Shekar Ayyar, Arrcus (00:07):
All right, welcome all of you. I think this is great. This is the first official Arrcus press conference. I guess I told Ray, we have so much news that we can't just kind of call you individually and talk to you. So thank you all for being here. So we have a number of exciting announcements here at the event. I mean we've obviously been at Mobile World Congress for a few years, but we've grown as a company and we've got a number of neat things happening. So we thought we'll gather around, this should only take about five or seven minutes. We have a couple of our partners here, I'll share that with you as well. And then after that we have plenty to drink and eat. And between Sean and Sumita, they can help you if you want something else as well that we don't have.

(00:55):
So why don't we flip to the next slide. I'll just quickly run through the summary of the announcements here and then go one by one and start explaining what we're doing with each one of these. First and foremost, we are super excited to announce at the event that we have launched a new switch called the TGAX, which is a telco grade AI ethernet switch that we have jointly done between Arrcus and nvidia. Second announcement that we made was that Arrcus along with our partners Broadcom as well as UFI space, and Vincent is here from UFI Space. Thank you for being here Vincent, and thank you for helping Arrcus as well. Along we are supporting a tapio, which is a network provider in the US supporting a number of customers including Japanese customers like Yahoo, Japan, et cetera. And this is a solution that is a next gen networking fabric to build AI clusters and particularly for running large language models using ethernet based networking.

(02:09):
And so we are super excited to get that announcement out. The third announcement that we shared was relating to Liberty Global. So Liberty Global is a customer where they are showcasing both here as well as in their booth in Hall eight a solution which is volumetric video and video applications that are running on top of an AI infrastructure platform. And that platform is composed of Arrcus ACE operating environment along with Fujitsu ran running kind of side by side and on top of an Nvidia infrastructure and running applications from both Eviden as well as Philips. These are essentially the video applications that are running on top. And then finally our partner Lanner and Gene is here from lanner. Thank you for being here and thank you for being an Arrcus partner. So Lanner announced today a solution that incorporates both their hardware as well as Arrcus software to transform the Telco Edge network so that they can run both AI as well as 5G on that infrastructure.

(03:18):
So now I'm just going to go and share a bit of detail on these announcements and then of course open up for questions and then like I said, a good amount of social networking that we can do after that. So first and foremost, the announcement that we have made with Nvidia on the telco grade AI switch. This is the industry's first telco grade AI switch. Now fundamentally this switch essentially allows the partners to do three things, which I'll jump to the bottom of this slide. First, a lot of the telcos have been looking at AI and somewhat like what they have been doing with cloud is kind of scratching their head saying, okay, how do we actually take this and monetize this? What is it that we are going to start generating revenue? So what Nvidia and Arrcus jointly decided is that we are going to give them first the capability to start monetizing using network functions and services.

(04:18):
So immediately they can take this switch and start basically providing connectivity services network as a service, multi-cloud networking as a service, all of which is done by the orcas, a platform. They can then also use the same infrastructure to provide AI inferencing as a service. And then whenever they are ready, they can see this as a seamless path to also dropping the AI ran on top of the same infrastructure. So this is kind of a three in one product or technology. It integrates well with the aerial architecture that Nvidia had previously released and described. And the key highlights of TGA X in the middle are it's based on Nvidia spectrum for silicon, which is a high performance ethernet switch. It's got ACE networking software from Arcus. It's completely programmable and flexible as you have known arcus to do and it's deployable not just at the edge, but it can be deployed at the edge core data center as well as cloud.

(05:23):
So once again, this is extremely exciting for us. And then if you want to know more details about this, of course we would love to find out how many of these switches you want so that we can take orders today, but we can be happy to share a bit more about this as well. Go to the next one. The next announcement is what we have done for a tapio, which is essentially creating a next gen networking fabric to build out clusters for large language models for LLMs, et cetera, but using ethernet based switching and routing from orcas. So specifically a tapio use cases using this for training LLM on top of Nvidia GPUs for a personalized use experience. Arrcus essentially at the networking layer is the one that delivers the high performance lossless network that is required for GPU connectivity. We are doing this one based on our partner, other partner Broadcom.

(06:20):
So Broadcom's Tomahawk five silicon essentially defines this networking hardware underpinning and UFI space Vincent's company who's our partner, they are building the box essentially that is the networking routing box that goes along with this technology in order to make this all come together, the hardware built by UFI space. The specific SKU in this case is the S 93 21 dash 64 E, I think that's correct, right Vincent. And from an advanced Arc OS capabilities, this includes intelligent congestion control, it's got Rocky V two, it's got priority flow control, EC and dynamic load balancing. This is all a part of the infrastructure architecture that we are deploying. It's got twice the bandwidth and six x fewer switches I should ask Sanjay compared to what? Where is Sanjay compared to the earlier generation? And it's got of course improved power and cost efficiency and we'd be happy to share statistics on this if folks are interested and sorry, stay on that one.

