9. Hyperscalers, NeoClouds, Sovereign Clouds, and Service Providers with Jeetu Patel and Guru Shenoy

Understanding AI's Transformative Impact on Service Providers Hello everyone. Welcome to Cisco's Inside Product podcast. Today I have with me Guru Shenoy. Guru is our product leader for hyperscalers, Neo Clouds, Sovereign cloud service providers. It's a big job, my friend. Speaker 2 Thank you, Jitu. Yeah, pleasure to be here. Speaker 1 Yeah. And so we're going to talk a lot about what's happening in the industry with AI, with data centre build outs in these four kind of customer constituencies that you have specifically in data center networking. And so walk us through what's happening, set the stage for us. Why is it important for for what's happening specifically in service providers right now? Because that's actually a huge area of growth in addition to hyperscalers. And so what's happening over there and why do you think it's the it's seen a resurgence? Speaker 2 Yeah. Now look, we are undergoing a massive transformation as we know with AI and it's having major implications on traffic. And if you look at where the last three years or so have been, a lot of the energy around AI was around training, you know, mainly the hyperscalers building big data center networks. Now what we are starting. Speaker 1 Which we were involved in. Speaker 2 We've been heavily involved in that, helping those build outs and investing in technology and so forth. But now we are starting to see sort of the consumption of AI, if you will, right, inferencing the agentic economy arising. And so enterprises, consumers are all starting to use AI. And so all of that AI is now starting to get delivered. Most of it is over telco networks, service provider networks. And so we're starting to see massive shifts in traffic, not just volume, but changes in the patterns of traffic. And so service providers, they both have, they have a huge opportunity here because they're at the in the middle of this. They're the backbone of the AI ecosystem, as we like to say. And so delivering intelligent fabrics to serve these new AI workloads, participating more actively, offering potentially things like inferencing cloud. So there's huge opportunities for service providers and that's what's causing a resurgence because, you know, post 5G, this is the next big transformation and this is a durable one that we see. So there's a lot of energy amongst the. Speaker 1 Service provider community, This is a bubble which is going to go away at some point because. Speaker 2 Absolutely not. I'm 100% convinced that this is a durable trend because we are seeing value. We are seeing products and, yeah, you know, services that are actually adding a lot of value that was hard to see in the 5G era. 5G was all about, you know, hey, we need better latency. But no applications ever arose to fulfill that need, if you will. But here, it's a different ball game altogether. Speaker 1 The the demand had to follow the supply on that one was over here the demand is almost instantly yes you know serviced as a result of the supply and we are in the supply constrained world right now. Speaker 2 Big time, yeah. We're not able to keep up just in terms of the infrastructure that needs to be built out to deliver all of these new capabilities that the agentic world is is demanding. Characterizing the Exponential Growth of AI Network Traffic OK. So I want to, I want to talk about each one of these four markets real quickly, but let's start with the strategic shift you said is, hey, the, the, the amount of volume of traffic is going up exponentially. Give us some stats on the volumes of traffic that are going up. Speaker 2 Yeah. So this is, it's interesting because very recently, just a month ago, we published our AI impact on the van report, right? And it's the first of its kind. What we actually did is we didn't just look at third party sources and so forth, but we actually ran measurements with sensors in some of our service provider customers to identify what does AI traffic look like on the Internet today, right? How is it growing? What is the nature of traffic distribution and patterns? And this is a public report that we've shared and there's some really interesting stats. First of all, we're starting from a very low base because inferencing is only now starting to take off, but the growth rate is phenomenal. We are seeing 10X year over year inferencing growth rates. And obviously at some point it's going to kind of slow down a little bit, but it's going to go on for quite a. Speaker 1 While I think it slows down because you've got about .1% of the world that's using agents right now. Speaker 2 Yes. Speaker 1 For sure. Like you're only going to see that compound over time. It's not going to slow down. Speaker 2 You're right about that. It's the slowdown will happen many years and by that time, it's going to be a significant portion of the overall traffic. And we haven't even started seeing AI in a lot of video yet, right? Once that happens, we're going to see even more. But we are seeing this incredible growth rate in inferencing traffic 10X year over year. Moreover, we characterized how does this traffic behave, right? Because now we have agents that are actually driving and it's things like people using codecs or cloud code and all of these capabilities, right? That's one of the big contributors to the AI traffic. And we found that agents actually generate 450% as much traffic as humans do when they perform web transactions. And that's simply because they can do things so fast. They need a lot more context to reason and so