theCUBE + NYSE Wired: Mixture of Experts Series | Robert Brooks, Lambda

Robert Brooks, Lambda

Join John Furrier of theCUBE in a captivating discussion with Robert Brooks IV, a founding team member and VP of Revenue at Lambda. In this insightful conversation, Brooks shares his deep expertise in AI infrastructure and the pivotal role of open-source initiatives in driving innovation. Conducted by theCUBE Research and analysts, this interview delves into Lambda's strategic position in the AI space, backed by substantial financing and the burgeoning demand for specialized infrastructure. Key takeaways from the interview reveal Lambda's approach to AI challenges and solutions, emphasizing speed, scalability, and the removal of traditional barriers for developers. According to Brooks, the company’s unique position allows it to offer a comprehensive AI infrastructure that meets diverse enterprise needs without the common hurdles. He highlights how Lambda's strategic partnerships and deep involvement in open-source projects propel their collective forward, with ramifications for broader market trends and technological advancement. Follow theCUBE's wall-to-wall event coverage at https://siliconangle.com/events/ and learn about the latest theCUBE events at https://www.thecube.net/. Find more SiliconANGLE news and analysis at https://siliconangle.com/. #CyberResiliencySummit #Microsoft #AWS #Cybersecurity #AI 00:00 - Intro 00:01 - Unveiling Lambda: Growth, Innovation, and Market Dynamics 03:00 - Strengthening Market Leadership through Innovation and Infrastructure 05:23 - Leveraging Lambda: Navigating the Evolution of Cloud Development 09:47 - Enterprise Challenges in AI Deployment 12:04 - Addressing High-Powered Computing Needs 15:24 - Impact of Open Source on AI Development 17:43 - New Kind of AI Workloads 20:19 - Deep Customization and Partnerships 23:07 - Strategic Expansion and Growth Initiatives 25:59 - Reflections on Growth and Competition 28:54 - Conclusion

Share this session

Clips
More from theCUBE + NYSE Wired: Mixture of Experts Series

Robert Brooks IV

Founding Team & VP, Revenue

Lambda

play_circle_outline Importance of fast, scalable infrastructure for AI

play_circle_outline Lambda's solution for simplifying AI infrastructure

play_circle_outline Lambda's Abstraction & AI Workload Challenges: Ease of Use for Developers and Setting Up Hurdles

play_circle_outline Differentiation in the market with no rate limiting

play_circle_outline Experimentation with different types of workloads in AI

play_circle_outline Unlocking AI Potential: Economic Advantages of Open Source Models for Users and Companies

Info
Transcript

Robert Brooks, Lambda

Robert Brooks IV

Founding Team & VP, Revenue Lambda

In this theCUBE + NYSE Wired: Mixture of Experts interview, Robert Brooks IV, founding team member at Lambda, joins theCUBE’s John Furrier to unpack the realities of scaling AI infrastructure as enterprise demand surges. Brooks details Lambda’s $480M Series D equity round (taking total funding to “over $800M”), participation from investors including NVIDIA and why capital intensity, power density and liquid cooling (50–150 kW per rack) are redefining data center strategy. He shares how Lambda abstracts DevOps for math-first ML teams with a plug-and-play stack... Read more

explore Keep Exploring

What recent developments can you share about Lambda and its impact on AI infrastructure? add

What advancements has Lambda made to simplify the setup and use of computational resources for AI developers? add

What are the advantages of using Lambda's platform for machine learning compared to other platforms? add

What distinguishes this inference platform from its competitors in terms of user scalability and access? add

What activities and projects is Lambda currently undertaking to enhance their platform and support developers? add

What are the advantages of using open-source models in AI development compared to proprietary models? add

bolt Powered by CUBE AI

Robert Brooks, Lambda

search

>> Hello, everyone. Welcome to this special CUBE conversation. I'm John Furrier, your host of theCUBE here in our Palo Alto Studios. Got a great interview here with Robert Brooks IV, the founding team member of Lambda. AI infrastructure is super hot right now. As we all know, we need to get faster, more power, literally and figuratively, to run the software. Open source is driving a lot of change, certainly on the application side. Agents are on the doorstep waiting to come in at a massive clip. We've been covering some of the early wins. Robert, thanks for coming on theCUBE. Really appreciate it. Got some big news. You guys got some big fat financing. Thanks for coming on.

