in two folks that were really good with AIOps as a technology stack to live this message that Adam was preaching of. There's about 30, 40, 60, 80% improvement for every individual almost and every department if you can start to wrap your mind around how am I going to use genera of AI to do my legal job better, to do my BDR and SDR job better, to do my engineering job better. But there's progression and you got to help teach people. And it's like the cloud had a big impact obviously on IT people. But now exactly what you're saying, it's everybody. It's like when the PC hit, we all got affected. I mean, we're going to see productivity exceed that most likely. Yeah, the web, the app store, virtualization, these are key inflection points. And what's interesting about the genera and what you said is that there's no discrimination against the value proposition every vertical. And Adam led the FinTech healthcare life science automotive clearly areas that will come together. We were just talking on the queue earlier that the dot-com bubble, everything was like over hype and then pop, but all that stuff happened. They went over a period of time. Now, we've been thinking about for the past decade, self-driving cars, that's kind of happening, the smart home kind of happening. Telco Edge, which you brought in later in his talk, which gets to your point. These things are now coming together because the compute's there, the amount of compute to render an image these days with AI wasn't available five months ago. So just in the past five to eight months, images, computer vision, major computation value is now out there. For sure. And compute ultimately is going to be one of the big constraints, I believe. And so what you see Nvidia doing for sure. I mean, they are still in the spotlight, but there's a couple of interesting companies out there that I've run across that are trying to do, to radically improve the cost structure, the power consumption structure, and the processing power. A lot of the models are dependent on compute if you really, really want to do the recursion and the inference on a more affordable basis. So, okay. Like an off-the-shelf kind of compute. Much more of a classic CPU and FPGA model so that you can actually, so that the cost structure comes down. And as you were saying, John, power consumption and like, where are you going to locate the compute centers and the data centers? But it does feel like mid 90s, maybe 93, 95, 96 when the internet was beginning to take off. And it is, you know, everything gets over-hyped and I'm sure it's almost too much on JNAI right now. But the promise over the next decade is, I get goosebumps. Yeah, yeah. Back to your networking comment, I want to get your reaction to the Jensen on stage with Adam Sileski. It was a big moment where they had stage. The whole side comment about what the posture was. It wasn't a little awkward. Well, we can go there now, but I want to make a point on this networking piece because I think this is the leading indicator. Amazon was the first cloud provider to have the GH 200 grace hoppers to the cloud. But now they're going to bring in 32 grace hoppers and connect them with an MV link, a one terabyte per second, process working together hand in hand. This is essentially telegraphing. And when I met with Adam for the one-on-one for an hour on the 17th, we talked a lot about what's around the chips. He was already telegraphing. That's not just about the GPUs. So you got the connections. They're basically optimizing the GPUs to kind of work together to get more out of the big bang, more bang for the buck. But this is going to be everywhere. And networking is one of those connective tissue things that matter for latency and velocity of packets. Yeah. So if we just go back to practicality, so I'm really excited about what AWS is doing and others on bringing Generve AI to the masses. And now let's go to the average company, a multi-billion dollar company in the Midwest that's a manufacturer or something. And how do they take advantage of that? There's all the education on how do I use the tooling, but they're a landscape of what they're operating today on their IT stack for both internal efficiency and for customer connectivity and regulator connectivity has a lot of ways to go. Like that's where I, what I'm so excited about with Aviatrix is we're finally at the point where the average corporation realizes we have a very complex landscape. We've still got proprietary data centers. It's hard to move those mainframes to the cloud, hard to move those AS400s to the cloud. We've got a whole series of edge compute centers, Equinex, Megaport, you name your favorite vendor. We've got usually more than one cloud and how do we get a network mesh that is reliable, transparent, optimized so that I eventually can take advantage of all this great Generve AI that I'm gonna manufacture so I can do better drug discovery or more efficient manufacturing or serve my customers better. And that just goes back to that foundational layer of you cannot connect all those pieces together and provide that seamless experience