forth. So we see this clear sign. Moreover, there's some interesting characteristics about the traffic distribution itself. There's a lot more upload that happens with AI traffic instead of just. Speaker 1 Because your markdown files, your skills file, your memory file have to keep getting uploaded. Speaker 2 Exactly. You need to send it to your inferencing model and then that needs to be processed as context and then you need it needs to respond to it. And then imagine as we start looking ahead, we are already starting to see signs of physical AI with a lot more robots and 3D spatial sensing and so forth. That's going to introduce a lot more video in the mix. So you'll have a lot more video upload as well to your closest inferencing location. A lot of it will happen on Prem, but some of it will go over a service providers metro network into a local regional inferencing cloud and so forth. So upload is a lot more traffic tends to be a lot more distributed as well in the sense that these agents and imagine a world where you have millions if not billions of agents, every enterprise application agentic enabled, every consumer application agentic enabled. All of these are talking to inferencing cloud endpoints. They don't just talk to 1 cloud location to download data as. Speaker 1 Very chatty goes across. Speaker 2 Multiple chatty. So the net effect of all of this is that the network over which all of this is serviced, service providers network needs to adapt significantly to handle that traffic load, to handle the multi directionality of traffic. A lot of it will be latency sensitive. We don't think latency is going to be the only factor like everybody thought in the 5G age. Speaker 1 Basically, latency, performance and power will be the three. Speaker 2 Right. Those are going to be the three primary ones, absolutely. Powering AI: Efficiency and Service Provider Monetization So Guru, one other question that comes up a lot is especially in data center build outs is around energy efficiency. Because if you think about it, we just don't have enough power, compute network bandwidth in the world associate the needs of AI. And that's, that's been a pretty commonly known fact at this point. We're doing a lot on making sure that these these infrastructure elements are built in a sustainable way and that we actually have very, very high efficiency on power consumption. Talk to us a little bit about that. What's happening? What are you doing and give us some stats. Speaker 2 Yeah. So power has been probably the number one constraint when we talk about data center build outs right now we're starting to see some others, but power continues to be one of the key ones. And so the desire to build extremely power efficient network infrastructure and compute infrastructure has been a very high priority for data center builders. So what we have done from our side is it begins from the silicon itself, right, building extremely power efficient silicon. And this is what we pride ourselves upon when we build Silicon 1, making sure that the silicon itself is superpower efficient, but then also all of the software that goes on top of the silicon and the systems, the nature of systems we build. We are because we are now building extremely high capacity silicon. We have the ability to deliver in small fixed form factor boxes, 123 RU that would earlier take big modular chassis to do right, 50 terabytes of routing and more. That's a huge difference because a chassis with all the fabric cards and line cards and everything. Speaker 1 The size matters on power consumption. Speaker 2 Big time and it's, you know, 8910 kilowatts for the chassis and we can deliver that in less than a couple kilowatts when we talk about fixed systems. So there's a lot of innovation there, but that's not where it stops. That's actually only the start because we have optics and optical that is being integrated more and more. Yeah, One of the big things that we've been pushing for a couple of years and now the entire industry has really started embracing it is something we call routed optical networking. And that is this idea that you bring together IP and optical layers together. So we have coherent plugables sitting in the routers, and that means you're eliminating an entire layer of hardware the. Speaker 1 Coherent plugables are the ones that go data center interconnect, and so you have these rather than having the big transporters, you can just actually have a bunch of plugables. Speaker 2 We can drive distances of hundreds and thousands of kilometers just with these little plugables that sit inside the router. And so you know it's 120 kilowatts in a transponder card versus 23 kilowatts is sitting in the router, huge power savings when you put these coherent transponders in. So that architecture and then more and more optical integration into our systems is something that we are really focused on. And then additional things now we have already started building liquid cooled systems, right, highly efficient liquid cooled systems in our 50T generation, 100 T generation and so forth. So there's been a huge amount of investment in just the core technology needed to support infrastructure build out. And then for the likes of New York cloud, Sovereign cloud builders who get who also buy software from us, right, the software that's integrated on those systems. We have capabilities in our software as well for intelligent power optimization, fan control, all kinds of things designed to make sure that we keep the power profile, the the real time consumption of power as low as possible. So, so yeah, this is just scratching