theCUBE + NYSE Wired: The AI Factory - Data Center of the Future | Diane Bryant

theCUBE.net

- Links
  
  theCUBE Network SiliconANGLE Media
info Help

Clips
More from theCUBE + NYSE Wired: The AI Factory - Data Center of the Future

Diane Bryant

Chairman, CEO, Independent Director

play_circle_outline Generative AI needs large infrastructure; enterprise must increase adoption for growth.

play_circle_outline Enterprise adoption of AI is slow due to cultural and workflow changes.

play_circle_outline Overall optimism about AI's future impact on business operations and efficiencies.

play_circle_outline The evolving landscape of AI may lead to market consolidation among key players.

play_circle_outline Unlocking Enterprise Efficiency: Leveraging SaaS Solutions and Startup Innovation in AI Integration and Competitive Advantage

play_circle_outline Need for enterprises to manage change effectively for AI integration success.

Info
Transcript

Diane Bryant

Diane Bryant

Chairman, CEO, Independent Director

In this theCUBE + NYSE Wired interview from AI Factories – Data Centers of the Future, theCUBE’s Dave Vellante sits down with industry leader Diane Bryant to break down the new playbook for AI-scale infrastructure and the shifting power dynamics across chips, clouds and colos. Bryant shares candid perspective on Lip-Bu Tan’s leadership at Intel, including deal-making momentum (figures cited in the discussion include multibillion-dollar government support and a strategic NVIDIA investment) and why a durable U.S.-domiciled foundry matters for resilience and nat... Read more

explore Keep Exploring

What are the key trends and statistics related to infrastructure spending and the adoption of generative AI? add

What are the challenges and pathways for enterprises adopting AI solutions? add

What are the challenges and potential advantages of implementing generative AI in enterprise settings compared to consumer-focused models? add

What are the potential future outcomes for generative AI solutions and their infrastructure in terms of consolidation and financial viability? add

What are some sources of quick wins in technology integration, and how do they compare to the challenges of transforming enterprise applications? add

What percentage of downloads today is attributed to enterprise versus consumer use, and what steps are necessary to facilitate the adoption of generative AI tools in enterprise settings? add

bolt Powered by CUBE AI

Diane Bryant

search

Dave Vellante

>> Hi, everybody. Welcome back to the NYSC theCUBE and NYSC Wired's coverage of "AI Factories: Data Centers of the Future." Diane Bryant is here, CUBE alum. It was great to see you again, Diane-

Diane Bryant

>> Thank you....

Dave Vellante

>> after our session out in our Palo Alto office.

Diane Bryant

>> Yeah, this is pretty posh out here at NYSC. I'm impressed. .

Dave Vellante

>> It kind of bougie. We love it. The best is yet to come. So thank you for coming into the studio.

Diane Bryant

>> Thank you.

Dave Vellante

>> I really appreciate you being here. So a lot of action since we last talked.

Diane Bryant

>> A lot.

Dave Vellante

>> Your Lip-Bu Tan, your boy has been making moves. We saw a big, big announcement last week with NVIDIA making an investment in Intel. What are your thoughts on that?

Diane Bryant

>> Well, I'm super proud of Lip-Bu, obviously. The big difference is he's a businessman, and you need a technologist to run a tech company like Intel. It's very complicated, but it needs someone that actually knows how to run a business. And you can see how quickly... I mean, he's a great businessman. Look at what he's done at Cadence. Look what he did in his VC world. He knows semiconductors, but he also knows how to run and make deals and make things happen, and that's what the company needed. So he, obviously, had time to think through given he was on the board and through that process. So he understood Intel so he could hit the ground running, and poof, off to Washington, 8.9 billion. Then off to Jensen, another 5 billion. He's turning the company around. There's huge debt, and he's got to turn it around.

Dave Vellante

>> Well, the turnaround he did with the President was amazing. I mean, one day the President saying he should leave, the next, a month later, he's walking away with 14 billion in cash, which like you say, they need. So I think that underscores his ability to interact with people. I mean, you know him. I don't know him.

