RAISE Summit 2025 | Robert Brooks, Lambda

Clips
More from RAISE Summit 2025

Robert Brooks IV

Founding Team & VP, Revenue

Lambda

play_circle_outline Superintelligence Cloud positioning highlights advancements in AI and machine learning capabilities.

play_circle_outline Lambda's Rapid Growth: Scaling AI Infrastructure from Megawatts to Gigawatts to Meet Skyrocketing Demand

play_circle_outline Navigating Energy Constraints: The Future of High-Performance Computing Beyond Traditional Supercomputer Limitations

play_circle_outline Democratizing Software Creation: A Vision for Neural Operating Systems, Encrypted Communication, and Secure User Data Sharing

play_circle_outline Plans for larger customer announcements and AI lifecycle integrations are in progress.

Info
Transcript

Robert Brooks, Lambda

Robert Brooks IV

Founding Team & VP, Revenue Lambda

In this RAISE Summit interview, Robert Brooks, founding team member at Lambda, joins theCUBE’s John Furrier to unpack how the “Superintelligence Cloud” is scaling from 100-megawatt data centers to multigigawatt AI factories. Brooks explains why four-fifths of Lambda’s revenue now comes from serving hyperscalers and elite AI labs, including a recent 10-figure deal that will deploy 10 000 GB300 GPUs in 2025. He shares how sectors such as drug discovery, financial services and automotive lean on Lambda’s global cloud footprint to meet soaring demand while naviga... Read more

explore Keep Exploring

What is the current state of development and demand for superintelligence and how is your company responding to it? add

What is the current revenue distribution for Lambda's public cloud and AI services, and how does it relate to their overall operations? add

What advancements have been made in accessing high-performance computing resources, and how does this compare to past experiences in computing? add

What is the purpose of the GPUs deployed by Lambda and how do they plan to utilize them in the future? add

What recent developments or announcements are being made regarding customer wins and product usage? add

bolt Powered by CUBE AI

Robert Brooks, Lambda

search

>> Welcome back, everyone, to theCUBE's live coverage here in Paris, France. I'm John Furrier, host of theCUBE, two days of wall- to-wall coverage. A who's-who is here. It's an international global event as scale of AI comes together with the builders, the entrepreneurs, and the leaders formulating their plans to continue to create value. But also, we're starting to see value extraction and clear lines of sight into the economics of AI, and of course sovereignty, and many more conversations. All the top people are here. Robert Brooks is part of the founding team at Lambda. Very, very successful, growing, fast-growing company in high demand. Robert, great to see you. Thanks for coming on theCUBE.

Robert Brooks IV

>> Thanks for having me, John. - So,

>> it's good to be popular. One of the cool kids, "I got GPUs. What do you need? " In all seriousness, we have a massive tsunami of demand. You guys are hitting a demand curve, straight vertical S curve. Clearly, everyone knows that power is constrained, AI and blockchain bounded by power. That's the scarce resource. It's causing all kinds of disruptions and enablement, country, geopolitical relationships to architecture around how I build my data centers or AI factories, whatever you want to call it. Essentially, supercomputing centers are being built and architected, by the way. It's not rack and stack. This is like a whole other level service providers that are coming out, the new brands like Lambda emerging. And what I love about your company is that, in every transition or wave, new brands emerge. Yahoo search engine, Google, and you got the iPhone, App Store creates massive brands. So in all aspects, there's a thousand flowers blooming. But you guys are in the category where it's massive growth, so give us a quick update. We're at a midpoint right now. We're in Paris. So two things, update, midpoint year for Lambda, and then Paris, what's the vibe here? What are you doing here?

Robert Brooks IV

>> Yeah, of course. So again, thanks for having me. We are here in Paris. The scene, I like to do that play on words, related to the Seine river. But ultimately, we're at a really good inflection point for our business. We've historically thought about our business in terms of megawatts, 50 to 100 megawatt data centers. We're now entering the world when we start to think about our business in terms of gigawatts. And that's Lambda's next entrance. That's why we're being referred to as the Superintelligence Cloud today. And we are a global cloud that serves customers all over the world, have a large presence within Europe with our public cloud. And ultimately, it's good to be close to our users, it's good to be here with the industry.