the surface. There's a lot we're we're doing here. Speaker 1 And then this I'm assuming that because the advanced packaging of the chips where you we also make our own zurdies and you know, kind of high speed IP that actually really helps with power efficiency as well. Speaker 2 Absolutely. So if you look at what goes inside a chip, you have the packet processors, you have the the IO, you have the zurdies, you have the buffers. So because we control the entire technology stock there, the packaging and the design, we can bake in power optimization and sustainability as a core characteristics of those designs. Speaker 1 That's great. Speaker 2 Yeah. And and then there's a whole discussion around how can service providers be more effective participants in intelligent token delivery, not just over the top token delivery, right. So, so there's a lot of opportunities to rethink how network infrastructure. Speaker 1 We, we, I try to talk to, you know, way I think about the business that service providers are in, because a lot of time service providers will ask, hey, what do you think we should be doing? I think you have to be in the path of token generation. And if you're just in the consumption point of token generation, it's, it's marginally interesting. But if you're in the path of token generation, by the way, the amount of you know, kind of infrastructure that they had laid out with 5G is, is actually this might be the mechanism for them to go out and monetize all of that, that they haven't been able to fully. Speaker 2 A lot of the infrastructure, that's a very good point actually, because both both great points. One, how can they be in the middle of token generation and intelligent token delivery and not just be, you know, providers of over the top tokens, but also the infrastructure to deliver all of this by definition is very distributed because unlike training clouds, which are a few hyperscalers, big data centers, gigawatts, the inferencing ecosystem is all, you know, kilowatts and megawatts of capacity from a power perspective. And telcos have already built a very rich set of distributed locations, points of presence or pops as they call them, where they have just the right amount of space power, 20 megawatts is plenty and the fiber to deliver all of this. So. So you're right that they can actually realize a lot of the. Speaker 1 CapEx that has already been spent can naturally be monetized. Speaker 2 Now. Cisco's Unique Value Proposition for Hyperscalers OK, so now let's talk about Cisco and what Cisco can do to help. And let's talk about each one of the class of customers. Let's start with hyper scalers because they're the most sophisticated on the kind of things that they're doing at scale. And then we'll go hyper scalers, neo clouds, sovereign clouds, and then service providers. So let's start with hyper scalers. Tell me what's why hyper scalers are coming to us. Tell me why we've actually gone from what was virtually for AI workloads, almost zero in orders taken three years ago to we thought we were going to do a billion last year we ended up doing 2.3. We thought we were going to double that this year at 5:00. We're going to end up doing close to 9. Why that level of surge and what is happening over there? And why is Cisco uniquely differentiated? Speaker 2 Yeah, no, a lot of credit to our engineering teams, to our sales teams, just a whole cross functional set of teams for having pulled that off. But if you think about what hyperscalers are looking for, they build first of all, technology is paramount, making sure that you have innovative technology, leading technology. For example, they're looking for 50 Terabits, 100 Terabits, which is beyond 1. Speaker 1 Thing I have actually told people is you should expect over the course of the next few years we should have A1 petabit switch with a trillion transistors. It's not. Speaker 2 Unrealistic. It's not unrealistic at all because we are already in the 100 T era. And as you can imagine, we are working with our customers two or three generations beyond. So having the right ingredients for delivering technology, silicon optics, liquid cooling, various kinds of optical packaging, right, NPOCPOXPO, all of these, we have all of those assets. We have our own in house silicon with Cisco Silicon One which has been a phenomenal investment and it's really paying paying off now. Speaker 1 Started that a decade ago and it seems like a genius move now, but I'm, I take no credit for it, but I think it was, you know, you have to thank all the people that made that happen. Speaker 2 Two of the things that have worked out really well for us, Silicon One and then our Acacia acquisition gives us the capability to bring coherent optics technology, not just plugables, but a lot of those technologies are now coming together. Silicon and optics are coming together. So technology is a big piece of it. We have that. We are also able to build the kind of systems that hyperscalers want, right, because a lot of them want very bespoke systems they Co design with you. So you need to have those skill sets and those abilities. We have that, so we're able to deliver that to them, right. They look for diversity of supply chain of technology. We have that silicon one and our optics is something we build. We own the entire supply chain right down to you know buying wafers from TSMC. So that is an advantage. We also engage. Speaker 1 Yeah, we're not just building ASICS. We have a full customer owned tooling and customer owned tool kit for you know go that goes direct to TSMC. So that exactly hops to TSMC matter. Speaker 2 And