So, thank you for being here on day three of CES. I'm Brian Markwalter. I'm a senior vice president with CTA. So, we own and produce the show and I love having everybody here. I do market research and I also do technology and standards. So I get to see a lot of different technologies including spatial computing and work with most of the companies here. If you've looked ahead at this session, you know we have a rockstar panel, an incredible lineup of companies, and we want to have an information-packed session here and cover a lot of territory. It's a big topic. So AI and spatial computing, we've merged like two of the biggest things we could find. If we could throw robotics in here, we would have everything, I think. But so we're going to get going. I want to kick off, rather than me trying to introduce panelists, I'm going to let them say who they are and where they fit in this ecosystem, because we really are coming at it from all parts of the spatial computing ecosystem. So we will just go down the line and let our speakers do their thing. Neil, we'll start with you.
Sure. Yeah. Thank you. So my name is Neil Trevet. My day job is at Nvidia. You might have heard of Nvidia, I guess. So we do engines for 3D graphics and compute acceleration and of course AI acceleration. But as normal, I'm here with two hats. With Mark with the standards track, I'm also the president of the Kronos Group. We do open standards for 3D graphics, compute acceleration, and AI acceleration. So the two are overlapped and the question is what is AI having an impact on your activities? It's a fundamental impact. We're finding that the Kronos standards, for example, are used for spatial computing and therefore the metaverse. Metaverse just being spatial computing users connected together. And rather than being competitive, AI and the metaverse, they are two threads of the same story. They're totally connected. AI is enabling the metaverse or spatial computing because it's helping solve all the really difficult problems that seemed impossible just a few years ago. How to recognize scenes, get context-aware of the environment, understand users' gestures, language. But also the metaverse, spatial computing, virtual worlds are going to be where we train AI to interact with the real world. And spatial computing in the metaverse is going to be our user interface into AI as we use AI in the real world. So really these are two threads of one story. So the interaction is pretty fundamental.
Good morning guys. My name is Ziad Asghar. I am the general manager for the extended reality, wearables, and personal AI businesses at Qualcomm. You know, this is an amazing time. This is the time when we are seeing this transition from personal computing to mobile computing to what we call spatial computing. And this aggregation of XR and spatial with AI is really making certain use cases possible that were just not possible at all in the past. And especially what I'm excited about is these devices. Devices like smart glasses and pendants and smart watches and, you know, I call them camera buds or earbuds with cameras on them. These are all devices that are actually able to interact and understand the environment around you and they're able to do certain things that a typical device that you might have in your pocket cannot do. These devices, and at a very high level I'd say is we need to do them in such a way that they need to give the consumer superpowers. And I think with this merger of AI and spatial, it allows us to do exactly that. So very exciting times.
Hello everyone, my name is Anne Hopson. I'm the policy lead for our metaverse products at Meta. I've been working on VR, AR, and immersive experiences for several years at this point. First in academia and then in industry. I have a PhD in applied economics, and the applied part is to policy. Policy broadly refers to legislation and other formal or informal rules of the road that may impact how products are developed or how they are used. My role at Meta is to counsel our product teams with an eye towards legislative trends. At this point, we're well on our way to building the next generation of the internet, the spatial internet, where billions of people can feel a sense of presence and togetherness, no matter their device or their location. How many of you have used a Quest headset? Raise your hand. Pretty good. Pretty good. How about any of our AI glasses like Ray-Ban Meta, Oakley? All right, we're getting there. What those things allow us to do is blend the physical and digital worlds and Horizon is the social fabric that allows people to move across these experiences on different devices. Quest is not the final form factor. Devices are getting thinner and easier to wear. Wearing some right now. These are the Meta Ray-Ban displays. They have a Qualcomm chip, which I learned a little bit earlier. So lots of collaboration up here on this panel. What they do is they give AI a first-person view of the world as you are experiencing it in real time, which allows AI to be more useful in the moment.
Just to start, I just want to say thank you to the audience for attending. This is day four for me at CES and after four days at CES all I could say is what an amazing time to be alive. My name is Patrick Hadley. I'm the sponsored AR product leader for Snapchat. Sponsored AR is our ad solution that has augmented reality for brands and advertisers. Snapchat has been around for about 14 years now. We have 943 million monthly actives and about half of those are daily active users. Snapchat brands itself as a camera company, not as a social media company. And really we believe that the camera is the best way for users to visually express themselves, learn and have fun with family and friends. So really spatial and AI has been part of our DNA ever since the initial conception.