>> Yeah. I love the Superintelligence Cloud. I mean, first of all, AGI is a great conversation. It's all the forward-thinking. But we're starting to see... I mean, we're talking about superintelligence in years, not decades. So Eric Schmidt said on stage, six years, Silicon Valley people are more optimistic, three years... It doesn't matter. It's fast. I mean, six years is like college for me. Okay, maybe four. But in all seriousness, this is really happening fast. And how are you guys handling it? Because you're serving a lot of customers, there's huge demand. Take us through on what's happening with you guys right now. What's some of the makeup of the deals you're doing? What's the company doing to service those deals? How are you expanding the footprint? I mean, it's like the scene in Star Trek, "Scottie, I need more power. " Literally, more power, horsepower, but also physical power.

Robert Brooks IV

>> Yeah, of course, of course. So Lambda, our brand is really associated with our large public cloud, and that's who we were and still are today. But it only represents 1/5 of our revenue. And I think the secret out there that hasn't been told, and maybe is being told right now, is that 4/5 of our revenue is really related to serving large superintelligence AI labs and hyperscalers. And we're going to be talking a lot more about this as time goes on. But ultimately, we're in a position to be speaking in, again, hundreds of megawatts to gigawatts. And it's not just about hiring a team to deploy these types of GPU clusters and rack and stack, to your point. These are armies of people, not AI agents, that have to stack these things up and make sure that they run when you're operating clusters at this type of scale. So for us, the deal-making, the deal sizes have gotten bigger. We've just closed on a 10-figure contract. We deploy that this year in the form of a 10, 000 GB300s handed off to a customer, clustomer, in 2025.

>> Super clustomer. - Yeah, super clustomer, right?

Robert Brooks IV

>> And so ultimately we're really excited about that, and our position in the industry.

>> I'm just doing my math in my head. 10-figure, okay. I think you bring up a good point. I want to, if you don't mind, go a little further on the Superintelligence Cloud direction. One of the things that's obvious from the current... Just look in the past couple months and in the news, all the headlines about Meta offering the $ 100 million, one-year deal. Actually, Information copied theCUBE by calling them athletes. I think, Frank, we called tech athletes 16 years ago. You're truly seeing a lot of tech athletes out there with these kind of free agency contracts. But what it speaks to, to me, and I want to get your thoughts on this, it's a new categorical shift that's a first-generation development of your world, which is, the producers of AI are centers of excellence in the sense of, there's a lot amount of intellectual capital, almost like a university vibe. And the companies have that same thing. "I went to Brown," "I went to Stanford," "I went to Northeastern. " I mean, depending on what school you go to, you are in that, quote, "club". It feels a lot like college, but they're producing massive amounts of value and they're fueling the models, they're fueling the acceleration, and they're doing hardcore computer science. This is where the action is. Not just OpenAI, it's the people behind OpenAI. So talk about that. I mean, did I get that right?

Robert Brooks IV

>> So computer science is one aspect of it, but the people that are signing $100 million offers are experts in math. So why is Lambda named Lambda? It's after lambda calculus, functional calculus. So the story about machine learning engineers is, yes, they can maybe write a little bit of code, but you would never employ them just to write code. These are folks that are experts in math, and these are the folks that are signing $100 million bonuses. So it's the resurgence of the mathlete, right?

>> Yeah. - That's signing bigger deals than Lionel Messi.

Robert Brooks IV

>> I mean that's a fantastic place to be. But ultimately for us, we see a world in which these types of people need tools to ultimately build super intelligence, that tool's in the form of compute. But superintelligence for us, I think is a much better definition than AGI and ASI. For me, it's a better cognitive ability than a human in a particular place, whether that's social or scientific.

>> Yeah. What's your thoughts on this? Because first of all, do you agree it's a new category of research, as a, I won't say product, but as almost a virtual entity?

Robert Brooks IV

>> Yeah. - So that's one.

>> That's happening or not happening? What do you think?

Robert Brooks IV

>> Well, there's two applications that I see. One is more in the theoretical. There are people that are adopting current tools, which I would define as superintelligence today, AKA the chatbots that exists from OpenAI and Anthropic, to crack math problems that have historically not been able to be worked on. They were just too complex. And we are making inroads to those mathematical problems that have never been done before, and those models are helping us do that. That's one area that's a little bit more playful. On the realistic side, with drug discovery, we won NVIDIA's Healthcare Partner of the year in 2025. We have a ton of drug discovery companies that are using our platform, and those folks are able to use the drug discovery tools with the AI models to get to just a lot more tests faster, and ultimately to a curable disease much faster. I mean, that application is the greatest application I can possibly think of.