that has huge value. And then of course we engage with them on a variety of different business models. They buy chips from us, they buy systems from us. We give them software support for what they want, right. Some of them put Sonic, others put F boss. So we are able to engage with them in a variety of different ways. And so the combination of technology, the ability to engage the way they want us to engage, Co designing the just the, the supply chain, right, not to be underestimated the the value of our supply chain, all of these factors together and. Speaker 1 Scale really matters in that dimension. Speaker 2 It is absolutely paramount because the the, the order at which they consume systems is it's not something you you need a really deep supply chain capability to be able to do that and, and Cisco has that. So all of these have been very useful. Speaker 1 In OK, So what you're saying basically is one of the big advantages is you've got a technology stack that is a fully vertically integrated platform that's a Co designed full stack everything from silicon and optics to systems hardware, system software and then you know on and on above that that's a huge differentiator. What else? Speaker 2 Engagement models. They need to engage in a variety of different ways, all right, responsiveness, speed of operation is extremely important. We have seen cycles, technology cycles compress. A decade ago, we would release new silicon every four years, not just as the entire industry. Now we are releasing new silicon every 18 months, right, and compressing even more. So hyperscalers need that kind of, you know, velocity. And then also looking forward, right, there are all these technology investments. It's a huge amount of investment, as you're well aware, that we need to make in these spaces and Cisco is willing to make that investment going 345 and further, right. So all of these are critical. Scaling AI: Solutions for Neo and Sovereign Clouds So, okay, so that's how you start with the hyperscalers. Clearly there's momentum over there. Let's go to Neo Clouds and solvent clouds. Yeah. Why should they think about Cisco and what are they going to get from us that they don't get from elsewhere? Yeah. And by the way, silicon diversity is a big, big reason as well, because you don't want to get locked into just one provider. So you need to make sure that you have multitude of providers. Speaker 2 That is absolutely true. Speaker 1 And there's only a few handful of people that make silicon for the networking silicon site. Speaker 2 That's right. Yeah. And so and Cisco obviously now we have proven this right, having been in the most demanding hyperscaler environment. So we definitely have I, I I believe earned our credibility that we can deliver these neo clouds slightly different, right? Neo clouds look a lot of the time, some of them, the biggest ones are looking for the same kinds of things that hyperscalers are. So we can meet them the same way. But these? Speaker 1 Tend to be, you know, Co design projects where it's almost semi custom work that needs to get done because everyone has a different architecture, different roles that they need to go out and fulfill and so on and so forth. Speaker 2 Absolutely. And so we need to get engaged much earlier in the cycle with them even as they are conceptualizing design. So, so we do that. And so it's with a few of these big ones, Frontier Labs and so forth, It's the very same kind of engagement model. But a large majority of the Neo clouds actually are looking for fully integrated solutions. They want the silicon systems software even for us, they may not necessarily go with open source options. So they need a fully integrated stack because for them being able to monetize those GPUs they're buying ASAP is most critical. So they want the networking that just works management for all of that. So we have the ability to do that. So with our Nexus portfolio, for example, we bring in the complete stack including our Nexus dashboard, which is now being integrated as part of Cisco Cloud Control. And what's more, we also have partnered with the likes of NVIDIA so we can make sure our networking technology and our switches and a fabric works really well with NVIDIA GPUs, right, with their NICs and and so forth. Speaker 1 And spectrum. Speaker 2 X spectrum and we have Spectrum X support as well for those who want that variety. We support both. So it's a very the, the, the key value they get from Cisco. There is the ability to move from concept to implementation very, very fast because we bring in the whole stack. And by the way, we connect these to our service provider portfolio as well because if they're building vans and data center interconnect as well, right? And that's a key topic we should talk about because some of the technologies we are building, especially in data center interconnect and scale across use cases, we are uniquely. Speaker 1 Let's talk about all those three because there's scale up, scale out, scale across. On the scale up side, it's largely within the rack and you have to make sure that you connect the GPU's within the rack, right? Scale out is when you do row level connectivity of multiple racks within a row. That's right. And we have special chips for that. So talk to us a little bit about how hyperscalers, Neo clouds and Sovereign clouds are thinking about that. Speaker 2 Yeah. So, you know, the simplest way to think about it is first all these AI cloud providers and Neo clouds and hyperscalers, they try to maximize throughput inside of a rack that scale up. But you run out of capacity very fast 'cause you