Hi everyone, my name is Azad Balaban and I'm a product manager at Niantic Spatial working on 3D reconstruction. You might have heard of an app called Scaniverse and Pokémon Go. Those are both products that we've made. We have a long pedigree in mapping and augmented reality technologies. Building games in the real world all around getting people to, you know, leave their house, go outside, do really interesting things with XR technology. Nowadays with Niantic Spatial we're really interested in taking all the data that we've been collecting from around the real world and creating a large geospatial model. So the way spatial computing and AI connect for us is we're all about building a living and breathing model of the world that humans and machines can talk to. And so that's about creating a shared coordinate system at the very minimum between headsets, robotics, drones, anything essentially with a camera and putting it all together into a real-time experience that people can interact with. And lucky enough that we get to work with all the companies that are sitting here including the Kronos Group on Gaussian Splatting, Snap with VPS for their spectacles, Meta with also the Quest passthrough system so that we can create a cross-platform VPS system. And I'm blanking now. Anyway, yeah, and Qualcomm actually we built a reference AR headset design so we got to be a neutral third party and work with everyone that's in the field in all these platforms.
All right. So, you can see why I'm excited about this panel. Like, we truly have every player here that is bringing this to life. So, but this is not easy stuff. We've seen versions of VR and MR and every name for a few years at CES, but it does feel like, you know, it's really being deployed. Sort of the same story. I was only half joking about robotics, but that same story. Like as I talk to CTOs, they're really impressed with sort of like the motion to market, you know, and seeing this stuff really get turned into reality. But there are hard problems to solve. So I would like our speakers to address some of the challenges. And so anyone, I know we're working well, this is why some of your standards groups exist, right? But if anybody can chime in here and let's address some of the challenges you see in merging AI and spatial computing.
Sure, I can go first. I think one challenge that's unique is legislation and regulation and it's increasingly an obstacle to innovation globally. Realizing the full promise of AI is as much a matter of policy as it is technological investment. AI needs talent, energy, data, and compute. And it's going to take government and industry working together to make those technologies accessible. One forum for advancing that is the Metaverse Standards Forum and also companies like CTA are bringing folks together to make sure that we're collaborating on industry best practices and also conversing with governments to ensure that these technologies can come to fruition.
I can go next. I think, you know, these devices we talk about like smart glasses and, you know, Meta and Snap are great partners for us in that. To be able to have the level of compute that you need in these devices, I'll come much more from a product angle on this, which is basically now you got to be able to have the amazing ability to give eyes to AI, great camera technology. You need to be able to have the best-in-class ability to do AI processing on the device because you don't have the ability to send all of that data all the time to the cloud. So you have to do some processing on the device. So you need to have pristine best-in-class performance per watt for doing AI processing. You need to have great connectivity because that's how you get data from one place to the other. You need to have best-in-class power consumption. And you got to make all of this fit in a small form factor like smart glasses. That takes some amazing amount of innovation because just imagine a typical smartphone has 20 times the battery size of a typical battery in a smart glass. But each and every consumer, one of you, the guys who are basically using it, they expect the same experience that they have on the smartphone on the smart glass. That is where we have to do a lot of innovation. I call this, I like to call this a Qualcomm-sized problem because we have been working on basically doing more compute for any given amount of battery life for a long, long time. We have devices and we build devices that are not plugged in. So we really know how to sip on battery and then be able to bring in the best-in-class camera, AI, rendering capability to be able to bring them to very small devices like smart glasses. But what's also very interesting, if you walk around the CES show this year, it's about devices that are not just smart glasses. It's also devices, for example, that might be an AI pin. It might be something that sits on a different part of your body or it might be a smart watch or earbuds with cameras. These are all tiny devices, but at the same time, people expect them to be able to do everything that you get on your PC. And that's where the challenge comes. That's one part of it. Another one is we need to be able to make sure that these devices have to be fashionable. They need to make you look good. If it makes you look like a cyborg, probably a problem there, right? And I think that's where a lot of people are working together. And we have the technology today to be able to miniaturize and to make these devices in the form factor where people would feel very comfortable wearing them and be able to benefit from them. The part that I really also like a lot with these challenges is the amazing opportunity that comes with it. Right? Just imagine a person who might be hearing impaired and now they're wearing glasses like the one that she's wearing and you can have basically the transcript of the words coming on the display right in front of you. That's an amazing benefit for a person like that, right? Just imagine somebody else who might be visually impaired. You get to a cross-section and the speaker in your glasses is able to tell you that hey the light is red at this point in time. You shouldn't cross. But that's just scratching the surface. Each and every vertical has a different challenge. See today if you look at it the glasses and devices are coming much more from a consumer angle but we have verticals like industrial, education, healthcare, elderly care, productivity, each and every one of them have a different challenge but an amazing opportunity as well and that's why this space is ripe for innovation and that's what makes me very excited about it.