>> It's interesting, you solve the math problems that get put into the models to do better models. So this is a nice little research flywheel. So I was talking to someone here in Paris, I won't say their name or reveal the gender, but this person said there's a real conflict right now in AI between product- led companies and research. And you have an appetite for the alphas. I call them alpha engineers or math folks, whatever you want to call them. Call them alphas. They want more research, because they know there's more research to be done and they don't want to just get something out and sell their company, and then be in the cadence of, "Let's get ship that product. " Are you seeing that psychologically happening? Because we're seeing, again, this research class emerge as not just academic. I mean, it's academic in the sense of it's research, but it's really robust. I mean, if you're in that market, you're sinking your teeth into some serious hard problems. That's going to attract the best talent. Why wouldn't it? People want to work on hard problems.

Robert Brooks IV

>> I think people easily forget that OpenAI, a year and a half or two years ago, launched Laundry Buddy. It was like this little app that helped you do your laundry, and everybody's like, "What the hell is going on right now? " And that's when Ilya Sutskever started to trend towards leaving OpenAI, because he wanted to focus more on research and less product. And I think that played out in that particular scenario. Do you want to launch Laundry Buddy? No. Do you want to focus on safe superintelligence? Yes. And that's why he had to go start his own company to do that. So there is a dynamic there.

>> It's a nice... Well, it's a

>> balance, as well as needed on both... Again, that's why I think the flywheel is interesting, because the research on the math side gets better, only makes the models better, thanks to what you guys do. All right, talk about what you guys are doing with the Superintelligence Cloud. You got a nice customer base on these big research and big clients, drug discovery, life science is booming. Automotive, hot right now or kind of good?

>> Automotive is hot, financial services is hot, bio is hot,

Robert Brooks IV

>> drug discovery, that's where we're focused a lot on. Robotics has had a nice little kickoff, we're waiting for that next second wave on our side. So we'll talk in the future a little bit about some big robotics customers we've signed in the past two quarters. But ultimately, that's where we see a lot of the compute, and power, and research, and math going into. And for us, we're really focused on letting those customers get to market a lot faster. That's the name of the game right now.

>> Just dial up some compute and GPUs.

Robert Brooks IV

>> Yeah. Research does become the model, does become the product.

>> Interesting. It's fun to cover the AI, because I've been covering HPC, high-performance computing, since it's been around. And cloud comes in, it looks a little bit HPC, but it truly wasn't. And then the high-performance computing market's been like, "Yeah, I got a better server. " It's like inch by inch it progresses, it's kind of boring, and then all of a sudden, boom, supercomputing is actually here. It doesn't look like what it was before. But all those people dreaming up and working on the hard problems were just constrained by having the horsepower. And now it's unleashed, it's like freedom. I can now go faster. And so that's a huge dynamic. What's your thoughts on that and how has that played out?

>> So I'm really glad you brought that up.

Robert Brooks IV

>> A CEO of a company called OpenPipe recently tweeted out that within Lambda's 1-Click Cluster, which is essentially on-demand, high-performance clusters, you don't have to talk to a single human being to get access to these clusters. You can get 1,500 B200s, and not talk to a single person to get them. That is more computing power than the largest supercomputer in 2021, and it's available online, at your fingertips, swipe of a credit card. So this is how fast we're moving. And the democratization of access, I think, is the most important part where Lambda's solving that.

>> I'm so old, when I was in college, you had 30 minutes on the mini-computer, "Time's up. Oh, I didn't compile. Save." But that's gone.

Robert Brooks IV

>> That's gone. - This notion of waiting in line, waiting

>> for my turn, hours, go home, "See you in two days," is over. That literally was two years ago, maybe three years ago. If Boeing wanted to do a simulation on a wing design, they would have to wait literally days. This is now gone.

Robert Brooks IV

>> This is gone. - Okay, so now let's fast-forward.

>> What's the long-term vision? You guys have an operating system vision, which I love. This is where you guys are looking at, now, I'm guessing here, tell me if I got this right, collective intelligence of all the data you got and are serving into some sort of neural fabric-

Robert Brooks IV

>> Operating system. - ... operating system.

>> Yeah. - Okay. Explain what this is.

>> So this is the master plan of Lambda.

Robert Brooks IV

>> We've built this entire massive fleet of GPUs.