can have 72, maybe more in the future, but not hundreds and thousands of GPUs inside a scale up environment. So you need to interconnect them and that's the scale out that you refer to, right? We have switches typically connected in leaf spine configuration, so within a data center typically. And so Cisco has our G series silicon and we build, you know, high performance switches 50 Terra, 100 Terra more. With the right optical interconnects and so forth for that scale out environment. But what's becoming interesting is in some cases now even the confines of a data center are too small to accommodate some of the large training workloads. So you need to actually connect. Data centers often spread hundreds of hundreds of kilometers across and run workloads across these two data centers, and so you need we. Speaker 1 Built a separate chip. For that we built a. Speaker 2 Separate chip, P series chip and it's the P series chip. And there's a reason why we built a separate chip. It's the same architecture. It's still silicon one, but the P series also has deep buffers and that's critical because once you start going over long distances, there is higher possibility of jitter. There is higher. Sometimes you cannot control the traffic. You know, if it's pure back end traffic, maybe you can control it a little better. But if it's front end traffic, you cannot control 'cause there's multiple flows, there's a lot more variability. So the the you know the impact of dropping a packet is very very high and so shallow you. Speaker 1 Have to restart a training run. Speaker 2 We have to completely restart the training run, so it's very high. So shallow buffer switches don't actually work very well. In some very highly tailored applications they do, granted. But in a lot of the applications where you need to go across, you need deep buffers, you need the ability to temporarily hold packets while the jitter clears, and so forth. So you can guarantee a much higher overall throughput for your for your run, whether it's training or inferencing. Speaker 1 OK, so that's you've now covered Neos and hyperscalers Solver clouds is just a variant of Neo clouds just with a nationalist agenda as well, because they need to make sure that they optimize for the nation that they're in. Building AI-Ready Infrastructure for Service Providers with Cisco That's correct. And so you think about companies like Humane, you think about companies like G42, those that happen to be, you know, kind of, you know, great partners of ours, right. Walk us through now on the service provider side, what has changed in the way that our technology is not just perceived but implemented within service providers? And what are the major things that if you were to tell service providers, here's what you should know about Cisco? Yeah, this is what it would be. What would that be? Speaker 2 Very good question. So we two years plus ago when we started seeing when we had the ChatGPT moment, we began thinking really hard about how do we make sure we build AI ready infrastructure for a service provider? What does that mean? And, and So what that has resulted in is an entirely new portfolio that we've built from the ground up Cisco Silicon One base, but we have a new line of chips just for our service providers. These chips have embedded capabilities, telemetry sensors, etcetera, specifically designed to identify AI traffic, characterize it and so forth. So what this has allowed us to do is, and it's not just the chips, we've built the right systems on top of it, the management and so forth. But what this allows us to do in terms of outcomes is 3 things, right? First, for service providers, they can now actually make what I would say intelligent AI fabrics, not just pipes and networks that connect endpoints, right? So they can get visibility into the AI traffic that's flowing through the network. They can make decisions on that AI traffic. They can slice AI versus non AI, create different kinds of low latency, high performance slices that they can monetize by selling to enterprises that are sensitive to those kinds of traffic patterns and so forth. So they can actually take a lot more intelligent decisions based on the AI traffic that's flowing through the network, right? To your point, be part of the token generation and not just carry it over the top. We've also done some interesting things. For example, if service providers. Now if you imagine a world where you'll have these millions of agents, these agents need to talk to inferencing clouds. Someone needs to tell these agents which inferencing cloud to go to. Speaker 1 So you need some kind of intelligent routing technology on that. Speaker 2 Front. Exactly. So if you think about, you know the closest proxy I can think about is DNS, right? Domain name servers, where when humans type a name, it gets translated to an IP address. Someone's determining where that should go to. These billions of agents will need similar kinds of capabilities because you need to be able to provide cheap tokens, high performance tokens, complex reasoning tokens need. Speaker 1 To know which model to go to which is going to be the most fit model for the prompt that you've actually posted. Speaker 2 Exactly. So you need a proxy function and service providers can provide that. So we are bringing in capabilities like that, that allow them to be very much part of that intelligent token delivery. So that's one of the things and there's a lot more we're doing there. But just to give you that example. Then the other thing that service providers can do is actually leverage that infrastructure. You