Basil Chatha: I don't think we're in a bubble because I I don't think we've really discovered the full value potential of AI. What if we built one big chip and then left it? This is the chip. This is the wafer scale engine. Nearly one million cores. They're gonna buy 750 megawatts of compute by Cerebrus to power their applications. The coding model that uses Cerebrus now is Codex Spark. Inference is where you ring the cash register, medium speed tokens at really low cost. But that's not a market I'm interested in pursuing. The market for fast tokens for the future of AI applications that are at a competitive cost, that's what I want to go hunt and win. I know Dario and the team really well. Extraordinary team, extraordinary technology. I hope we get the opportunity to do more. Most computer chips are smaller than a postage stamp, smaller than a bottle cap. Cerebrus builds one the size of a dinner plate. In this episode, I chat with Andy Hawk, who's currently the chief strategy officer at Cerebrus, which just IPO'd and secured a 750 megawatt compute contract from OpenAI. Andy makes his case that the trillions of dollars pouring into AI infrastructure isn't a bubble. He explains why memory stocks are ripping, why DPU prices have to follow, and why their chip actually benefits from rising memory prices. He tells me why the stat that ninety five percent of enterprise AI projects fail is measuring the wrong thing, and he talks about the US selling chips to China and the impossible job of the people who make that decision. Let's get into it. So Andy, you are the Chief Strategy Officer at Cerebris. Can you tell us about what that means and what Cerebrus is? Absolutely. ⁓ It's a pleasure to be here, Basil. Thank you. So Cerebrus, for for those members of your audience that aren't familiar, is an AI computer systems company. We observed nearly 10 years ago that AI had transformative potential, ⁓ not just for things like cat videos and consumer applications. But also for fundamental science, industry, society. So we saw this big potential, right? And we can go into this more a little bit later. ⁓ what we saw limiting that potential was the speed of computation. Took too long to build models. Once you had a capable model, it took too long to deliver to deliver answers. ⁓ at Cerebrus, we developed a new chip and a new system that were specifically designed for artificial intelligence compute. ⁓ fast forward to today. ⁓ those systems are delivering AI inference and training ⁓ more than 10x faster than legacy general purpose processors like GPUs. and enterprises are starting to come to us to work with us because AI has moved from being a curiosity to being existential. ⁓ and once a capability like this is existential, everybody wants to be fast. ⁓ and that's where we live in the market. That's what we do. In my role, I'm our chief strategy officer. ⁓ I started with the company nearly 10 years ago and ⁓ actually started as our first head of product. So working directly with our engineering teams, our founders, our customers to help define what we should build in terms of the hardware where it sits in a data center, the software stack that meets users where they are, and work with the team to build towards those market and customer requirements. a a computing solution, hardware and software stack, ⁓ that would deliver on that promise of accelerated compute for AI. ⁓ These days, under the strategy banner, my charter includes product strategy, so thinking about where we want to build into the future, but also many of our largest strategic customer engagements. So think US government, think ⁓ large commercial and international like G42 and the UAE. Our recently announced partnerships with OpenAI and AWS. ⁓ the kinds of big partnerships ⁓ and big technology moves that are gonna carry us from our first decade into our second and beyond. So when Cerebrus started, I'm assuming that was before LLMs. Pre L O ⁓ Pre L O So what were you building towards back then? Like what was you know, what were you thinking was going to be your customer or who I love that question. I mean, I think for many of your ⁓ your your your viewers, your listeners, this is gonna take us back in time, right? So let's let's put ourselves in 2017, 2018. The canonical AI problem of that era was training convolutional neural nets like ResNet for image classification and object detection problems. So image net, right? ⁓ find a cat. Hot dog, no hot dog. A hundred percent hot dog, no hot dog. ⁓ so that you know, that was the problem. So it was training comnets for image problems. ⁓ and and at that time, that was the canonical problem of the day. But we and so we we thought about that as an an archetypal problem to inform our hardware and software development, but we also knew that we didn't want to be limited about that, because by that rather. Because in in some sense, we had a we had a hunch, a hypothesis that AI was going to be transformative, as I mentioned before, but we didn't know exactly how. And so we'll get into the the the chip and what makes it special, I think a little bit later. But long story short, we didn't design the chip ⁓ at the heart of our systems specifically for training or for inference. And we didn't design it specifically for ⁓ convolutional neural networks or transformer networks. We we designed our chip at the heart of our system to be a a ⁓ general purpose accelerator for all of AI compute. And I I don't think we we really articulated it this way at the time, but I think what that bought us as we then move from twenty seventeen to twenty twenty six and present day, what that bought us is durability and flexibility to accommodate large language models, inference as well as training. ⁓ and and be able to accelerate all those workloads by building a general purpose AI accelerator built around the first principles of all AI compute, rather than building a a truly sort of problem-specific chip that would just be good at one thing. Yeah. So when so I guess was GPT-2 on your radar ⁓ back when it came out? I'm assuming so. And if so ⁓ did you have to change anything about the chips you guys were building for hey, like LLMs are going to be huge over the next few years? ⁓ so yeah. Yeah, great question. This is I mean, this is the fun part about this history. I think when when you're building a a fundamental technology that is a a new technology that's sort of at the base of the stack, I think you have to acknowledge, particularly in the hardware world, that it's gonna take some time. ⁓ and GPT two, sure yes, when it came out. it was definitely on our radar. But I think as you just articulated the question, I'm not sure if we really knew at that time that that like LLMs were going to be what they are today. Right. We knew that it was a it was a really compelling answer to ⁓ I think the fundamental value question of AI. Because that was the moment in AI where AI went from being sort of an interesting curiosity to being demonstrably valuable, right? Was was that moment, the sort of chat GPT moment. Yeah. ⁓ But we just we still didn't know that it was gonna that that architecture was going to endure ⁓ and that it was gonna be as large in the market as it is today. ⁓ so I I'd love to say we did. But I don't think our crystal ball is is is is particularly better than anybody else's. ⁓ I I think we just built a technology intentionally that could be durable and flexible to changes like we've seen in the field over the past decade. ⁓ Okay, so then to your question, did we have to change anything? The short answer is yes, but not in the hardware itself. So the the chip, the heart of our system, has remained the same throughout, through throughout all the way, from you know training conv nets for image problems to inferencing language models for you know coding and agentic and generative AI. ⁓ the chip at the heart of the machine has stayed the same. Generation over generation there's been improvements, but the fundamental architecture has stayed the same. ⁓ at the software level though. That's where we were able to pivot and respond to to the market. So our original systems were ⁓ were were delivered with a software stack that was optimized for training, ⁓ and that had a ⁓ a lower level library of compute primitives oriented towards ConvNets, right? That makes sense. That was the market of the day. That's what we were delivering and building for. ⁓ at that time, we were also integrated with the ML framework TensorFlow. To sort of go back in time, shout out to Google and the TensorFlow team before PyTorch really came into the market and became the de facto ML framework for ML model programming. ⁓ As the mod as the market shifted then from training and commets to ⁓ to language models, still training, language models then got big, right? You might remember sort of the transition from ⁓ from from BERT. to larger language models, right? Language models started to get beyond, you know, one or ten billion parameters and a hundred billion parameter territory. And Bert was like two hundred million or something? I if I recall correctly, yeah. Something