Robert Brooks IV

>> By the end of the year, I'm sorry, by the end of the decade, we'll have well over a million GPUs deployed. Eventually, those are all sold to customers and contracts, whether they're three- year contracts or five-year contracts. But eventually those contracts end. What happens to that compute? We are going to use it, Lambda. So we're actually founded by machine learning engineers. We have a machine learning team. We're published at all the largest machine learning conferences in the world. And we intend to make use of all these GPUs for our own research and our own products. We're talking publicly today about neural OS, neural operating system. We believe in a world in which you'll be fundamentally interacting with neural networks, and your operating system will be built like that. You will no longer need to procure SaaS from other vendors to do things. You will fully control the application and you will create it yourself. And it may be for a short-term need, it may be for a long- term need, but ultimately it's prompted in your existence. Now, this feels like a toy today. It should. But we believe that that's the future of neural OS, is going to be prompting software into existence, you having full control of it, no longer needing to rely on other vendors for that SaaS application, and then being able to do things we haven't thought of. So today it's a simple messenger app, maybe tomorrow it's a creativity tool that we can collaborate on.

>> Or something that fits a perfect use case of a workflow or a product. So when you say codifying prompt codes, are you referring to, say, I want to rebuild Salesforce, or will vendors telegraph their agents to have code talk to code? Is it replacing existing, or is that more of just, if they don't step up to the plate? I mean, what would be the model? I mean, if I had theCUBE app, right, that was built by a supercomputer, say, "Hey, I'm going to give you access to codify at will, " almost like a marketplace store, is that the vision? I mean, what's the vision of the interaction and partnership? Or is it just going to build it from scratch?

Robert Brooks IV

>> I think in the year 2025, if you set out to build a SaaS application for a very niche thing, and you force the user through your model and your security, it's unclear if that'll be a successful business. That would've been a successful business in 2015. I don't know if it'll be a successful business in 2025. I'm pretty sure it will not be a successful business in 2035. And so this is the world we're building towards, and it is disruption, which is sometimes an uneasy word to hear. But ultimately, it's the way that we move as a species. And we believe that a neural operating system leveraging neural networks to create software in your own design, and not having to rely on anyone else, is the future.

>> Yeah, or environment, preferences, pre-existing

>> stuff you have, customize on the fly. It's interesting, you talk about a software, you had general purpose software where you just get it out of the box, shrink wrap, and you plug it in and you use it. Then you had SaaS, and onboarding was, fill out those forms, it kind of gets personalized, but you still have to onboard. What you're talking about is auto onboarding, superintelligence automatically onboards it on your behalf.

>> Correct. - That's what you're getting at.

Robert Brooks IV

>> Yes. - It's almost provisioning on the fly-

Robert Brooks IV

>> Correct. - ...

>> either at run time or at design time.

Robert Brooks IV

>> Yeah. And under that will be a bunch of GPUs making sure

>> that you can actually build these applications in real time.

>> And today, again, it might be a simple messenger app,

Robert Brooks IV

>> but tomorrow it could be a collaboration tool within CAD, where you and I can design something that's going to outer space. And maybe we need it to be a certain custom way, and maybe certain existing tools don't use it that way, but I can prompt it into existence.

>> All right, take me through the roadmap on that. We're riffing here. We're on the whiteboard. What's the first building block for the neural OS?

Robert Brooks IV

>> Yeah, yeah, yeah. So it exists with a simple sort of, almost Linux kernel, probably a Linux kernel. And ultimately, it gets to a point... We've done some demos today where there's two neural OS computers that are owned by two different people, and they have a tunnel to each other that we can create, and we can do certain things between those computers. Right now, we just want to do encrypted chat. Maybe we don't trust what Meta's telling us, maybe we don't trust what other apps are telling us, that it is actually encrypted, there is no back doors. This way we can verify that these two applications are built on that technology and it is ensured, encrypted technology.

>> It's truly encrypted. Yeah, exactly.

>> Not fake encrypted or just-

>> Yep.

>> And that's true. I mean, people are nervous.

Robert Brooks IV

>> They want to be confident. All right. Talk about the bounded by power discussion, because energy is a huge factor. You guys are going to gigawatt plan. What does that look like? What are you guys doing at Lambda to continue to maximize the power you have available? And what happens with these constraints?

Robert Brooks IV

>> Yes. I'd say we're in a privileged position because the company's been around since 2012. So we've seen the wave coming, and we made decisions in the early 2020s to allow us to be able to sit on multiple gigawatts towards the latter half of this decade. And there's not many companies that can be in that position, because they're making those decisions today. And those decisions today are forced to be reconciled in the 2030s, whereas we've made decisions much earlier than other people, and so we're in a privileged position in that sense.