talked about where they built things out during the 5G era. They can actually repurpose a lot of those physical assets, power fiber to actually host inferencing clouds because they they'd have the natural advantages already. In fact, if you look at so many neo clouds, that's exactly the market they're trying to get into, set up these data centers so they can provide inferencing. Well, service providers already have natural advantages that they can capitalize on. So and so we've built an entire stack. So our Nexus based secure AI factory, everything we talked about earlier, right with the right management inside the data center connected to the AI fabric, we have that full end to end solution. It's a value they get from Cisco that they would otherwise have to, you know, engage multiple parties to Plumb together. Speaker 1 So what would what would the service provider have as an objection and why not to work with Cisco? Speaker 2 Honestly, I can't think of any at this point in time. Jitu, because I think if you look back five years ago, four years ago even we didn't have the right technology blocks. Now we have the silicon, we have the systems. Speaker 1 What was missing stack? Speaker 2 Four years ago, so AI think our existing platforms were getting to be legacy. We didn't have the latest, but now we fix that. We have edge access score and entirely new portfolio. It's getting to be very feature rich. We have an agentic op suite which is a very important component of all of this that's getting integrated into a Cisco Cloud Control. Speaker 1 Cisco Cloud Control is the unified management plane for those that don't know that can actually have every single, every single technology we have is now going to be accessible through that. Speaker 2 So you can do multi domain management, you can bring in security and troubleshooting, observability, all part of a unified ecosystem. So we have that, right. And then the other thing we've done is we've actually brought and you like to talk about this, the platform effect, right? So we've brought all of these capabilities together across our different pieces of the Cisco portfolio, integrated security into our infrastructure, right? Security has become a huge thing, being able to react very fast to vulnerabilities. There's a functionality called Live Protect that we've been talking about, which allows you to apply compensating controls really quickly and you know, improve the security posture dramatically. Things like that we have introduced into the infrastructure and we have reference architectures now that can connect our enterprise solutions, SD, Wan and others to our service provider, intelligent AI ready networking into our data center so our customers can get the full stack end to end from Cisco. Again, it's something that is very hard to Plumb if you are doing it piece meal, right. So these are the kinds of things that we have today that we didn't have just two years ago. Addressing Unasked Questions and Future AI Challenges What questions do you think am I not asking that should be getting asked so that the world actually has a view that's congruent with the level of innovation that you folks have done? Speaker 2 We've been through a lot, but I think the one thing that I see with the service provider community in particular is I think they are underestimating how much opportunity they have. And so just the education of what Cisco brings to the table, all of these things we talked about a full portfolio for the AI economy, helping them be a big part of the token economy. That is what I think our service provider customers are perhaps not asking enough. Some of the big ones are now and we are partnering very closely with them. But a lot of them are only now becoming aware of what they can do and how they can be a big part of the AI ecosystem. Speaker 1 I I would say there's probably another area that we are, we're going to need to have many, many more conversations. There's a lot of conversations that have started, but we are living in a post mythos world. Speaker 2 Yeah. Speaker 1 I think there's a lot of aged assets in service providers and the life and the support, these can cause a fair amount of risk. And we need to make sure that we're actually working and partnering with our service providers to make sure that it's not financially prohibitive, but not also something that poses, yeah, risk at a national level because the there's dated aged infrastructure that's not actually that that's being, you know, sweated out for a few, few more years. Speaker 2 That has become a huge topic now with Mitos being released. You know, I just came back from our New York service provider customer Advisory Board where we had thirty of our customers, right, Tier 1 telcos, regional telcos, enterprises, education networks, pretty diverse group. This was one of the key topics on how do we make sure that we can actually start moving faster in terms of plugging security gaps in a network, but also figuring out how we deal with legacy infrastructure because that is one of the biggest, you know, access points for, for the bad guys. So and we we have a bunch of solutions there that we're working through. Speaker 1 Well, Guru, thank you for everything that you're doing. My friend and I, I have to say that the, the progress that's been made just in the past two years since you and I have started working closely together, I've been blown away. And it's we're lucky to have you and I can't even wait to see what the next 5 to 10 years hold so. Speaker 2 Thank you Jitu, I echo that sentiment. Thanks for all the support. Speaker 1 Thanks. Yeah.