Robert Brooks IV

>> Thank you so much for having me, John. Big fan of the show and super happy to be here today.

>> Yeah, we love what you guys are doing. We've been covering Lambda for a while. You guys have been around since 2012, original Lambda Labs, but you couldn't ask for a better tailwind with NVIDIA's growth, GTC's coming right around the corner. The some demand for infrastructure and AI infrastructure specifically is massive. You guys have some news in the past week or so. You guys have just raised a big series D equity financing. Give us the news. What's the number? How much did you guys raise and what's the momentum look like?

Robert Brooks IV

>> Yeah, so we're super happy to have closed our series D with a $480 million equity raise. Definitely had strong participation across a broad amount of investors. In no certain order, Andra Capital, SWG, Andre Carpathi contributed to the round, G Squared, Air Street Capital, NVIDIA, so we're really fortunate to be in this position and have the armchairs ready to really dominate the market.

>> What's the total raise now for you guys? You guys got to be close to a billion in raise. What's the number?

Robert Brooks IV

>> It is getting close. I don't know the exact figure, but it's definitely over 800 million in aggregate.

>> That's awesome. It takes a village, as they say, certainly the application side of the business with agents and AI specifically seeing massive enthusiasm. Confidence though is just now getting there. You're starting to see people knock down some use cases. You're starting to see demand for new kinds of systems that don't look like the old infrastructure and I think this is one of the keys that we're seeing certainly in the general, I say, cloud market hyperscaler startups and growing companies, but also in the enterprise. One, it's hard to get gear. Two, the systems look differently and I think DeepSeek was equivalent to the ChatGPT moment. If you had to pick some moments in this AI wave, ChatGPT couple of years ago clearly woke everyone up. DeepSeek wakes everyone up on the internet. Hey, you can do some clever things with software configurations if you design around the constraints and there are many. This is a big part of what you guys are part of. Could you share what's going on in the market because we're waiting. Come on guys. Go faster, more scale.

Robert Brooks IV

>> Yeah, so you'd mentioned in the introduction that Lambda had been around since 2012. We've been in a position where we can sort of see where the puck is going to and skate towards it and ultimately six months before ChatGPT launched in November 2022, it was when we turned on our cloud and I had to basically beg and convince people to take an A100 back then. Ever since that launch, it's been quite the opposite. And moreover, with DeepSeek and what they were able to achieve by showing sort of test time compute and reasoning at an open source level, well, Lambda six months before that launch their inference API, so we feel very fortunate to be in this space and be ready for the next wave. Ultimately, to answer your question, where we see enterprises going and really sort of the most forward facing AI labs is starting to use this infrastructure in Lambda Cloud to be able to not only just scale their training workloads, but also scale their production workloads. And that's really the next chapter for Lambda. We can full service an entire ML team with all the compute that they need in the full training cycle and machine learning life cycle of their applications and ultimately we can grow with their business as well as they continue to get more users onto their platform.

>> Robert, talk about the trend you guys are riding on right now. As Dave Vellante says, the trend is your friend. Certainly the AI is a big trend for you. You guys talk about speed and scalability. It's a big part of the narrative, but speed's a double-edged sword. Speed, obviously processing power, more capability, but also speed to deploy. Developers and end users want to go faster, number one. Number two, they can't get the gear in some cases or two set up the environments properly to even run the workloads because you got the system design thinking around the platform and then you got to have configuration workload testing, not trivial. Could you share some of the constraints you're seeing that you guys fill the gap for. Lambda Cloud certainly has got great uptake. Revenue's good, you guys are doing great. What's the reason why? What is the key driver for your business?