>> That's called option pricing. So when you made those bets, obviously that's a good call. Obviously you guys saw that you guys were in the infrastructure business, and you were doing the Bitcoin mining, right?

Robert Brooks IV

>> We were never on the crypto side. We did on-premise infrastructure, and then we moved to be a cloud company.

>> Got it. You were already in the games, procuring.

>> You saw that the capex, facilities, classic real estate investment trust game was going to go value add, and you went after and procured, and not sat on, but have.

>> We have, yeah. So power is a big thing.

Robert Brooks IV

>> I don't think it's going to be... There's not a world in which nuclear reactors come online for a good amount of time that's going to solve this in a big way. I believe that's the next wave. But that's a 2030 wave.

>> You think 2030s is when we see any nuclear of any kind?

Robert Brooks IV

>> With a cloud like ours and our competitors, that's when it's going to meaningfully happen. Maybe with a Microsoft or some of these folks that invested in the Three Mile Island, it'll be a little bit faster for them.

>> Yeah, yeah. We saw some Chinese reports about, they've got

>> foolproof nuclear. Yeah, I believe that.

>> Yeah. It's a different scale right now.

Robert Brooks IV

>> So we're lucky to be sitting on roughly a lot of megawatts in America today, because we made those decisions early on in the decade.

>> Robert, what's your view on and vision around... Okay, do you have to make an assumption... Well, I guess, let me rephrase. What's your bet going forward? Because you got the power, you're sitting on some good power, that's going to give you some headroom. And you also got other assumptions, NVIDIA is going to create better products. They got their five-year roadmap. You guys are working closely with them. Other innovations are coming. The combinations of maybe even using retro H100s.

>> There you go. - Which is now retro NVIDIA.

>> So there's other leverage points.

Robert Brooks IV

>> As architects and infrastructure engineers, how are you guys thinking about the hardware- software combination of, "Okay, we can actually squeeze more value out of existing gear, product"? Given the architecture you're going after, you're basically building data centers, a computer basically. So okay, I'm sure there's going to be some innovation around all kinds of design criteria. What are you guys thinking around that? Because you're going to get the blaze and graze, that's going to give you the advantage, but can you write some software to make the other stuff work too, run policy, price- performance-based configurations?

Robert Brooks IV

>> Yeah. So we're one of the only clouds... I mean, every single one of our competitors has gone multi- chip, and they've gone beyond the NVIDIA ecosystem. We've decided to stay in that ecosystem, and I think that's a really unique path for us.

>> The NVIDIA ecosystem? - Yeah. So we

Robert Brooks IV

>> literally bleed green.

>> I mean, that's what we talk about. We believe in that technology. We believe that running GPU clusters and the millions of operations that concurrently have to go right to make that cluster run is already a hard enough problem, which is why the CUDAs and the NVIDIA GPUs of the world solve it for us. But ultimately, we think that there is leverage, to your point, in developing software for distributed inference that allows us to get to a point where there's some gains that can be had on older architectures like the H100, which, it's so funny that it's an older architecture, to your point.

>> Yeah. No, it was kind of a- - 2023, it came in mass.

Robert Brooks IV

>> But it is an older architecture.

>> It's kind a joke, but it's true. I mean, it's-

>> It's an older architecture.

Robert Brooks IV

>> It's modern legacy. - H100s, H200s, B200s, we have a lot

Robert Brooks IV

>> of those available online, but we're already selling GB300s, that are coming online next quarter. So distributed inference, there's a lot that we can do there in order to maximize the amount of tokens that are coming through these large clusters to serve an even wider array of OpenAI or Anthropic's revenue. I mean, they're projecting what, 80 billion, 100 billion in revenue by the end of the decade?

>> You guys are a poster child for Jensen's AI Factory?

Robert Brooks IV

>> Yes. - You guys are an AI factory?

>> Yeah, we're called the Superintelligence Cloud,

Robert Brooks IV

>> that's our headline, the byline is,

>> Gigawatt scale AI factories for training and inference. So we're there.

>> All right, talk about the second half. We're at the midpoint of 2025. It feels like a whole year has passed this year, so much has happened. A lot of action. What's on the agenda? What's your focus? What are you optimizing for?

Robert Brooks IV

>> So we're going to be doing some large announcements just based on customer wins that we've had. More recently, we talked about how ServiceNow leveraged our 1-Click Cluster product to train a model called Apriel-5B. It's a really small model, but it's really efficient. They then hosted that model on Lambda chat. So we have this full circle AI development lifecycle cloud that we're offering, and companies like ServiceNow are adopting it, which is great. You're going to see more enterprises and AI labs talking about that.