for your customers and your employees than you can realize the promise of all this investment that Adam and Jensen and everybody else are making. I go back to what you were saying about networking's not invisible. And it kind of felt, I mean it's trite to talk about the pendulum swinging but the pendulum is swinging back and the cloud is expanding. Networking was kind of invisible when it was in the cloud. I think about Infiniband, right? It was invisible inside of an exadata. And see it. And now as the pendulum swings back to the edge and the hybrid world, it's becoming more obvious and it becomes exposed to more and more people. And it's a key component of getting the value to your point out of those investments. It has $250,000 GPU investments that weigh 70 pounds. And part of the problem that CIOs and CISOs and the head of networking have realized is you lose some visibility and you lose some control with networking when you go to the cloud. The benefit is, hey, there's a bunch that the cloud is doing but there's still, if you want an optimal experience and you really want to get the speed and the flexibility that everyone's trying to drive by going to the cloud, there is augmentation necessary around those awesome networking services that AWS and other clouds have. And so what the Aviatrix is able to do is insert, take advantage of the networking services that the clouds are providing but then add a whole new set of dimensionality on the visibility, on traffic optimization, on cost reduction, on efficiency and operational consistency making it easier for a non-networking guru to actually operate these very complex environments and meet their customers goals. We sort of stopped talking about digital transformation. I mean, people still do talk about it but that was all the rage when Gen AI hit. But we're actually seeing it come into focus now because you've got all this public cloud infrastructure out there that's just, it's there to be utilized and you can build software on top of it. You can digitize your business and you've got to connect it. That's- Absolutely. I mean, the digital transformation that was so fascinating is the first wave, a lot like the waves for Amazon was lift and shift, get stuff in the cloud to understand it and then begin to modernize those apps and then from that learn what you can build from scratch. I think we had the same thing with digital transformation was what we'd add to a Splunk. Take a monolithic on-prem vertical scale out app and get it to the cloud and then modernize it so that it actually is serverless and goes horizontal. The next wave of that is now, how do you build something brand new? And what I'm seeing with digital transformation in our own AI app stack is that you, what Salesforce is great at is getting you to a transaction. I had a new opportunity. I closed an order. All the interesting stuff is what happened before the transaction and then what happens after the transaction and what Gen AI can help with with a host of other tools is provide that literally minute by minute context of what happened with your BDR teams, your SDR teams, your ISR teams, your SCs, your account managers, your CSMs around that account pre anything happening in post so you can be a much more effective of learning organization. Each one of those functions gets better but that's true digital transformation. Now you've got a record of what's happening in a film of what's happening every minute of every day instead of just snapshots every four weeks of something that committed to relational database. It's always on. I mean, this is the concept in real time. I want to get your thoughts on it because I want to just pivot off that and say, okay, the three layers of the stack that Adam laid out, okay, Gen AI stack, cool, classic stack, hardware, middleware app. Amazon hates why it's called a hardware. It's not hardware. It is. It is. They're just managed by them. And it's got services. It's a motherboard, okay. The data engineering conversation has come up a lot over the past year and the past six months and the past two months in particular at supercomputing and here, the data tsunami that's required to support AI is going to be massive. Now you have a developer tsunami happening where there's going to be a feeding frenzy on this foundation layer. Everyone's going to start playing around with it. So okay, now the data models have to be flipped upside down and reset. So you've got this platform engineering, data engineering, engineering particularly because it's got to be engineered going on. That means the infrastructure's up for grabs. It's not giving more boxes in the old days. Hey, I got developers, I need five more boxes to stack them on the rack. This is the computer there, check. But the data engineering becomes the focus. If you don't have the right data strategy, you're screwed. What's your reaction to that? Because that's going to affect the architecture of how networking works, how data pipelines work. How do you see that and how do you play into that? I think there's going to be a ton of iteration and playing with the data. What