Diane Bryant

>> He's very business savvy, a logical thinker, very charismatic, very honest and high integrity. Who wouldn't want to do business with him. Who wouldn't want to give him a few billion.

Dave Vellante

>> So I may have overstated the impact of that deal. I felt like it was more than just money, but you were explaining to me off camera, look, we've always put the CPU and the GPU together, so you don't see that as necessarily a big change. At the same time, the foundry maybe gets a little boost building the SoC. I feel like AMD maybe gets boxed out a little bit with the holy water that Jensen just threw on Intel. Any thoughts on that?

Diane Bryant

>> Well, so you're right. You can separate it and look at two parts. One is on the product line part, the x86 processor. The AMD is there. Intel is there. That's been true for since the eighties when I was driving up the data center business. So you want choice, and so you have two x86 microprocessors. Andy Grove was one that said, hey, we can't have a monopoly. We need someone else in the industry, and he licensed the x86 architecture at AMD. I mean, that's how. It was an intentional. So you have those two in the market. The server has a GPU to boot up, a CPU for boot, and then you have the GPU to offload your matrix multiplication, your AI or high-performance computing workloads. So you need the CPU. And Jensen and everyone else in the industry wants Intel to survive. Nobody wants to see Intel go away. And Lip-Bu is sitting with a company that has incredible debt, so Jensen putting in money to keep an industry icon up and running is great. Then you look at the foundry side of it. When I was at Intel, we tried to start a foundry business literally five times and were never successful. So now this is times six, the sixth time to make it happen. And every time we created a foundry business, of course we went to NVIDIA and asked them if we could build their product in our foundry. You have to have a robust, solid foundry to get someone like NVIDIA's business or anyone else. Again, the world would like an alternative to TSMC. TSMC is running, 60% of all silicon in the world goes to TSMC. Okay, so someone needs to grab the other 40%. So Intel has the opportunity to grab the other 40%. So everybody would like a second source in every area. And so now Intel needs to go off and build a foundry. And again, if Intel fails, that opportunity is gone. And that's the pitch to the US government too. We can't build an Intel foundry without the capital to build it, and it's a national security issue. It's a national icon.

Dave Vellante

>> You'd like that foundry to be US-headquartered.

Diane Bryant

>> Absolutely.

Dave Vellante

>> US-domiciled, for sure. Okay, so we've seen a boom in data center spending, but that's been a one-trick pony, Diane. I mean, it's hyperscalers and neoclouds spending on GPUs and networking. How do you see the enterprise adoption evolving?

Diane Bryant

>> Okay, so there's a lot in there. So number one, you're right. I mean, there's going to be $500 billion in infrastructure build out this year. I mean, that's a crazy amount of money, and that's projected to be a trillion in 2030. So there's massive build out of the infrastructure, which is all, obviously, step one. Then step two, you've got all of these generative AI companies that scale is the game. If you don't have scale, you don't have a generative AI solution. You need very large language models. So they need all that infrastructure to run. If you look at it today, if you look at who is using generative AI, who's using Claude and Gemini and all these folks, ChatGPT, 80% is consumer, and they're not paying. 20% is enterprise, okay? And so 80% of that revenue is coming from enterprise, 80% of their revenue. So you've got to grow that. You've got to get more and more enterprise adopting AI solutions. That is generally a hurdle to get enterprise to do anything. I was CIO of Intel for four years. Enterprise IT moves very, very slowly. And it also, adoption of GenAI, or any other AI solution, is a behavioral change. It's a workflow change. You got to get people to change. So that's going to be a hard slog. The first way that enterprise is going to get the advantage of AI is through the SaaS providers. So SAP, now they've integrated AI into their SAP solution. I'm a CIO, I get SAP through the cloud, poof, I now have new solutions coming-

Dave Vellante

>> Salesforce, Workday.

Diane Bryant

>> Salesforce, and they've announced a lot of that.

Dave Vellante

>> ServiceNow.

Diane Bryant

>> They have new capabilities, new features that you, CIO, can buy. So that's the first way that enterprise is going to get access. The second is through all these startups. I mean, look at how many VC funding, over half of it is going into AI, and 90% of that is SaaS software. So CIOs enterprise will now have the ability to leverage some of these new applications that are coming up and give them a benefit. Then the very last way they're going to take benefit is to build their own AI solutions on-prem, build their own data sets, build their own capability. That's the long, hard slog.