Robert Brooks IV

>> Yeah, so I really love this question because a lot of people think of an AI developer and a machine learning engineer is predominantly a software engineer and that's actually not the truth. The truth is they're an expert in math and so they want the DevOps and software engineering abstracted away from them so they could focus on building and conferencing models. And Lambda's understood this for a decade now, so any sort of Lambda computation that you purchased from us, whether it's a workstation or a cloud instance, has this plug and play experience where you have all the sort of softwares, libraries, and frameworks and pre-configured for you. So again, you can focus on the workload and not actually setting it up and then ultimately we're in a position today to bring that to the next level. So last year we launched one-click cluster. This allows you to get hundreds of GPUs all clustered with NVIDIA's InfiniBand, which is the fastest networking in a click of a button and use it for as little as two weeks. This is a democratization of compute that allows AI labs and enterprises to move faster. They can do a POC, then they can scale, and so we're in a world in which we're trying to bring down the barrier of entry to these large computations in data centers, which by the way, there's a million things concurrently happening when you spin up a GPU cluster and you need not one of them to go wrong and this is where Lambda excels. We give you this sort of uber experience of, hey, the black car just shows up. You just get in and you're right at the restaurant. You don't have to think about it. That's what we're trying to deliver to the AI developer market.

>> I love the Uber example. It's like get Uber and get Uber Eats. Uber Eats is kind of like if I have Uber, I use Uber Eats, but Uber, they've got XLs, they got Uber X, Uber Comfort, black car, so variety of different kind of configurations. You mentioned the enterprise and I love that abstraction run software because I think you're going to see an in migration of new kinds of software developers, you're going to have the democratization obviously seeing that with open source, but all the I call the Alpha software developers are going to go down to the hardware level. You're starting to see that now. Talk about the importance and if you could scope the alternative to not having a Lambda Cloud. Because from what I can see and from our research, it's get it right in the system and then you've got to run software, which is a whole nother thing. Yeah, I've been using machine learning for some use cases like fraud detection. That's been around for a while, but now you've got generative AI, a whole nother thing. What is the alternative? What are people missing? What's the shark fin or the iceberg? What's under the water, if you will, on the complexity and the configuration? All the hassles involved in an enterprise or a large scale provider who's got a workload that's demanding?

Robert Brooks IV

>> Yeah, that's a great question. So ultimately there is an inflection point with any sort of AI team and really for enterprise, it's actually a lot more broad in the sense of, okay, we found the use case. We want to go ahead and test out and get the data to validate that we actually need to spend money here and then we need to spend a lot of money. And that's the way the enterprises think, right? They want to see some results before they invest and we put that platform in place that allows them to go ahead and tinker without blowing their budget and knowing that they can scale on our platform. So to your point about the shark fin out of the water, which I really like, because I'm scared as hell of sharks. There's a world out there in which you can grab GPUs from other platforms and they do promise a platform that allows you to scale, but ultimately you get in the environment, it's bare metal, the software is clunky, it doesn't allow you to scale. Maybe they're only focused on training, you have to install your own Kubernetes platform or some platform and you're not in a world in which you can actually build out and go to production from training. You're sort of stuck and you have to piecemeal between different vendors. Lambda is trying to get you into a place where you don't have to think about that. You can be in a one-stop shop and ultimately if you look at the interfaces for some of these clouds, they've been around for decades. They predominantly serve software engineers first, and again, we all know the profile is not predominantly a software engineer in the ML world, and so these interfaces have several sub menus, all these different sort of quotas that you have to meet. You have to talk to a human being. With Lambda, we're putting the engineer in the power and we're allowing them to choose how they want to scale on our platform without any of the barriers to entry.

>> I love the example of how you guys are targeting that persona. If you go back 15 years ago when we started theCUBE, obviously I call it the , he kind of understood the cloud. Cloud was pretty straightforward. If you were born in the cloud, again, a bunch of no-name brands started in the cloud, Airbnb, you name it. Now they're huge brands, right? They're dominating. The whole value proposition was provision stand-up infrastructure without provisioning a data center. Okay. That was a great wave. Now we're in this next AI wave where the challenges are similar. I don't want to have to stand up all this stuff and provision it if I don't have to. So is that where you guys see value? Because democratization aside, totally agree. Devs are coming in, open source is driving it, but enterprises have pre-existing stuff too, so they have to deal with startups And they don't have any money. So though I'm blowing your budget, why would I want to get all this gear to run a POC with five startups? Talk about that dynamic of standing up fast because I think this is where I'm seeing movement in the market where the pretenders are not winning because it's harder. If it's hard, if it creates friction, it doesn't work. What's your thoughts?