Cool. Yeah, the folks here are doing an awesome job on the hardware. We also have big problems and opportunities on the software side. I think one of the biggest challenges is creating an open and interoperable content and software ecosystem. You can imagine someone wearing awesome smart glasses finding geospatial located services every few paces you're going to find a new service. There's no way you're going to be able to stop and install the app every few paces. You need to be able to access services as easily as you click on a 2D web page today. And in fact, another framing of the metaverse, it is the spatial evolution of the web. We need to, I think, have a web browser just like Netscape ignited the internet revolution. We need the same ignition point for the metaverse, spatial computing, AI ecosystem and it's quite a big challenge because a 3D metaverse browser does completely different stuff to what the 2D browsers are doing today. A 2D browser is taking single source content putting it in a 2D rectangle. A metaverse browser is going to be taking multiple services that are geolocated, compositing them all into a 3D environment with multiple users, multiple AI agents that are geolocated and synchronized over the real world if you're using AR glasses. And that's going to take a whole bunch of standards. We need to create basically a sister stack to the 2D web stack and now we're going to need the metaverse web stack. It's going to take a bunch of standards which is why incidentally we formed the Metaverse Standards Forum to get all the standards organizations we need to actually talk together and to cooperate to build this. So despite the rumors you've heard, the metaverse is not dead. It's just waiting for its Netscape moment and it's going to come soon, we hope.
Yeah, one thing that's unique about Snap is as soon as you open the app, you're landing on the camera, right? And all you have to do is just swipe to the right and you're immediately put into some of what we call AR lenses. So one stat that's really interesting is our AR lenses are used 8 billion times per day. So not per year, not per month, but per day. So we really have kind of a playground where people are coming into our app to look for these type of experiences. So one thing that I would say is there's maybe two different ways of challenges. First would be the early kind of innovation in technology. There's a risk of leapfrogging or investing too early. Let's just use maybe like NeRF versus Gaussian. You don't know which one's going to potentially be the winner. So how much should you invest into that now in a technology standpoint? And then the other standpoint is around the interoperability. There's reasons why you don't want to stifle innovation by putting standards in too early. Gaussian's a great example. So, you know, getting assets from outside of our company, what we're putting it in and how we're doing it. Those would be the main challenges that we see from the early innovation. And then because we've got that scale we kind of have that test bed. The other thing is for the more mature type of, you know, generative AI experiences where you can act like you're an Egyptian, you know, in the pyramids, it's really about compute really is a problem for us. And then what I'd say is as the models continue to evolve, we're replacing them continuously of what's going on in our back end. What's the newer one that has the best example? So just having the organization aligned to have that mentality. It kind of comes from the leadership down, has to be really bought in at every level. It can't just be kind of a core center of excellence about AI. It really has to be the whole company's bought in because it's going to impact kind of your entire product.
For us, for the large geospatial model that we're building, the data that is composed of that composes the model is very important to us. It's all about capturing the real world because we want to build and understand, you know, what the real world is and how to reconstruct it, how to localize to it, and also understand what's within it. Thankfully we have been collecting data over a decade now and understanding, like we have over a million places scanned around the world over many months, seasons, years even, and so we get to build a good understanding of how a place looks and how it changes. But of course compute and inference, like those are the challenges that everyone on this panel understands. We have ways of dealing with things locally and things in the cloud and I think it's a very tricky interplay to determine what should be something that is, you know, at a minimum needs to run in a GPS-denied environment or with no network connectivity locally versus something that can run on the cloud that has the benefits of that and also the latency hit that you can get from that. But yeah, for us the data is very, very important and we continue to collect it and to train our models. Thankfully there's research that's happening, you know, every single day, and especially when it comes to 3D reconstruction on the radiance field side, that field is advancing every single day and as long as we have posed images of places we get to just retrain the same data with new models and gain the benefits of that.
All right, fantastic. So, I want to, this may be for, so in the AI space, CTA, we've been very active. There's, last year there were more than a thousand bills proposed at the state level. So one of the issues we face is states running off and doing their own thing to try to regulate certain areas. Is there anything that's spilling over in the spatial computing that you guys are, are there any, do you want to touch on something?
You betcha. Yeah, so I think our recommendation is that AI policies need to be grounded in science and evidence. The approach we see being taken is legislate now, ask questions later. And that is a surefire way to make sure that the potential of AI is regulated out of existence. Specifically there are overlapping regulations especially in Europe which is seeing the macroeconomic consequences of introducing those regulations and slowing down the advent of that technology in those spaces. Specifically the AI Act, the ePrivacy Directive, the GDPR, and the new batteries regulation. And this is a really important one to make a point on because it touches on what my partner here from Qualcomm mentioned, which is that in order to have all of these small form factors from a pin to these glasses, they are not viable without shrinking down that battery technology. So something like the EU batteries regulation really prevents innovation in this space.