Dave Vellante

>> Do the startups need a better route to market for on-prem because they've just been going to the cloud for the last 10 or 15 years because they didn't have to spend on the CapEx? What is their route to market?

Diane Bryant

>> Oh no, it's definitely SaaS solutions. All those will be delivered as SaaS solutions. There's so much low-hanging fruit to improve efficiency in enterprise through AI, and that's why you see just a massive number of startups. There's just so many opportunities. The one I love is a friend of mine is general counsel at a company, and they've adopted a startup called Brightflag. It's an AI software solution that when you get legal invoices coming in, it will analyze the legal invoice and tell you if you're being overcharged. It'll say, this contract should have been read in one hour, not four. This should have taken one lawyer to evaluate instead of three, and it'll tell you all the places that you have been overcharged. And she said they've never got an invoice where they weren't overcharged, and the legal firms never disputed the overcharges, always reduced the cost to send it back.

Dave Vellante

>> We need Brightflag.

Diane Bryant

>> We need Brightflag. And that's just one silly example, but you could look at that anywhere in your enterprise and see that. I mean, it's quite fun. It's fascinating.

Dave Vellante

>> To your point about opportunities in the enterprise, today it seems like the model is built it, and they will come. OpenAI really, they've got a consumer business model. They're losing tons of money. The scaling laws, people catch up in months. There's just seems to be no competitive advantage that's sustainable. But the enterprise can be different because there's proprietary data in the enterprise that can be applied. It just seems to take longer. As a former CIO, you know that.

Diane Bryant

>> That's true because enterprises have massive amounts of data. So with time, and I do mean time, like I'm talking eight, 10 years, because it's talent also, they'll start using their own data sets maybe integrated with publicly available data sets and improve their efficiency and operations, but that'll take a long time. Meanwhile, all of these generative AI solutions are going to go through a consolidation factor. All this infrastructure that's getting built out, it's all speculative, and there's multi-year commitments that may or may not ever come true because it is speculative. Not all of these folks are going to succeed. So maybe we'll end up with Anthropic, OpenAI, and Google's solution, and of course, Microsoft. That'll be it, but it'll consolidate way down. So now you've got all this infrastructure there's and it's trying to build out. Who's paying for it? First of all, those guys don't have the CapEx to pay for it, so you got to get one of the hyperscalers to trust that you're going to grow into that capacity, and we're going to have exactly what happened in 1990s with telco. Internet came along, all those telco companies were building out fiber to support the internet, look how many of them went bankrupt, defunct, Frontier, WorldCom, Lucent.

Dave Vellante

>> Enron.

Diane Bryant

>> So we're going to have... I'll end that though. But today would you say we have too much internet capacity, and there's still 3 billion people.

Dave Vellante

>> To your point, it worked.

Diane Bryant

>> Supply, demand. Demand will come, but I do think we've got this flurry of building out data centers that exceed demand today. And then over time, just like with the telcos, over time you're going to consume all that capacity.

Dave Vellante

>> Don't you think a big difference is that was funded with debt, today it's funded with cashflow, the hyperscalers. But at the same time, the hyperscalers could wake up one day and say, you know what, like the Oracle deal, they got a $450 billion backlog. Much of that is OpenAI. And if OpenAI says, you know what? We're going to dial it down. That's going to have a ripple effect.

Diane Bryant

>> Well, exactly. No, that is my point of, yes, those people are flushed with cash. Yes, they can build that capacity out. It's a risk decision though. Do they believe that the demand for that will come, or will it be like the nineties when the dot com burst and all of a sudden you didn't need all that fiber and all of those people went defunct? So it's speculative. You're making a bet. As one of the hyperscalers, you're making a bet, whether it's Oracle or AWS, that that demand really will come, even though today those folks are not making money. Their consumption is consumers that are getting it for free, so you're placing a bet.