like that. So it's still modest size by by today's standards. Right. Right. Now yeah, but a lot bigger. Yeah. ⁓ there's still a lot that can be done with those those I'll call them medium size models ⁓ and and smaller. But yeah, once training became fr went from images, image sorry, convolutional models to ⁓ transformers and language models, and then from I call them small to medium language models to large language models. We had to update our software library and then we actually had to change the way software used our system to train those much larger models. So we we basically did a ⁓ a a a significant revision on our compiler and software library to be able to train those very, very large models. Not because they were language models particularly, but because they were large. Yeah. ⁓ and then when, you know, in that post-GPT moment, ⁓ when the market became interested in inference, the same thing happened again. So we we we recognized there was there was an opportunity to accelerate inference. ⁓ but our software needed to use this engine a little bit differently than it had for large training problems. I see. And so ⁓ How do the machine stays the same? The software around it and and how we present the problem or how we compile the problem to that engine has has has changed. Yeah. Could you maybe now give us a one on one lesson on the chip and just like how does it work? ⁓ how did you know how did GPUs work? How do they differ? That sort of thing. Sure. That's a l that's that that's a long that's a long question. Okay, you can you you rein me in and I'll ⁓ and I'll start. So at the beginning recall that I mentioned We weren't building this device specifically for training or inference or commets or transformers, but that we took a step back and looked at the first principles of AI compute. So if you ask yourself then what like what are those first principles of AI compute, obviously you need a lot of computing operations, right? You need a lot of flops. ⁓ but not just any flops. AI compute tends to want ⁓ lower precision, ⁓ like 32, 16, maybe eight. 4-bit. ⁓ Also, it needs the ability to handle sparse operations very efficiently because there's a lot of zero-value data in the ⁓ input data and the model weights of AI problems. So you need a lot of compute, but you want it to be optimized for sparse linear algebra and relatively low precision. Cool. Lots of compute. Check. The second element that you need is ⁓ fast communication between computing elements, because AI is fundamentally a problem of handling data in motion. In the ⁓ sort of in in the analogy of a multi-layer neural network, the computing element that's computing on one layer needs to be able to quickly share the output to the next layer in order for the whole thing to go fast. So computing elements need to talk to each other at high bandwidth. So you need high bandwidth communication check. Last thing is you also need high bandwidth memory. ⁓ and I don't mean that in the capital HBM sense, but you need memory that can be accessed very, very quickly by those compute elements. Because you want your compute to be able to quickly pull up the ⁓ the the stored weights of the model and then combine that with inbound activations, push it out to the next computing element in the chain, whether that's training or inference, condom nets or transformers, all those things hold true. Okay, so then you say I want a computer that has a lot of sparse linear algebra flops, high communication bandwidth and a high memory bandwidth. And you look out at the computing landscape of you know 2015, 2016, 2017. ⁓ and you have X eighty six machines, CPUs, and and GPUs, the predominant compute engines. And and and they have a lot of flops. So they can bring the compute to bear. Maybe they have more high precision or whatever, but like they can bring the flops. But they're challenged, those architectures are challenged by the latter two attributes that we talked about. High communication bandwidth. They both are what are known as traditional von Neumann architectures. ⁓ so they have a a compute substrate, a logic die, and then some memory. ⁓ GPUs, for example, use capital HBM, HBM, memory, DRAM off the chip. ⁓ so there those are the memory is separated. And then if you want your processors to talk to each other, then you're wiring your processors back together in a in a cluster, right? ⁓ over copper, through a network interconnect, over fiber, across the aisle in a data center to like some other machine through another router, right? Like ⁓ so the the communication memory bandwidth is is really a challenge. So then we say, well, well. Well look, what if we just put as much compute as we could together on a chip, on a big chip, and then allowed the compute elements to talk to each other over silicon so we wouldn't have to wire a bunch of small chips together. ⁓ and we can if we have a big enough chip, we can also put a bunch of memory right on the chip. And then we don't have to reach out to some external or or separate source of memory for memory bandwidth intensive operations. We said, okay, so that would how big of a chip can we build? And then you you learn that the the way every chip starts is there's a three hundred millimeter circular ⁓ disc of silicon and use photolithography to basically ⁓ snapshot out or or stamp out circuitry on that chip. And then traditionally what happens after you have that full disc printed out is you then slice it up into lots of small chips and then Everything we talked about happen before happens. That is, people spend ⁓ their their entire careers and industries figuring out how to stitch those chips back together and get them to all work on big tasks. So we said, well, what if we just didn't do that last part? What if we built one big chip and then left it whole? That would give us all the ingredients that we just talked about. And it would be a fundamentally better design for the first principles of AI work and allow us to achieve significant acceleration on this class of work that we think is going to be so important for the future. So that that chip, it's just a bunch of transistors, but where is the memory? Yeah. Yeah. Okay. So let's talk about the chip. So ⁓ for for for your viewers, this this is the chip. This is the wafer scale engine. I think you've seen it, but you can take a look. ⁓ this particular one happens to be our third generation design. ⁓ but the first and the second were also the same size. The the fourth is also going to be the same size. This is the biggest square you can cut out of those primary 300 millimeter ⁓ circular wafers of silicon. Also random, but why a square? Yeah, so okay, so why a square? ⁓ it turns out A, it's a lot easier to physically package. That is, once you build this thing, then you have to figure out how to deliver power to it, how to deliver data to it, and how to keep it cool. And it turns out that a square is a lot easier to package. The second reason actually is software reason, because it it's a lot easier to to to tell a compiler how to compile a workload like a neural network to a a square array rather than an array that has all these sort of tangential parts to it and that's non-U. I see, I see. Yeah. So we conceived of this design. ⁓ our original CTO, Gary Lauterbach, ⁓ Our other co-founders, Michael James, Sean Lee, JP Fricker, and Andrew Feldman, conceived of this design. They they looked at other solutions like a a bunch of small chips on a common interposer. We looked at alternatives because we knew this was going to be really hard. ⁓ but we kept coming back to this because of its ⁓ performance attributes on the work. ⁓ and then we worked very, very closely with with TSMC, who's our fabrication partner, Taiwan. ⁓ to basically develop the fab technologies and the methodology needed to instead of dicing this up into a bunch of small chips, keep it as one. ⁓ and so this is ⁓ then the wafer scale engine three and in today's systems it has nearly one million cores all on one device and 900,000 in our production machines. They're in a 2D grid across the entire surface of the wafer. And then All those cores are directly connecting each other and across the entire wafer over silicon. And each core has its own local bank of SRAM memory. So the memory is on the chip too, back to your question. So we we coming back to those first principles, lots of sparse linear algebra compute, high communication bandwidth, high memory bandwidth. This design checks off all those boxes ⁓ and gives us what you might think of as the a cluster worth of AI compute on a single device. Literally three orders of magnitude more communication bandwidth and memory bandwidth than is possible with traditional chips designs. And I'm sure we'll come back to it later. But it's that con ⁓ that combination of resources, particularly the memory bandwidth, that allows us to just scream on today's agentic ⁓ inference workloads. Yeah. So it seems like this chip is better in basically like every dimension. So why do we still use GPUs? You know, it's a great question. ⁓ I think there's there there