we've seen over the past, well, 40 years, but certainly past seven, is we've gone from a handful of data structure types, relational database, object database, index, if you go to the Splunk arena, to if you, last time I checked DB engines, I counted 29 different database categories. And each category would have five, 10, 15 people on it, and including an open source. So vector DBs, graph DBs, time series. And you need those because when you are in a cloud world and you've got billions potentially of users, you can't have a general purpose data store. I've gotten so many debates through the years with CIOs if I want one, it's like, but why do you want one? Like what are you trying to get done? So that we've got the reality of data being fragmented across these different data stores for very good reasons. Now how do you work across those different data stores to actually bring additional value with GNAI? And I think that will create new layers of a fraction and new data. The semantic layers need it. Make it coherent and yeah, indeed, that's it. But it always happens recursively and iteratively, right? We don't know today, I don't know today. Others much smarter than me probably do. Exactly what data is going to be important and where do you get it and how do you store it to get optimal usage for the unseen use cases that are coming with the GNAI wave. But I'm confident there'll be a bunch of new companies. A builder who has the problem will solve it and that's going to be a lightning strike. We just had the VCs from Adrona and said, it's going to take a Travis to Mrs. Cab in Paris to start Uber. I mean, this is the momentum we're starting to see. How is Hadoop born? It was born because there's no massive data storage structures within Google and Yahoo and to actually take advantage of that data and I think we'll see a lot of that. And the modern data pipeline is a better Hadoop and so there's something coming that is going to blow away what we see today. It has to. It has to, right? The volumes are, again, if we're going with that movie moment by moment that we haven't even begun to see. I was talking about X-Bite and Zetabyte scale at Splunk. Yeah, that may happen. I thought it was going to happen 20 years down the road. That might happen in the next few years. Doug, great to have you on. Give us the update on Aviatris. What's going on there? Put a plug in for the camera. We got a couple minutes left. Sure. What are you working on? So I was on, I think, even before my first day but maybe even before my first day. In the studio. Yeah, in July and this summer. What I've seen in the past four months is I thought coming in from the outside that we were in the right place at the right time. So much of business success is luck and timing. That you can't, if you're too far in front of a wave you lose out. If you're too far behind you lose out. And what I've seen and to end this week has really confirmed it is we have got incredible product market fit. What we can do feature and function wise no one else in the industry can do right now. And I think the market finally is, I've had probably 14 one-on-ones with customers now. And the story on everyone it's crazy that we're almost a script on what does their landscape look like? What are the problems for them to actually get value out of their journey to the cloud and how networking plays into that problem set. So my takeaway is we just have to figure out a way to get the name out with Aviatrix what we do and allow the market to come to us the way that I think it's, Steve and others saw it early to get us to this point. And the core problem you're going to solve for them is what? So the core problem is we're giving them a high simplicity visibility and resiliency on their networking. Cause again, cloud is a very different beast on networking than non-cloud. We're giving them a security woven into the network. I think the entire security landscape is changing at the networking level because you can't separate those two anymore. It's part of how networks operate. And we're giving them a true hybrid mesh. We're really bringing edge, all the different iterations of where compute and data might sit. We're making that a natural part of what happens within the cloud. So your Equinex pops, your last generation data centers feel and operate like they're part of AWS, like they're part of Azure, like they're part of GCP, OCI, Alibaba, the different clouds that we play in. So simplicity, security, and then integration and seamlessness across your complexity. So location agnostic. Yeah, location agnostic. Got it, okay. Doug, thanks for coming on theCUBE. Really appreciate it. Great seeing you and glad you can join our special SuperCloud 5 special edition battle for AI supremacy running out of Palo Alto. This is our on location conference. We'll see you at six. Yeah, I'm ready to be at six. We'll start going to Roman Newell, like the Super Bowl. It's like. It always enjoys seeing you guys and be on a show. Thank you. Well, we appreciate you. Okay, theCUBE, be right back with more coverage from Location. Back to Palo Alto Studio.