(07:33):
And so this is a combination of aca, Broadcom as well as UFI space supporting a TAPIO as a customer. So go to the next one, this one. The next announcement is around what Liberty Global, I think most of you know is a UK-based communications provider. And they are, like I said, also having a display in HOL eight. So please feel free to stop by and visit them. Liberty Global is doing a demo here of an advanced visual recognition application that is deployed on top of this platform. The platform itself is essentially built on a combination of Fujitsu ran ACA ACE platform running on top of Nvidia and it's running two applications in this case, one from Eviden and the other one from Phillips. And both of these are applications that are basically used to detect video-based detection of things like security features or if you kind of look at break-ins in a parking lot or for that matter when you use volumetric video for sports applications. And all of this is being run on this platform. The highlights of the platform are that it is essentially a convergence of AI and 5G on an GPU based pool using SRV six based mobile user plane, the SRV six M technology on Nvidia Bluefield three dpu. Oh there's ANCO from Liberty Global, come on up upon.

(09:13):
So our partners from Liberty Global are here too, so thank you for joining. So you are joining Lanner as well as UFI space over here. So once again, actually there's a very nice cool video that actually shows cases this application we'll be happy to show share with you. And then you can see that in the Liberty Global booth as well. Now the highlights of this of course are that the platform by bringing all this together with the RS RV six mobile user plane along with the Bluefield three allows for low latency and low cost in terms of deployment of these sort of high fidelity real-time AI applications to being on top of this infrastructure and leading. One of the key things that we do is improved resource optimization as well as managing the power efficiency and cost that is related to that. Oh, one more of our partners is here. Moan, can I invite you for a second on stage? So you're just walking by. So I decided this is Moribayashi-san from Fujitsu. You're one of our partners in this announcement and so you're just in time. So we have all our partners here and

Masaaki Moribayashi, Fujitsu (10:37):
Oh, this is not a scenario. I had in mind, but fortunately.

Shekar Ayyar, Arrcus (10:44):
Thank you so much. Thank you. So Lanner announced today, so this is their announcement and we are participating in it that the combination or collaboration between Lanner and Arrcus that we have created a telco edge device essentially allowing operators to run AI in 5G. And of course this is broader applicable even beyond telco operators and particularly this is useful for multiple different types of applications, industrial automation, healthcare, doing smart cities, et cetera. Combines the MGX Edge AI platform from NAR along with the ACE networking from Arcus and specifically it's got high performance edge computing from lanor using both GPU as well as DPU and then the programmable networking from Arcus with our SRV six map for realtime AI applications. So once again, congratulations to lanor for the announcement and thank you for the partnership. So that essentially kind of completes our quick round of announcements here and of course Arcus, we have a booth here. As I said, we are also in the Liberty Global Pavilion out in hall eight, and then you can likely see us in some other partner pavilions as well, including Intel, red Hat and a few others. Let me pause here. I don't know if anybody from Arrcus wants anything to add anything or let me also hand the mic to Vincent to see if you want to say anything. Vincent.

Vincent, Lanner Electronics (12:20):
Yeah, I think that was it. Arrcus is the most good partner we have. Thank you so much.

Shekar Ayyar, Arrcus (12:26):
Thank you.

Guest speaker #2 (12:30):
Yeah, behavior, but then I think that we look the partnership with Arrcus as a manager milestone to reach out the AI powered edge compute for telco and enterprise space. Yeah, we do appreciate the opportunity to join the Arrcus partnership. Thank you.

Shekar Ayyar, Arrcus (12:51):
Thank you Lanner. An I don't know if you wanted to.

Guest speaker #3 - Liberty Global (12:56):
Sure, yeah, thank you. So we've been investors in Arrcus and now Liberty group of companies a customer as well. So we're really appreciative of the partnership, but overall as investors, we just think that this is a fantastic opportunity for next generation networks to really power transformations like AI and cloud computing. And I think these guys have done a phenomenal job and we're looking forward to future partnerships. Thank you. Thank you so much.

Shekar Ayyar, Arrcus (13:25):
So thank you all. So let me open up for questions in case there are any questions. So

Audience guest question #1 (13:32):
Ask me something I can ask. Sure, of course. Yeah, congratulations on the announcement regarding Liberty, the partnership, how long did it take to launch that service? Is it in production or just being launched today or created? Is that a good one for you?