there's two factors here. One is that different parts of the AI ecosystem and problem landscape need different engines under the hood. ⁓ like what people have done with GPUs in industry to date has been incredible. And there's still lots of problems that that might not require, you know, blistering fee speed, instant answers. I happen to think that the market is going into a direction where Blistering speed and instant answers are going to be the default, right? ⁓ and we can come back to that as well. ⁓ but I think the future of AI compute is is heterogeneous, right? There's there's going to be different kinds of compute engines that answer different problems, just like the compute industry today. ⁓ but I do think the market needs a a different and faster solution for AI training and inference, and that's that's what we've built here. ⁓ I think the the second thing, honestly, that was ⁓ A a challenge and opportunity for us in the beginning was just building a software stack to meet users where they were, right? This is this is a very different kind of device. We couldn't use existing compilers, existing software libraries. We had to write ⁓ our our team had to write the software to make this device easy to program and use. ⁓ and as we fast forward today, ⁓ it is, right? We we meet users exactly where they are. For training problems, you can program the machine in standard PyTorch. And for inference problems, actually the the the programming challenge gets even easier because for most inference problems, ⁓ what application developers want is just an API to some inference model under the hood. And so for inference, we just surface a a open AI standard API for inference on our machines. So the ⁓ what what was once a ⁓ sort of a a challenge for us to to build and something that made a adoption trickier has all but dissolved. And today it's easy for customers to to do either training or inference on the machine with standard framework. Yeah. You mentioned earlier that maybe a few years ago you were more focused on training. When did you have to make that transition to, ⁓ like inference is the more important problem now? Is that the case, would you say first? I think yeah, yeah, yeah. ⁓ more important is interesting. I I what I would say is inference is our focus and inference is a i is w the the primary area of growth in in the market right now. But I think actually one of the un under told stories is is the the the remaining importance of of training. The training market is still really growing. ⁓ but inference is is taking off extremely rapidly because that's where AI delivers value. Right. So as as soon as AI is valuable for a problem, then inference just takes off. Just because the demand is like insane. Yeah, I mean I th like the the way I think about it in in in normal terms is ⁓ training is a is is a cost center. That's where you build your model. That's where you teach it how you to do what you want it to do. ⁓ inference is a is a value center, right? Once the model does the thing you need it to do or the thing that's valuable for you or your end users, inference is is where you ring the cash register, right? And so ⁓ that actually goes back to your question. When did we make that shift? We made that shift to start ⁓ building the optimized software stack ⁓ and our cloud services for inference when ⁓ when when inference started to become ⁓ the the predominant problem in the field. So that was probably about eighteen to twenty four months ago. Yeah. And I'm would you say most of that usage right now is coding? ⁓ yeah. Most of the usage right now is coding. So I think the the other areas where we get a lot of usage are ⁓ Voice models and reasoning models for various chat and assistant applications. ⁓ and ⁓ timely that you're here today. We actually ⁓ just launched today a private preview of Google's Gemma 4 DeepMind model, which is a multimodal model. Open source. Yeah. ⁓ yeah. And it's ⁓ so private preview today, ⁓ and we'll come to this later too, but all the What like all the models that we bring up, it is now the the fastest frontier multimodal model on the market. ⁓ we're serving Gemma for at ⁓ 1,500 tokens per second. It's 10 times faster than any GPU implementation, ⁓ and 15 times faster than something like ⁓ Claude's haiku model. So yes, I think to where there's a massive value center for a lot of our customers today is fast coding agents. where speed equals software engineer productivity. ⁓ but I think that that will remain very likely into the future. But I also think in the future, ⁓ things like agent applications for other workloads, multimodal workloads, AI for science and security applications, right? These these are all sort of just over the horizon and I think are gonna be ⁓ really exciting areas for fast inference in the future. If you had to make a prediction on, you know, the next couple of years where the share of usage is going to shift. Do you think that coding is still going to be the biggest share in, I don't know, three years? Or do you think life sciences is going to be that bigger bigger share? Voice, whatever. So you're saying I do have to make a prediction. No, look, I am ⁓ I often joke I'm I'm I'm a recovering physicist. ⁓ I think it's a reasonable hypothesis. so much of the the the of large enterprise and consumer industry that we touch every day is digital. I think it is very likely that that will remain true. And if that remains true, ⁓ then how will we continue to update that, make it better, ⁓ develop new applications on top of that? It's gonna be coding under the hood. ⁓ and and so I think that's a I think that's a durable application. Whether it remains the predominant application in in three or five years, I just can't wait to see what people will build. So I don't want to rule anything out. Yeah, yeah, yeah. You know? Okay, cool. ⁓ so I wanna shift the conversation a little bit to everyone saying, ⁓ there's like an AI infra bubble going on right now. Do you agree with that, disagree with that? You know, all the big hyperscalers like Google or whatever, like in I guess combined, they want to spend trillions of dollars over the next few years on AI infrastructure. So how do you reconcile those two things that on the one hand everyone's saying that, ⁓ we're in a bubble. On the other hand, these huge companies who are supposed to be the most sophisticated buyers in the world are spending a lot of money on this stuff. So yeah, like how do you think about that? Okay. I I think people are investing behind a a ⁓ massive demonstrable value. That's that's what I really think. ⁓ is it big and have those investments sort of come online quickly and grown really rapidly? Yeah. ⁓ will there Will there be a settling in the market? And and I and I don't necessarily mean like a a correction, but will there be a a settling out or a reconciliation where some investments pay off and and and others have to be redirected? Yes, that's how markets work. But I actually I I I don't think we're in a bubble because when I think about a bubble, not an economist. When I think about a bubble, what I really think about is ⁓ a value that's inflated beyond the inherent value. of the underlying technology, product, services, assets. And and I don't think we're there with AI. I I don't think we've really discovered the full value potential of AI. And so I think these really sophisticated organizations that you mentioned, some of whom are our partners, I think they're I think they're investing into and ahead of a a a legitimate value hypothesis. And a lot of what I heard is where basically you're not able to build data centers fast enough to ⁓ to service all the demand. So is that a problem? ⁓ Yeah, it's a problem. It's a it's it it's a it's a great and interesting problem. So I think like you know, coming back to to Cerebrus and and what we're doing here and my own view on the market, right? Th there is a ⁓ a fundamental mismatch between the timescale of development for hardware solutions and physical infrastructure. And the software applications that people build on top of it. Right. So what we're seeing now is sort of this massive acceleration and uptake of AI software applications services that are that are fundamentally built on top of ⁓ this physical infrastructure of metal racks, power lines, fiber cables, big boxes with wafer scale chips inside. It takes longer in real clock time. To build these physical systems and pour concrete and and and raise up data centers. There's almost no way that you can build that physical infrastructure as quickly as it's being adopted ⁓ when when you're in a moment like this. And so yeah, I think it's I think it's a problem. But I think it's a really, really interesting one. Right. We at Cerebris, we happen to I I I think we we did get ahead of it a little bit, but recall that we started this project. almost ten years ago, right? With seeing that we would need to invest for years in order to build the right instrument for the future of this work. So we we we knew we had to be ahead of it because we thought it was going to be big and important ⁓ and it