Robert Brooks IV

>> So it's an incredible question, John. The reason why is the world up until today was predominantly in air-cooled servers and you could get away with a server that ran between six to 10 kilowatts each. So these enterprises that had these older data centers built out for the web and mobile scale, they could get away with their five kilowatt per rack, maybe hosting one AI server and stretching that across a large aisle. That doesn't exist today. Next wave in video architecture is predominantly liquid cool, and you're looking at kilowatt per rack going in sort of the 50 to 150 kilowatt per rack, and there's not many enterprises or companies that have that just sitting there. In fact, it's predominantly owned and rented out right now by Buzz Clouds. And so that expertise as it relates to not only just standing the hardware up but making sure that it's properly cooled is an even further complexity that we're having to manage in the space that an enterprise ultimately just does not want to deal with. I think that's a huge part of it. Then there's world of Slurm, which is sort of an HPC type orchestration platform that's still really good for ML today, but Kubernetes is also something that even some of the best and biggest AI labs in the world still need their hand held on, and this is where Lambda really helps. We abstract away the water cooling. We give you a managed Kubernetes or hosted Kubernetes platform that allows you to just do your workload and these things are extremely complex. We have hundreds of engineers on both sides of the equation solving that for you.

>> Yeah, you bring up the Kubernetes. That's good point. That's a trigger word for us. Obviously, every Kube constants was founded. Talk about that because I think this kind of ties together I think the rationale between why you're winning, which is the platform engineering cloud native community, it's not that obvious to the naked eye that hey, that community's intersecting with the ML folks. Because if you think about it, that platform engineering term is kind of what you've done, right? You're basically platform engineering as a service for ML and gen AI engineers. Is that right?

Robert Brooks IV

>> That's 100% right. Actually, our goal is to be one of the biggest users of our deployments eventually. So to just back up to how we started, we're started by AI researchers and ML engineers. We're published at the largest AI conferences in the world. We actually are performing these workloads on our own infrastructure, so we're solving for our own problem. That allows us to pick the best tools and the best versioning of the tools that we would want from experiencing a cloud and then give that to our user base. I don't know if there's any other problem in the world that can say that.

>> Yeah. It's interesting. Not to go on a tangent here, but you just highlighted to me what we've been saying on theCUBE, and you kind of said it perfectly, which is every big wave, every inflection point, the winners kind of came at it from a different angle. They didn't really say, "Hey, that's a hot market. We should build stuff for machine learning." They were. They had the problem. So you look at all the winners, go back 30 years, even in the computer industry, all the ones that were the winners essentially kind of came out of left field. They weren't obvious at that time and do it. And then they are the problem. They are the solution, and then they provide that service. AWS was formed out of Amazon doing data center for themselves. Why don't we offer service?

Robert Brooks IV

>> Exactly.

>> And same with theCUBE. We had a video problem, we built our own platform, similar kind of things, and you're seeing some of the starts. Even in the cloud wave, they were solving something that wasn't obvious. Airbnb even was passed over because their original business plan wasn't what it was. Talk about why that's important now because a lot of the creativity that you're seeing in the dev side on the app side and the workloads are either being retrofitted for AI or net new, and that's why the inference, the reasoning, the open source and the deep sea kind of play around that. Talk about that, because again, that's not obvious either. Talk about that dynamic of these new kinds of workloads. You're solve the problem for yourself now that problem is becoming obvious for the customer. What is that problem? I mean, is it the wave of software? Is it the net new? Talk about that.