>> My final question is a little bit more of a fun one for you, since you're here, since you're a mixture of experts on theCUBE, because we have our own little cloud that's got some vector embed and search. I was on stage here doing a little ad lib before Daniel Deniz came on, and I was asked the question how AI is going to affect the content business, and I said bluntly, "Media as we know it will be dead in three years. " The core was, "As we know it. " Which means that AI will do a lot of the content, the superintelligent CUBE, if you will. I said that humans will still be involved in original, authentic, trusted content. And I believe that's going to be true. Of course it's going to happen, and I think that's a clear prediction. But then I said, "Okay, let's just say I wanted to rule the media world. I would need to have a stockpile of GPUs, because if everything's available, the only advantage is scale. " So the question to you is... That's a little narrative, little kind of set the table. How does a poor guy like me get access to all the big gear from you? So how does the little guy, doesn't have the big $ 100 million checks... Do I stand in line? Do I cut a deal? Because I think you're going to start to see a tsunami. Or do I have to wait for the price performance to get democratized? I mean, it's a serious question because a lot of people think, "Okay, there's an arms race, or I have to partner, or I have to buy. Maybe rent." So what's your view on that? Because I think you're going to see that demand phase two of your journey, when you guys are already, escape velocity, public and all that good stuff, winning. There's going to be that wave two of customers.

Robert Brooks IV

>> So I agree that authenticity is needed, scale is needed, taste and curation is needed. And I think that's where folks like yourself come in. You could have a whole bunch of AI slop, but they need someone to trust that's going to vet through that slop. So I still think that absolutely taste is going to be the biggest differentiator for all of this. I can tell when something's AI generated, whether it's text or image. I've been in the industry long enough. It's going to take other people to identify-

>> Boy, John's grammar got really good.

Robert Brooks IV

>> He uses a lot of em dashes.

>> It's got a dash in there. That's too good English.

Robert Brooks IV

>> But to answer your question on scale,

>> we have an inference API available today, and you don't even have to worry about the GPUs. It's all abstracted away from you. It's the lowest cost inference API in the world. It has obviously thousands of GPUs behind it, but you're computing in terms of pennies when you're leveraging that opensource inference API from Lambda. And so you can leverage that as much as you possibly need for your work.

>> So I innovate on the software side. I don't necessarily have to get a stockpile of GPUs and run the biggest monster GPU-

Robert Brooks IV

>> It's just an API call. - ... media cloud?

>> It's just an API call. Yeah. - Yeah. And do you have a

>> lot of media customers out there?

Robert Brooks IV

>> We do. Yeah, yeah. Yeah.

Robert Brooks IV

>> And it's much more in the form of generative image

>> and video today. Not necessarily tech space, because there still needs to be a source of truth associated with that. But ultimately, in your business, an API can be leveraged.

>> Yeah. I know you got to go, but real quick, edge computer vision, since you brought up vision, that's a huge generator of value, massive data corpus coming in. We're seeing software manage that, but the edge devices will be bringing in a lot. Thoughts on that, directions, where are we? Is it coming in fast and furious? Is it just developing?

Robert Brooks IV

>> Yeah. I mean, Lambda, actually, our founder and CEO sold a company to Apple before this. That company is the reason why, when you go into your photos app and you search for "dog", it shows the pictures of dog. That was the software that they were working on in 2012, leveraging local neural networks they were running on the iPhone at that time. Think of how powerful our iPhones or Androids are today. And ultimately, I think that's where the edge play is. In my mind, my M4 Mac right now is just outrageously powerful. I think the next iPhone's going to show that too.

>> All right. Robert, great to have you on. Congratulations.

>> Great to talk with Lambda. You guys are doing great.

>> Thank you, John. - Congratulations on the success.

>> There's more to come. It's the beginning.

Robert Brooks IV

>> We appreciate it. - As they say, don't fall off the

>> rocket ship, and they got a rocket ship key.

Robert Brooks IV

>> Hold on tight. All right. This is the enablement of the big cloud providers, the big hyperscalers, the new hyperscalers. Infrastructure in the AI side is exploding in value, enabling use cases. And that data layer comes in, software, inference APIs. As the builders adjust their tools, you're going to see a lot more value. That's a big story coming here out of Paris on theCUBE. I'm John Furrier. Thanks for watching.