was different enough to be valuable. We also knew it was going to take time. And to be candid, I think we as an industry probably didn't have that same ⁓ lead time opportunity with inference in the in the post chat GPT moment. to like get an get a head start pouring concrete and building data centers and and developing like thoughtful energy strategies ⁓ for for for for global infrastructure. So I think we're as a as an industry, I think we're playing a little bit of catch up right now. Yeah. ⁓ but I'm optimistic on what that means that our sort of community of innovators around us are going to create, right? What that challenge puts pressure on creative people and investors. And we're starting to see people develop, for example, novel energy solutions, storage and generation. ⁓ we're starting to see people come up with, you know, clever approaches to to reuse physical facilities and finance data center expansions. Right. Like so I I think it's a problem, yes. ⁓ but but I think it's a a challenge that is is is gonna be really exciting for us as a community to solve. ⁓ and I'm smiling because I think if we solve it, then w we get together to see this potential of this technology ⁓ have have some really ⁓ I think positive impacts on the world. Yeah. So I guess one of those bottlenecks is data center development. The other one is memory. I think a lot of people are, you know, seeing all these memory stocks just blow up. Why is that happening? Like why is there a shortage in memory? ⁓ yeah, could you explain that? Sure. I mean so back to The the fundamental architecture of a traditional chip is that logic die and then the memory die. And what we're really talking about here is that type of memory, which is used by GPUs called HBM memory. ⁓ that memory ⁓ I I is ⁓ you know, subject to ⁓ fabrication limits, manufacturing limits, similar to what we talked about before, right? There's physical facilities, machines that need to be built in order to turn up capacity. You can't just turn a knob and produce ten X more. And so ⁓ that memory is is is a bottleneck right now for those types of chips. For us, it's not, right? We have all of our memory on the chip as right here. It's not part of that supply chain. So we don't have the same challenges ⁓ to the being subject to to HBM ⁓ supply chain limitations as ⁓ as other chips do. So why so this is called SRAM. Yep. Yeah. So Why is there not a shortage of SRAM, but there's a shortage of DRAM? It's it's really about the ⁓ the the fabs and manufacturing facilities, right? So for a traditional chip, ⁓ it goes through, say, the same fab that we do at TSMC or Samsung. ⁓ and it's it's it's one process, right? Building memory chips uses a a different fab, a different manufacturing facility, different supply chain. Because we can leverage the the ⁓ the traditional chip process. we don't we aren't subject to those we don't need the other manufacturing facilities for our systems. So I guess that's a that's really good for you guys. 'Cause then the 'cause the if the memory memory costs are going up five, ten X, then all the GPUs are also gonna have to go up in price, I'm assuming. But you will not have to. I think that's yeah. I th I I think that's that's about right. Right. When we so when we build big clusters, we have other systems in the clusters that do use that memory. ⁓ so the it's a it's a part of our solution, but we we're not nearly as reliant upon it. ⁓ so I would say that that's it's it's an opportunity for us ⁓ to be able to deliver sooner for customers, ⁓ and to be able to deliver ultra high performance, leading performance in the market at an even more competitive price point. So The other bottleneck in data center development everyone talks about is energy. Do you agree that that's a bottleneck? Or why wh why is that a bottleneck, I guess? Bottleneck is an interesting word. ⁓ I think Yeah. It is a constraint on the rate of development. And I think it's also an important constraint to be mindful of when we think about environmental impact impact on the communities. ⁓ is it a i is it a fundamental bottleneck that is i you know, is there enough energy on the planet? I don't I don't think we're quite there yet. ⁓ but I do think as a community, as an industry, we need to think about not just how to make AI bigger, faster, and more capable and more valuable. We also have to think about how to make AI more efficient. ⁓ and that by by efficient, I mean more efficient at learning. Right. ⁓ using maybe fewer weights to learn the same thing, ⁓ or using fewer compute flops or watts to to learn the same thing or less data. So more efficient learners. But I also think we need to make AI that is more efficient in terms of execution and output. So we need more efficient chips for that, and we need more efficient ways to deliver energy at scale, ⁓ whether that's through updates to the grid or improvements to energy storage in communities and at large data center facilities, or if it means changes in in how we generate, whether that's things like small modular nuclear reactors or, dare I say, at data centers in space, where you have all the energy of the sun pouring down on solar panels. Right. So I I I think that in in that world, right, we're very much a part of the solution. Because but by building this big chip, not only are we faster, ⁓ but we can also save energy on things like communication of data and memory access, because we're not accessing memory from a different chip. We're not pushing data between two different chips in a cluster meters apart. We're accessing data from our own S RAM on the same chip and pushing it ⁓ microns and and nanoseconds from core to core. So we're a part of this solution, I think, but th the as as a community, and I think this is what a ⁓ a lot of f sort of folks that aren't in this industry are probably thinking about, we also need to think about how to deliver energy more more efficiently, make this whole process more energy efficient, ⁓ and and and really think of it holistically as a problem. So I do think it's a a bottleneck, but not in the way that it's a sort of a fundamental limitation, but in the way where w we need to figure out how to build more efficiently And deliver this capability ⁓ more efficiently, particularly per kilowatt, kilowatt hour. Yeah. So I want to shift the conversation a little bit to like you building your own data centers. Are you guys building your own data centers? Or yeah, why why is that? Why is a chip company making their own data centers? Yeah. So to so to be clear, I I often joke that ⁓ working on this ⁓ at Cerebrus on the Waifor Scale Engine in our systems, I never thought I'd learn so much about plumbing. our systems are water cooled. So when we deploy to a data center, ⁓ we we we take water, cool water, we push it across the back of our wafers inside the machines, ⁓ and that allows us to keep the chip cool. As a result, we have a bunch of pipes running through floors and ceilings, and we have to think about how to keep the water cool. ⁓ sometimes things leak, so there's questions about plumbing, and that turns into conversations about ⁓ things like joint ⁓ you know, connecting pipes together and wrenches. ⁓ it's it's a it's a real physical infrastructure problem. ⁓ I think that in that sense we're not building our own data centers, to be clear, right? We're we're not actually pouring concrete from like trucks with Cerebrus logos and you know, setting up facilities that way. What we're doing is we're contracting with folks that either own facilities already or are developing facilities, and ⁓ we specify to them ⁓ a particular layout of the floor. Power requirements, inbound data requirements, cooling requirements. ⁓ and then we we work with them to to to basically contract and and lease that facility. I think that's your question, but I wanted to clarify that for the audience because I think when you think like building your own data centers like, man, what are these guys getting into? Right. I mean, like at the at the end of it, do you own it or does a different company own it? Are you kind of overseeing the development? I guess Really depends on the customer, but typically we don't own it. We're we're we're leasing the facility. Right. And and and and renting the power. There are cases where where we where we have our own our own ownership, ⁓ or we're operating a facility on behalf of a customer who might own the facility. But often what we're doing is we're we're leasing the facility. ⁓ in in that context, right, we we chose to to to ⁓ effectively roll our own data centers and build our own cloud, ⁓ fundamentally because we had so much demand for what we were building, that is serving fast inference ⁓ that we w we needed to sort of own our own destiny in rolling out data centers. ⁓ and we also needed a particular, you know, for for each compute provider this is true, but we we wanted a particular layout of power, network, cooling, and racks. ⁓ and and that allowed us to go quickly in response to the market. Okay. I want to shift the conversation back to the chips actually. Yeah. ⁓ but