Robert Brooks IV

>> Well, I think there's limitations in the way that companies offer compute and API services. So an example with our inference platform is we compete in the market, but we're the only inference provider that has something called no rate limiting. So essentially you're free to use our platform and scale up as much as you possibly want on the tens of thousands of GPUs that we have available. Whereas other platforms, they want you to enter their sales process, they want you to hit a rate limit, they want you to be frustrated by it, they want you to do an enterprise contract. Of course, we eventually want those things too, but we're trying to let the developer be in control, and so that's one very, very small nuance that our community loves. They're like head over heels in love with the platform and the thinking that, hey, I can be in control and I can scale out. In terms of workloads, not only are we playing with text models like Deep Seek and all that sort of stuff, we have a video leaderboard that we made very prominent within the community where we're constantly testing these things out and giving that information back to developers. And if you stop by Lambda's office, John, which you absolutely should in San Jose, we are currently programming a robot, a humanoid that's walking around. We're getting familiar with those interactions and we're trying to understand what those particular type of developers need from our platform. So when you're in the weeds, when you're playing with the tools and you're getting familiar with the platform or the problems firsthand, that allows you to solve for it much faster in a more eloquent way.

>> And you guys just getting started as Dave and I always say, what inning are we in? The game hasn't started yet. It's like spring training in AI. It hasn't even gotten going. I think the key is accelerating more value experimentation. This is why I think the open source angle and the inference around inference and reasoning are important because I was just watching a video this morning from MWC and some of the mainstream conversations are, hey, it's not going to be hundreds a months, it's going to be thousands a month. Well, we knew that years ago, so we saw that long tail coming. Talk about the importance of open source. Obviously DeepSeek is open sourcing and their stuff even though it's from China, but still there's so much good stuff going on in open source that has certainly activated the developers and that's just getting started. And you see the hype on agents as a tell sign that those apps are going to come in and want some workloads and you've got the reasoning get reinforced learning. You're starting to see the application side being enabled by this machine learning data layer that requires huge computation and agility, but also scale, right? So this is again open source reasoning. Why are these important? How do you guys see people using the models, not just LLMs. You got computer vision is going to be a big use case. Obviously it's going to power a lot of GPUs and of course NVIDIA is rolling out more gear too. What's your take on this? What's the commentary on the open source impact and how these models will start to route to each other, reason with each other, integrate with each other? How do you guys see all that playing out?

Robert Brooks IV

>> Yeah, of course. So I'll start by just talking a little bit about how Lambda has been involved in open source and then kind of get to that question. So our Lambda stack platform, which is really that plug and play experience that I was talking about before. That's all open source. We have multiple pull requests that are accepted in the CUDA, CUDA NN, NCCL libraries for NVIDIA, even the Linux kernel. So we're very involved in this particular community. What we've seen from Metaslamo models as well as the recent DeepSeek model is the ability to catch up to the proprietary labs. I think what you're seeing with GPT-4.5, although it's actually a really cool personal model, there is potentially a limit to just raw scaling compute and these things like test time and inference time compute are ultimately where the next wave is going. And for the user, there's several benefits that transcend also to the company as well, but being able to actually see the reasoning, how the model's thinking, being able to diagnose it, there's this concept of the high testers versus the low testers of these models. And essentially what that means is you're either technical or you're not when you're using these models. And if you're technical and you're using this to build an application or build a video game, I mean, over the weekend a very prominent developer in Twitter built a flight simulator literally by using Claude. They can get so much more value out of these types of models that are open source and they can actually see the reasoning and be able to scale out and not have to pay a pretty buck to do this. So we have something written out in our series D-rays where if you want to scale up training 10X, then you're spending billions of dollars. If you want to scale up conferencing in terms of definitely the reasoning models and test time compute, you're spending 13 more cents per token. So the economics are really starting to favor the broader AI developer community, and really that's attributed to open source predominantly. Yeah.

>> Great point on the economics by the way, that's huge. Because once you get the training, it's like going to school. Once you're in school and you graduate, you don't go back to school unless you get your master's degree or PhD. Why would I want to go back to the fourth grade? So once you get the training done, you infer in the real world, I think you're seeing that with AI. Brian Baumann and I always talk when we're at the NYSC, Brian Baumann from NYSC Wired, we've been observing through our AI leaders series, we've been talking to folks on, and I want to get your thoughts on this too because I think it's something that we've been seeing a lot more of. A lot of the people are going down to the hardware. So you mentioned the costs are getting better, obviously per token on inference and the general app, but then the developers are squeezing more and more out of the infrastructure. It reminds me of the nineties when I broke into the industry. When you deal with memory, memory management, the basic stuff around computing, again, different system model there and not as complex. What are you seeing around the customers and their engineers, are they getting down and dirty in the low level piece of it? What do you guys say to that piece of the market? Do you guys see that as well? Is that where the APIs are being used? What's your relationship to the developer once they get their hands on a cluster? What are they doing?