Shekar Ayyar, Arrcus (13:52):
So this is essentially, it's a demo. So the application is in demo. I think it's actually being worked through the process to then take it to production. And you can of course feel free to see the demo as well, but it's not yet in production. But let me hand it over to...

Guest speaker #3 - Liberty Global (14:10):
No, and I think we at Liberty we've sort of seen many use cases as well, and so I think depending on the use case, some will probably go faster than others. But yeah, you should come by our booth. Certainly there are more of my tech colleagues there who would probably be able to speak much better than I am.

Guest speaker #4 (14:29):
I can actually add a little to that. It was last year at this time that folks from Liberty Global met us over here and said this is our vision for an AI powered telco. We are interested in bringing all these applications to the edge. And so we ideated with them and came up with collaboratively with the intent of putting everything together on one common infrastructure that would host the 5G core. It would host AI applications as well as networking. And so our innovation in networking essentially enables the low latency delivery of these applications at the edge. So it started the ideation right about there.

Audience guest question #2 (15:18):
I'll have a go. Thank you. So I'd be curious to hear you expand on the monetization angle and in particular the AI ran opportunity. How significant do you think that will be as an opportunity?

Shekar Ayyar, Arrcus (15:35):
Yeah, so this is kind of core to what we had announced with the TGAX. We said that we're going to bring three things together just so that there is in fact a very solid path to monetization. I think if you look at just the RAN component of it, there's still a little bit of jury out there in terms of how quickly is the RAN going to get virtualized, how quickly is it going to run on A GPU, does it need to run on A GPU? And what does AI ran actually mean? So what we thought is that you really need to get a little bit of a phased path to getting there rather than just sort of telling everybody, oh, you have to forcibly move your ran over and then it's going to be an AI ran. So therefore what we've come up with is essentially a combination which says, look, you can start monetizing services that are here and now today, and these are simple connectivity services that everybody needs.

(16:29):
These are bread and butter things. And you can start basically seeing that as, like I said, it can be VPN services, it could be security services, it could be things that are related to cloud networking. Then when you start looking at it and saying, look, I want to actually build ai, it goes without saying that AI has both training and inference and inference is going to require a distributed architecture. It's going to require networking and it's going to require you to do things at the edges. So therefore this sort of a platform then becomes important to have that. And so almost everybody's thinking about doing something in ai. And so now that same platform then enables AI to add that. And then when you are ready to bring your RAN infrastructure on top of that, you now have networking and ai. So it's actually easy to then make the transition to AI ran. So this is now a very, what I would call as a credible path to monetization as opposed to saying build it and they will come, right? I mean that's the whole idea here.

Audience guest question #3 (17:33):
Excuse me. Hi there. It's a very impressive list of partners you've got and customers you're working with. I'd like to ask you however you want to answer it. I realize there's a lot of people in the audience. How do you decide who does what and who's paying for what? You've got so many different interests here and I'd love to hear how you as a player in this ecosystem using a very different model from others. How do you figure that out?

Shekar Ayyar, Arrcus (18:06):
I mean, I kind of break that question into two parts. One is sort of the technology behind it, and the other is really the ease of bringing everything together for the customer. So on the technology side, I mean it's not rocket science in the sense that we know what we do at Arcus, we kind of make really good software as a network operating system. It lands on top of different processors and it can work with different hardware vendors. And so we have been working hard to build that ecosystem of partners around us so that we can collectively then build and create solutions. And every once in a while there'll be a use case or two or three that will come out from our customer engagements that then tell us that this is going to be an application or a network service or a function that rides on top that we can start then marketing that as a use case. The good news with us as Arrcus is you can actually run us as a low end switch in a data center, but you can also run us as a high-end distributed network that is a chassis replacement for a large scale router.

(19:15):
And so to answer the first part of your question, I would say we know which partners we want to work with based on the technology stacks that are required in order to make the solution real for our customers. How that then comes together for the customers, I would say there are two paths to that. One is a slightly more traditional path around an integrator. So we are obviously working with a number of different partners that can then help pull these solutions together because ultimately customers still like to work with a single entity and make sure that service calls are routed in an efficient way and that they don't have to call six different people to get things done, et cetera. So we have now worked with our partners to make sure that these things can come together in an integrated way for the customer, but equally, we ourselves have also presented as a single throat to choke if that's what the customer wants us to do. So I mean between, for example, our partners like ufi and if the customer tells us, Hey, we want Arrcus software on top of a UFI box or on top of a liner box, we will pull that together and play the role of an integrator for a customer, should that be what their preference is. Okay, time to eat and drink and make merry, thank you all.