more I guess on the economic side. So how do you think the costs well, first of all, what are the costs involved in ⁓ building a chip? And then how do you think those costs are going to evolve over the next five years? This is another good one. So I I I'll try to keep it at a high level because e each one of these the questions are great. Each question could be sort of a a one on one one on one course. Yeah. Right. But I think the the the costs of developing a system ⁓ a chip and system are fundamentally the design, the engineering development, the prototyping, and then the manufacturing at scale. ⁓ and particularly the last one, the manufacturing at scale, ⁓ also involves the the inbound supply of all of your subcomponents, right? So ⁓ just to give you a sense of the sort of the timelines involved in this, ⁓ Company started in 2016. I started in 2017. ⁓ we at that time had designs for the first waferscale engine. ⁓ and if I recall correctly, we launched the first version of our Cerebrus system, CS1, based on that first waferscale engine in 2019. So you can imagine to yourself it took took about two or three years to get from a design into a a workable device. And then it took another few years, probably between our first generation system CS1 and CS2, really to iron out most of the ⁓ the the the the system and software challenges to make those units really reliable at data center standard and production scale. So it takes time and it has all those different cost centers. It's not a cheap project by by any stretch of the imagination. Now back to your question. Where do those costs go in the future? ⁓ one of the areas I'm most excited about is ⁓ AI accelerated, AI-augmented chip design. ⁓ some really interesting work building foundation AI models to enable faster chip design, or maybe even use AI to design chips that compute AI that can then develop new chips. ⁓ So that that area of of chip design, I think is one that is, in some sense, ripe for acceleration and potentially ripe for cost or barrier to entry reduction in in the future of computing. I think once you have a chip design, ⁓ even if it's faster and cheaper than it was before, you still have to ⁓ figure out how to package that device. That is, bring power, data, cooling, put it on a board that integrates into a system that can go into a server or a data center act. So still have to figure out the physical packaging and and the prototyping. I think those things will get faster, but probably not at a sort of order of magnitude kind of X factor, right? Maybe we get two X fac faster. ⁓ maybe there are tools or equipment costs or learnings from industry that drive the cost down by a factor of one and a half or two X, right? Just hip pocket numbers. Yeah. ⁓ Then the manufacturing at scale, that last piece. ⁓ I think there it's gonna get really interesting. I think some of the fundamental supplies, like you were talking about memory, right? There's a lot of conversations about rare earths and materials that go into these systems. The supply chain that goes into manufacturing at scale, ⁓ I don't think any of that gets cheaper, right? Or I would be surprised if it does. I think the entire ⁓ history of human species and industry would suggest that. as ⁓ as as as ⁓ interests as demand grows, right, and and supply sort of stays the same or or gets smaller, unless we can find new sources, right? That's that's gonna get more expensive. Then then the manufacturing that is the actual chip fabrication, system manufacturing, I think that's an area that's ⁓ also ripe for more efficiency and and cost reduction. We're seeing teams here in the Bay Area build up prototype manufacturing lines. ⁓ and build scaled production manufacturing lines far faster than they could five years ago even, ⁓ and use more automation and more data to drive not just manufacturing velocity, but drive costs down. So in summary, I I think cost of of design opportunity to decrease, cost of packaging and prototyping, meh. And cost of manufacturing at scale might might also be a wash if we see supply process supply prices go up and ⁓ and manufacturing prices at scale perhaps go down. Are there certain rare earths that you know y you see as like these are the most important in terms of or maybe these are the ones you're focused more on because there's some, you know, you know, some country has ninety percent of the reserves, you know, things like that. Yeah. I mean, just to be candid, not ⁓ at this moment, no. ⁓ but I I think we're a as a as an industry, we're keeping an eye on that. And we're also i i excited to find ⁓ other sources for those materials. Yeah. ⁓ so Google Google has like a full supply chain of you know, they basically own everything, like they own the software, they own the they're building their own TPUs, they're offering it as a service through GCP. ⁓ and that makes it probably like very cost efficient for them to offer it to other companies if they want to. ⁓ Do you think they will be the lowest cost producer of tokens, I guess, ⁓ going forward? It's ⁓ it's an interesting question. I I'm not sure I want to speculate on the on who will be lowest cost. Yeah. But I do think you're you're you're fundamental there on, you know, that they have a a ⁓ a vertically integrated product ⁓ that that they that they can offer at scale. That is that is true. ⁓ I came from Google, I know that team. Outstanding team. That being said, ⁓ I think rather than focus on cost, ⁓ what we're focused on is performance. So delivering the the fastest tokens per second, ⁓ at a competitive price and with greater power efficiency. And I think that back to our conversation about the future of AI computing, right? Well I th I think TPUs are are an extraordinary machine. ⁓ I think in the context of the future of AI, there's a place for a machine like that, and there's a a place for an ultra fast race car machine like ours. and to the premise of your question, one of the reasons why we vertically integrated is so that we can make the the fastest AI computers on the planet, but do it at scale and and be competitive to those costs. So I think there's gonna be ⁓ future systems that deliver, I'll call them medium speed tokens. at great, you know, at at a really, really competitive price. But for applications, for the user applications that that we were talking about before, right? Voice reasoning agents that are that I think are really the future of AI computing, their speed is paramount. Right. The those applications just don't get built or aren't feasible if they don't answer quickly enough. And so yes, there's going to be a future of ⁓ of compute that delivers medium speed tokens at really low costs. That's not a ⁓ that that's not a market I'm interested in pursuing. Yeah. But the the market for fast tokens for the future of AI applications that are at a competitive cost, that's that that's what I want to go hunt and win. Yeah. So why is speed so important? ⁓ look, we know this from the entire history of the internet. And I'll I'll I'll I'll give two examples. One, we just talked about the company. Let's talk about Google search, right? ⁓ would you wait for a Google search answer? When was the last time you saw the little Sort of pinwheel of death on a Mac pop up from a from a a Google search query. ⁓ I I think that and before you say it, AI summaries, right? I I think early AI summaries made us wait. And and and even just maybe six or eight months later or a year later, we don't want to wait for AI summaries anymore. So I think speed matters and and and actually Google search and and and the team there did a a lot of research to to show the effect of speed. on user satisfaction and user retention. And what we know from that work and by proxy the entire history of the internet is that every millisecond of latency matters for these valuable ⁓ or ⁓ day-to-day important applications. Every millisecond that you delay, your users get upset, your users walk away, they go to your competitor. ⁓ so so for any applications that's that that is like that, speed matters. And that's where we're seeing voice reasoning agents going. ⁓ the the second thing that we know about the history of the internet is when that when things get faster, it's not just that the the same ⁓ group of applications exist but go faster. We see entirely new industries. So the example that I have here that is fun is is Netflix, right? When internet speeds went from broadband or from dial-up to broadband, it's not just that Netflix delivered more DVDs by mail faster. When speeds got faster, Netflix went from delivering DVDs to being a digital production studio and streaming entertainment company. So they created an entire new, effectively, industry, and hundreds of billions or trillions of dollars of revenue and lots of user delight from people like me that like to sit on the couch and stream Netflix. So I think speed matters full stop. and when we're working with customers today. Like OpenAI, AWS, Cognition, Notion, ⁓ in those areas of coding agents, voice agents, reasoning models, they know, our partners know that if they can deliver output