Robert Brooks IV

>> Yeah, so we're actually going to be announcing a couple partnerships with enterprises. They're going to be hosting their open source models on our platform, which is our Lambda inference API. And that's a newer trend. So you've seen that obviously from predominant labs that have big AI research teams, but you're starting to see that focus, which is a new trend and something we're excited about. In terms of your question related to actually being able to control it down to the hardware. So we have really three pillars that we think about things at Lambda that's really new for us with this Series D-Race. We were always an AI infrastructure company, but today we're really an AI infrastructure company and AI conferencing API company and an AI chat company. And so we run something called Lambda Chat and you could go onto your browser Lambda.chat, there'll be an app for it soon, and we're able to basically achieve throughput metrics and cost metrics that no one else can because we actually physically run the hardware ourselves and have done all this sort of optimizations all the way down to the node and GPU level. This is something that's actually unique to a cloud provider and our ability to have all those optimizations and then pass that experience to the end user. If you go into DeepSeek's native app right now, you're constantly seeing, "Hey, we're out of GPUs, or you can only ask 10 questions a minute." That's because they're out of computation. And so we're in a world in which we can solve that for ourselves. And to your point, we have gone down to the hardware level to make it faster and cheaper.

>> And that's a great differentiator. Congratulations. Quick couple of questions to kind of close out. One, what are you going to use the funding for? What's the mandate? More go to market, more GPUs, buy more gear, get more data centers, get more kilowatt racks. What's the use of funds look like? Obviously you guys are well-funded, got a nice stockpile of cash, revenue's good. What's the use of funds for?

Robert Brooks IV

>> We have an office in San Jose, which is predominantly where our manufacturing supply chain and operations are associated. And then we have a new office in San Francisco that's about six months old where we're really scaling out our software engineering and go-to-market teams. So a lot of hiring on those fronts. And then to your point, it's really around the infrastructure. So NVIDIA's Blackwell GPU is really starting to roll off the line masse. We're going to have our first sort of B200 test cluster available by GTC. We're going to be one of the first platforms that's going to have that publicly available on demand for anyone to use. Actually, you can go into our website today and request a B200 cluster and you can get into the queue, which is really, really exciting. And then ultimately we're in a position where we are a capital-intensive market, and so being able to have megawatts not only in the next year but three years is strategically important for our business. So we're breaking ground, signing contracts, doing joint ventures in that sort of 50 to multiple hundred megawatt scale that allows us to be ready for even the next iteration of computation.

>> What's going on at GTC? A couple of weeks, we're going to be down in San Jose. It should be massive. Again, it's like our Super Bowl event. Supercomputing and GTC have become my favorite shows. I saw four years ago, supercomputing was clearly an HPC show and I'm like, what are all these rooms on the side over here? All the cloud, all hiring all these people I knew, like all the smart people were over here. It was clearly AI coming in. Now it's full-blown AI, GTC. Any news there? Can you share a little bit of taste of what's coming out for GTC?

Robert Brooks IV

>> In terms of Lambda's plans, absolutely. We're going to have our robot at our booth, so I encourage you to come by. It'll be able to pass you some Lambda swag. We're going to be sponsoring a coffee bar. We're going to be wearing track suits. We're going to be doing a lot of that stuff. But ultimately the B200 test cluster that I alluded to, that's going to be something that we're going to be talking a lot about that week. Certainly we're all excited to see Jensen go on stage and talk about the next wave and what's been happening in the industry. To your point though, John, it is just a fantastic time for all of us to get together. Lambda is hosting a bunch of VIPs that are leaders in machine learning across the space. We're be doing a little bit of a keynote from our CEO as well. So it's a great time to be in the space and I think GTC is one of those pivotal moments for us all to be together.