and answers to their users faster, their users will be happier, more creative, more productive, deliver more value, and use the product more. And ⁓ honestly, that's that's why I'm I'm so excited about where we are in the market. And to see what ⁓ what what our customers are gonna build on top of our systems net. Yeah. So on the open AI deal, could you talk a little bit about that? So is it when I use Codex and I see that they're offering me two X faster speeds for twice the I I forget the cost for it, but ⁓ is that two X faster speed option using Cerebrus on the back end? Yeah, so we're we're we're building, I'm happy to talk a little bit about that. We have a great partnership with OpenAI. ⁓ We got to know their team actually from the early days of Cerebrus. We we saw their ambitious vision to build state of the art frontier AI models in AGI. I think they saw our crazy ambitious vision to to build a computer system that would be the right system for the future of AI. And it's it it's it's great and serendipitous that we kept track of each other through those, you know, almost 10 years and then we're able to to reconnect ⁓ and partner. To deliver fast AI for coding and agentic workflows now. So we have a deep co-design collaboration with OpenAI engineering team. So the the public terms of the deal are they're gonna buy 750 megawatts of compute by Cerebrus to power their applications. ⁓ the coding model that uses Cerebrus now is Codex Spark, and we're working with the OpenAI team to bring up ⁓ additional models in the future from their ⁓ from the GPT ⁓ frontier models to ⁓ to next generation coding models, all of which ⁓ should be the the the the fastest examples of those models in industry. Yeah. And I'm assuming the thing stopping you from working on the bigger models right now is just the data center issue. Aaron You know, it it it's really a matter of of of product strategy and prioritization. ⁓ really? Okay. Yeah. I mean certainly More computers would be better. ⁓ OpenAI has ⁓ incredible reach into the market, incredible user base. If we could serve all of them today with the fastest tokens, I'd like to say that we would. Yeah. So yeah, you're you're not wrong, right? Like if I had more computers, it would be better. But w we're also working really closely with their product teams to say, look, here's the library of models and applications that are really important to your users today. Which are those that ⁓ w where where speed is the most valuable ⁓ or most enabling. ⁓ and then let's look at the compute resources that we have together and let's let's rack and stack and then develop a staged approach to ⁓ to bring the right fast models to your users first based on what we have. It's a fun exercise actually. It's really cool to work with their team. That's really cool. Are you planning on working with Anthropic in the future? ⁓ look I I know Dario and and the team really well. ⁓ extraordinary team, extraordinary technology. I hope we get the opportunity to do more. Yeah. So right now, I think you just mentioned you have a deal with Anthropic for ⁓ was it seven hundred and fifty megawatts of compute? Open AI. ⁓ sorry, open AI. Open AI. Yeah, yeah. If we're so like a few years ago, I think talking about seven hundred and fifty megawatts of compute, like that would be insane to talk about. And now, you know, people talk about gigawatts and where do you think that goes in five years? Yeah, I mean, I don't know, this is I'm I'm showing my age here, but for but but maybe you and and I'm I'm sure some of your viewers and listeners remember Back to the Future. ⁓ there was a a joke about one point twenty one they call it gigawatts, but one point twenty one gigawatts to get the car to go back to the future, right, from a from a bolt of lightning and you know, i i in that, you know in that sci fi movie of my youth. It so it sounded like a made up number, you know, it sounded crazy. And funny, but unbelievable in just the right kind of way. And and now we're talking about those kinds of power envelopes in a very realistic way in the next, you know, twelve, eighteen months for for some of these systems. I I actually think it's sort of an an awe inspiring ⁓ trajectory for the industry that and and a signal of fundamental demand and value that we're talking about building infrastructure at this scale. ⁓ the the only analogy that I can think of is through the the the building of of the railways or or laying of undersea fiber to enable the internet. Right. This is a a generational infrastructure project that's being taken on by a coalition really of individual members of industry and governments ⁓ to to to build this the in some sense the the rail lines of the AI industry future. ⁓ what do I think we're gonna be talking about two or three years from now in terms of power? I I'd I I'd I'd venture may maybe I'll get some some blowback from this, but I'm I I'd venture we're still talking about gigawatts. maybe maybe ten gigawatts. But I think we start to run into some limitations of physics. Yeah. ⁓ and w without new generation or or significant grid updates that that do that does take some time. I think in a few years we'll probably still talk. So you're also leading US federal programs. Could you talk about like what does that mean? Sure. ⁓ so At Cerebrus, the US government ⁓ was one of our very first customers. ⁓ we worked with the Department of Energy, Argonne National Labs, Lawrence Livermore National Labs. We're now doing a lot of work with Sandia National Laboratory. ⁓ we also did some early work ⁓ through ⁓ microprocessor RD with the Department of Defense and DARPA. ⁓ so the the US government has been a longtime customer for us, but also a technology development partner. And I think for For folks that are building new companies today, I'm happy to share more about this ⁓ in another forum or if people want to reach out directly. But I think that the the US government can be a great partner. ⁓ Many organizations within the government actually have the charter to be early adopters of new technologies. In our case, the Department of Energy, who builds world-class supercomputers for scientific simulation, also has the charter from the US government to be an early adopter of new computing technologies. So that the government and its research scientists can sort of get a early peek at what industry is building and participate in it and also inform its development so that it can contribute back to science and society and public interest. I maybe I'm dorking out a little bit too much over it, but I think that that's a really, really incredible mission and function that the government plays. And for people that are building new deep tech or other technology startups, I would say look, don't be afraid of the government. Lean into the government and and then there may be ways that they can help you ⁓ get revenue or get access to technologies or or partners that your your industry partners may not be able to. So at Cerebrus government has been ⁓ an early customer and a great technology development partner. We're currently working with different parts of the US government on future IO or interconnect technologies as well as future memory technologies. ⁓ Coming back to what we talked about before. So we're continuing to do work with the government on those kinds of projects. We are also a part of the US national AI initiative, which is called Genesis, which in summary aims to build AI computing infrastructure to fundamentally accelerate how we do scientific discovery. Everything from biology and life sciences to physics and chemistry and drug design to space and the environment. It's an incredible mission and actually I think it's something that the US needs to do in order to keep pulling industry forward and help us as as a nation and as an American industry lead in this technology that in in some sense we had a big hand in inventing. Right? If in other words, if we want to continue leading and extend our leadership of AI technology industry in the world, It's gonna take a a big initiative from leadership in the country to do it. So we're a part of the Genesis program. ⁓ and we're also working with the Department of Commerce to figure out how to take what we're building here in the US with fast AI wafer scale systems and cerebral supercomputers and figure out how to package that with other US AI technologies and quickly and easily export those technologies to friend and ally nations. So that's where our our work with the federal government also intersects with international sovereign initiatives. That is, ⁓ how how can we enable our friends and allies with world-class AI technologies that are built in the US? ⁓ at least for me, I'm I'm I'm a believer that ⁓ we get closer to realizing the the full and positive potential of AI if the right AI computers and AI technologies are in the hands of more, not fewer. organizations and countries. so one of the reasons that I'm at Cerebris is to build that right AI instrument, build that right AI computer, and then put it into as many developers' hands, as many organizations' hands, companies and countries' hands as I possibly can, because if everybody can build faster, then we're much more likely to discover that next great thing that fundamentally changes our our health, happiness, ability to do work, productivity, than we would be if all these computers just lived in one place. So the work with the federal government helps us do that and helps us get access to foreign markets. Cool. ⁓ what are some of those projects that you're working on to export to foreign markets if you're able to talk about some of that? Yeah, absolutely. So this is ⁓ so there's a there's actually a a a new, relatively new program ⁓ l led at the US federal level by the Department of Commerce ⁓ called the AI export tech stack. ⁓ and so That that project is basically the US government providing a a ⁓ framework for industry members to come together, define a a common stack or a or a a menu, if you will, of AI technology solutions. And then the US government will work with industry to streamline the export policy for those so that ⁓ industry members can quickly package and sell and export those technologies. To friend and ally nations. That's one that's that's the sort of the the general overarching goal of that program and and and where we fit in as as a technology provider and partner. And we're also helping the government define what that stack looks like and what the menu of solutions might look like. ⁓ for for us though, that's actually ⁓ strongly informed by the work that we started four or five years ago now with our strategic partner in the UAE called G42. ⁓ There's there's lots of information out online about our work with G forty two, but long story short, they came to us about five years ago and we wrote together a an MOU, a memo of understanding, that articulated a vision that th that they wanted to build world class supercomputers and world class AI models in Arabic language, in health, and science and finance. ⁓ and and the the MOU was that we would do this together. ⁓ fast forward to just a few weeks ago when we had our IPO. I was I was sitting down with the the CEO of of G forty two and we were reminiscing a little bit that we we wrote that MOU and all of it came true. We've built with G42 as a as a commercial national champion for the UAE. We have built for G42 world-class supercomputers. We've collaborated on state-of-the-art model development. We worked with ⁓ UAE researchers to build the the the first and still best. open Arabic GPT model so that ⁓ world c so that state-of-the-art AI can speak the language of of local people and reflect the cultural interests and social and commercial interests of that market. We've worked with their healthcare teams to build state-of-the-art AI healthcare models. That work ⁓ I I think was reflected this ⁓ ambitious vision of the UAE to be a leader in AI technology. And I'm I'm humbled to say that I I I think we played a a a helpful hand role in bringing them onto the the world stage and where they are today. And I I think if we can do that for more of our international partners, ⁓ through, for example, the current US Department of Commerce program, ⁓ then then we're we're all winning. Yeah. Is it easier to build data centers in the UAE than the US? Is it faster? You know, that's a good question. I d I I actually don't know the answer to that. I I let me put it this way. I'm looking forward to seeing how fast they can build data centers. ⁓ what's the most surprising thing about working with governments? Is this your first time working with governments? ⁓ no, actually for the past twenty years. So I I I I used to work at Google. Before that I was building satellites, and before that I was I I was building image processing algorithms and software for environmental surveys. And ⁓ in all cases actually I was I was working with ⁓ with governments. ⁓ wow. ⁓ but I think what what surprises me most here, ⁓ and in AI particularly, is I think some governments are really moving fast. And by design, moving fast is not a thing that most governments are good at, right? In fact, th th you I think as a citizen you generally want your government to be deliberative, right, and and fair and thoughtful and take its time. ⁓ but I think some i some cases velocity really matters. And I think ⁓ emerging as a leader or developing state of the art AI, velocity matters right now. And so I think the surprising thing is like is is seeing some governments really, really lenient. I think UAE is a great example of that. ⁓ I I I've been thrilled to see ⁓ over the past twelve, twenty four months how the US government has sort of put the proverbial pedal of the metal on AI. ⁓ and and I think other nations around the world are starting to do the same thing. Yeah. So on that note, should we be selling chips to China? That's a as you know, that's a complicated question. ⁓ we've had the opportunity to work with China in the past, sell chips to China in the past. ⁓ and and and we decided not to. ⁓ I think there's an incredible developer market. There's an incredible research market in China ⁓ that I think as as a society we would love to see go faster. I think if we can answer some fundamental questions and come to agreement collectively on things like ⁓ security, ⁓ then we absolutely want to be able to work together. Yeah. I've heard some people say, hey, we should sell to China, but we maybe shouldn't sell our best chips to China. What like w what do you think about that? This is so I I I also work pretty closely with the the part of the Department of Commerce that administers export policy. It's called the the Bureau of ⁓ Industry Security, BIS. ⁓ huge shout out to the people that work in that part of the Department of Commerce. They have what I think of is is is probably one of the hardest jobs on the planet because they it at the end of the day, they are charged with balancing US national security interests with the indust the the interests of US industry and our economy. Right. So on one hand, you might say, open up the floodgates, sell everywhere. It's good for industry. It's going to be a boom for our economy. On the other end of the spectrum, you might say, we have fundamental questions about how these chips are going to be used. That could threaten national security, or we just don't know. So let's just stop for now. Both of those endpoints have challenging consequences. So finding the right middle ground for any given global geopolitical state of play on any given day is incredibly difficult. Right. If you hold back too much, you over incentivize the development of competitors. If you deliver too much, what you know, what security might you put at risk? It's almost an impossible problem to solve. And so I think the the nuance of getting that just right is incredibly difficult. I will I I will not claim to have the answer. of of of what is the right thing to do. ⁓ I think, as I said before, if we can answer those fundamental questions about security, if we can have the right diplomatic engagement, ⁓ if we can agree on the markets, we would love to be able to help the folks that are working on fundamental science AI, new commercial AI, new open source AI in those markets go faster. Yeah. So what are your thoughts on Europe? Everyone says, hey, Europe is so far behind, you know, their policy is Or there w there's way too much regulation, way way too much taxation. Have you guys worked with Europe? Yeah. And where are you seeing that headed over the next couple of years? Yeah, in fact, ⁓ I'm I'm going next week to to Germany to a conference called International Supercomputing, where we're gonna be sitting down with the with with European owners of state of the art ⁓ government public sector supercomputers, but also owners and developers of new commercial AI infrastructure and data centers. ⁓ we have ⁓ A long relationship ⁓ and extraordinary system in partnership with ⁓ EPCC, which is the supercomputing center in Edinburgh, Scotland. ⁓ so we actually have systems in Europe. We also have systems in Germany. ⁓ and we are working to deploy data center for our commercial inference services in France. That's all just context. Back to your question. I'm trying to encourage European. AI infrastructure developers to think differently and move faster. I do think that there is an awesome opportunity. I think back to our conversations about our conversation about government, I think European governments are trying to be really deliberative and make the right choice, which is typically in the interest of the population. But I think right now industry is moving faster. And i if they also want to lead as friends and allies. I I want to encourage them to to to move faster. And so that's that's that's that's one of the the soapboxes that I'm gonna be on next week is let's like as a community, let's think a little differently about AI infrastructure, right? It's not just GPUs. ⁓ speed matters, where speed matters, there's solutions like ours, but also how how can we help you move faster? ⁓ whether it's through commercial data centers or public supercomputing projects. Cool. I wanna