>> It really has been an incredible run, been loving the NVIDIA action. Final question for you guys is obviously we're early days. I call it spring training. The game hasn't even started yet. It's going to get better. You guys aren't going to stop either. What have you guys learned over the past few years looking back, kind of where you guys came into the market, it developed really, really fast. What are the key learnings? And then the second question, part of that question is, what do you guys do that's different than the competition? Can you share why you're winning, where customers need to squint through the requirements? How do you know Lambda versus other? So talk about what you guys have learned and then competition.

Robert Brooks IV

>> Yeah, of course. So my answer actually potentially is for both questions. So one of the biggest learnings we've had is how much we need to say no. And I think Steve Jobs sort of famously talked about this, but ultimately we set out to build a platform that strictly serves one customer profile. There's actually not a lot of companies that do that. They want to horizontally expand, especially for something like a GPU that can serve CAD workloads, VFX workloads, rendering workloads, crypto workloads, and AI. A lot of our competitors have chosen all of the above to serve, and that ultimately leads to a much more complex software stack, a much bigger support staff, and it means that you're not focused. And that has allowed us to actually move faster, because again, we're serving ourselves, we're serving one profile, and we intuitively understand what that means. In terms of differentiation, if we can cover the entire development lifecycle of a machine learning team with our platform, which we are at today, this is something that no other platform has. They say they have it, maybe it's a bullet point on their website, but ultimately we're living it in the sense that you can actually experience this today with signing up with a credit card and under five minutes you have access to a full training platform, full inference platform, and you're ready to go.

>> That's smart. I got to say, our research is showing we're going to come this out probably in the next couple of weeks. The enterprise production kind of window, if you will, or road to production is narrow. I mean, it's not obvious. I mean, certainly all production workloads aren't equal. Some are mission-critical, some aren't. Marketing market tech stack, obviously easy to get. Nice chat app for that. Rag is hot, okay, data's out there. But if you look at the top enterprises where the bar is high for production, it really is on gen AI. There's been machine learning workloads in production, right? So smart to target the machine learning piece because that enables everything because you got to deal with the data. Was that kind of by design or is that just you were there already or do you agree? First of all, do you agree? And then two, how should companies think about this? Is the ML team the gateway to Nirvana? Because I think if you look at the sequence of events, where's the resilience bar? They've heard all that already. What's your thoughts? Do you agree and what's your commentary?

Robert Brooks IV

>> Absolutely agree, John. And then I actually like to tell this story a lot internally. So again, we've been around for a while. So one of our first enterprise customers was actually using machine learning, not generative AI, but machine learning in material science to develop something related to the energy sector that allowed them to increase efficiency of a particular battery cell way over 50%. That was using machine learning. Matrix multiplication, neural networks, bespoke type stuff. And we still obviously have those customers today, but they don't necessarily need of GPUs to get that running. Generative AI is really where you start to see the scale come through. And again, that's been a boom for our particular business as there's customers that don't think in terms of hundreds of GPUs, they think in terms of thousands of GPUs. So for us, that's been the big change. But to your point, we still very much serve the machine learning market and it's still doing things that probably affect our lives way more than we even know in terms of the efficiencies related to material science, chemistry, drug discovery, and not being able to produce an avocado chair when you put it into the sex test. So we're excited about that.

>> Yeah, it's all good. I think the Gen AI is going to be the general purpose app dev market, and that hardened top, if you will, is the platform engineering. That's where the hard work's done. We've seen this before in other waves. Robert, thanks so much and congratulations on the funding to your team and pass along our congrats from theCUBE and SiliconANGLE. And again, thanks for coming on theCUBE.

Robert Brooks IV

>> Thank you so much, John. This is an awesome conversation. We appreciate the love and looking forward to being on here again.

>> It's easy when you're doing great work. Robert Brooks, the fourth founding team member, also VP of Revenant Lambda. Again, solving the hard problems is what has to happen in the industry. If you look at where we are right now, AI infrastructure has to get it right. That enables the next wave, and everyone's waiting on deck to come in, and that's going to be developers and agents. App dev will explode. That's coming next. Thanks for watching theCUBE. I'm John Furrier, your host.