shift the conversation. This is gonna be like the last couple of questions. ⁓ what predictions do you have about where enterprises or how enterprises are going to be using LLMs over the next couple of years? Like right now, maybe coding is a is a I mean, I guess we talked about this a little bit. Talked about it little bit, but but you know, we talked about it a little bit before the podcast too, that y right now AI is showing this massive potential. And and clearly there's a a value proposition to be ⁓ to be harvested in coding. ⁓ But how we as a community build those tools to put them in the hands of enterprises so they can use them at scale, that's a really tricky nut to crack. And so I think it even in coding today, we see massive adoption, massive value, but we still see large enterprises like sort of figuring out how to adopt it at scale, right? How how do we integrate these tools with our workflows? How do we Integrate these capabilities with our current teams. Right. So I think step one is we will see enterprises working with industry to adopt things like applications like coding and fast coding agents like those on our systems at scale. So step one is like increase adoption, right? Or absorption of that capability. ⁓ What what we're starting to see from some of our partners in enterprise is development of new, I'll call them domain or industry-specific models. Right. Maybe maybe it's not a language model for coding or chat, but maybe it's an LLM or a a sequence style model for drug development ⁓ or for material design. ⁓ or an AI model for physical industry for say manufacturing to say control robots or analyze data off of the manufacturing line. That realm of ⁓ broadly speaking, physical AI and domain-specific models for industry, I I think is a hugely exciting and active area of of development that I think is over the next couple of years, we're gonna see some big changes. So we have longtime partners like Glaxo Smith Klein that for the past five plus years have actually been building AI models to accelerate and improve the development of new therapeutics. ⁓ We have customers like the national labs been working on fundamental AI models for the physical and life sciences. And consumer agent ⁓ models, I think that area of physical AI is is is is gonna be big over the next couple of years. Another tangential question off that, why do you think like you've probably read those stories that, hey, like ninety-five percent of enterprise adoption projects for AI are just failing. Why do you think that is? I in some sense I think they measured the wrong thing at the wrong time. Right. I think those studies it it's it's important that we that that we measure ⁓ sort of value return, but I think it's far too early. To ask that question really at scale, or not really to ask the question at scale, you can ask anytime. It's too early to come to that conclusion. Right. The the the way that enterprises typically trial and then adopt and scale technologies takes 12, 18, 24 months, right? Or longer for for some big enterprises, especially if that enterprise is the government. Right. ⁓ you know, you you you start a pilot, you run maybe a series of pilots. It's intended and known that. Maybe eight out of those ten pilots will not show results. But then you have some subset of those pilots that show positive results. You expand the trial audience, you start to scale it up. Only after you scale it up, and then maybe after a a year of it operating at scale do you actually start to see returns, right? That that's that's how sort of the the life cycle and timescale of enterprise adoption and new technology works. And Like I said, I do think it's important to ask the question of are are we getting value out? But I think it's a little too early to come to that conclusion. Yeah. I mean to me, I definitely see a lot of people getting value, especially startups are definitely seeing that they're creating value. I think the problem with enterprises is maybe they're a little slower to adopt than like startups. So ⁓ yeah, and exactly what you're saying, right? You have to run ten experiments, two of them are gonna work out and then you just scale it. So you know, it's funny, right? Like the I I think I I like where you're going with that and I was chuckling about like the problem with enterprise. ⁓ I I happen to like working with big enterprises because they have incredible reach, right? But to your point, they also tend to ⁓ be more deliberate and move a little slower. There's just more people involved. There's there's more people involved, right? And there's there's ⁓ in in intentional controls in place. It's all good. It does take time. It's all good. ⁓ I think you're right though to bring up startups, right? Like i in in for a green field problem. or a brand new team. We're we're seeing people build from scratch with AI. And I think those are gonna be some really, really compelling disruptors ⁓ is is is those teams that are AI native from zero. Yeah. I've been talking to a lot of engineers at some startups and it's kind of crazy to me that, you know, all of them are telling me, I mean, these are like really, really amazing engineers and they're like, I haven't looked at a line of code in seven months, eight months, because they just use agents for everything. Yeah. It's pretty crazy. No, I mean look, we we we do internally at Cerebrus as well. Over the past six months, we're we've we had ⁓ probably a half a dozen different programs across our entire engineering teams to from from design in hardware to program management and product management to software engineering and AI. All of them now ⁓ are are are effectively bec being AI native for for all the code design. And I gotta say, that's really reinforcing to us here too, right? Because back to the OG hypothesis, if we can build the right instrument for this work, we we think this work could be really important and transformative. And now we see this that that same workload literally transforming how we're doing what we're doing. And so ⁓ it it's it's awesome and encouraging to see that sort of substantiation, even in our day to day work. Yeah. What are some projects that you are currently working on that you're really excited about that you want to share? I think I mean everything is a cheater answer. But I I I come in to this company every day and I am just at awe and humbled by the ambition and curiosity, hustle and resilience of everyone of our staff, right? Not just our engineers, our program managers, people that are working at the office. It's it's awesome and everybody is. incentivized in some sense to be a part of of building this and changing industry. And ⁓ there's no task too small, and there's no vision that's sort of too too too big. ⁓ and so a across the board, I'm excited, ⁓ from model research to software development and systems. But I think what came to mind immediately is it's also really cool to work at a hardware company because for those of us that spend most of our time in front of a screen or working with documents or PowerPoint or God forbid, spreadsheets. ⁓ being able to take a break and like walk into the back and see people turning bolts on things and see the the the things that you you're creating in a digital realm actually take shape in the physical world. That's it it's it's an incredible opportunity. So I think to your question, I'm really excited about our future hardware, actually. At the end of the day, we are a hardware company. Yeah. And so you we are on CS Three now, you can imagine that there is a four and a five and a and and more beyond that. ⁓ I think what we've seen from the market is that there's extraordinary value in fast inference and fast training. In some sense, I think we've proven out the value and burned down the risk of the the wafer scale processor. ⁓ I'm really excited to see what we build in our next generations to go faster. ⁓ as I mentioned, we've got some fundamental investments going right now in I.O. and memory technologies and future systems. So in some sense I think the ⁓ wafer scale engine of generations one through three where we are now is ⁓ it's been incredible, but it's just the beginning. And so for for the audience out there, I would say stay tuned because we've got some really, really exciting stuff in the hardware pipeline and ⁓ I I I think that could transform computing very much in the in in the same way that the wafer scale engine did in the first place. Awesome. Well that's all I had. Thanks for coming in. It was a lot of fun, man. Thank you so much. ⁓ I hope we can continue the conversation. We should check in on those future predictions sometime and see see what came true and see what was a puff of smoke. Yeah, yeah, yeah. Awesome. Thank you. Is there any topic that you wanted to cover that, you know, I didn't cover? I thought it was great. Okay. I thought it was great. No. ⁓ I mean I'm sure I'll think of something later, but I I I just enjoyed the conversation. I hope it ⁓ I hope it it sticks with your audience. Yeah. ⁓ and I really appreciate it. Yeah, I learned a lot. Awesome.