Dave Vellante

>> So one school of thought says maybe it doesn't blow up like the dot com because these companies are so flushed with cash. On the other side, if they dial it down, it's almost like the VC's funding Yahoo portal ads, and you remember Silicon Valley between 2001 and the Google IPO. It was like a ghost town. And so-

Diane Bryant

>> Unfortunately....

Dave Vellante

>> if we knew the answer to that, we'd be making-

Diane Bryant

>> Well, it'll be fun. Definitely watch. The build out is fun, fascinating. I mean, nothing more exciting. All of the hyperscalers building their own XPUs, or you see one by one. Obviously, Google started with a TPU. AWS has Trainium. Microsoft is deploying theirs. So all of that is happening. All that goes to Broadcom, on the board of Broadcom. All that benefits Broadcom, of course. But then you got to ask yourself what happens to NVIDIA's volume? If almost 50% of all compute in the world is within the hyperscalers, and the hyperscalers move off of NVIDIA onto their own XPU, what happens to NVIDIA? So that's I think if I had a crystal ball, they've got to build their own cloud. They've got to maintain that stickiness they have with CUDA.

Dave Vellante

>> And thinking about when you were at Intel, it was that virtual integration between Intel and Microsoft, in some ways, NVIDIA is Wintel because of CUDA.

Diane Bryant

>> Them both.

Dave Vellante

>> So to the extent that CUDA becomes the programming model, they have an advantage there. There are alternatives. I mean, AMD is trying to create an alternative. There are open-source alternatives. I mean, it takes a long time, as you know, to build that type of software and make it robust.

Diane Bryant

>> That is NVIDIA's stickiness or lock is CUDA. But if you're one of the hyperscalers, and the hyperscalers have 50% of the market, and you've decided I'm going to stop buying umpteen tens of thousands of dollars per GPU, and I'm going to move to my own, I'm going to pick up my software stack, and I'm going to port it. I'm not beholden to CUDA. Now, you're correct. If you're enterprise, if you sell the mainframe, you're pretty committed to IBM. You're not moving off. If you have CUDA in your enterprise, you're probably stuck. You're going to keep running on NVIDIA. You're not going to waste your time porting your app onto AWS or somewhere else, but that's not the majority of the market.

Dave Vellante

>> And that's why you see these neoclouds emerging, like CoreWeave and Lambda and others, and NVIDIA, of course, funding CoreWeave, even taken some of their excess capacity if they need to.

Diane Bryant

>> Well, that is smart. I think the other part of the build-out is not just the data center for the models. It's also the inference, all the compute that's going on at the edge. If you look at any of those hyperscalers, they are not just building their own data centers. There's leveraging all those colos. I mean, Microsoft I think is 20% is their own physical data center. 80% is leasing space out of all those colos. Why build all that silly data center infrastructure? So you're absolutely right. If for those folks, the neos, they built a data center, not just a colo, but they've actually loaded the infrastructure into it and they've loaded CUDA, well now NVIDIA doesn't have to go build that themselves. It's a great partnership. It streamlines the operations for NVIDIA.

Dave Vellante

>> I want to ask you about AI productivity. I know that's a passion of yours, and you touched on it before. I was listening to Michael Dell. He was on the BG2 podcast, Bill Gurley and Brad Gerstner, and he said, because there's a narrative that, wow, we're way overspending on AI and there's no ROI. There's no return. And Michael Dell said, well, the global GDP is like 120 billion, let's call it, and people talk about 10% improvement in productivity as a result of AI. So that's $120 trillion GDP. So 12 trillion would be 10%. People are talking 10%, 20%, even 30%. So his point was, look, we're probably spending 70 or 80 billion on CapEx for AI. We may be underspending. Now, to your point about the internet before-

Diane Bryant

>> It's an absorption. It's .

Dave Vellante

>> Right, it's the timing is everything.

Diane Bryant

>> It'll happen. Eventually all that capacity will be absorbed, but there is a consumption delay.

Dave Vellante

>> So there'd be some pain in between.

Diane Bryant

>> There'll be a little bit of pain in between.But on your productivity comment, so we talk about AI and the fact you hear a lot of naysayers saying that now AI will come in and it'll eliminate jobs, and that's the negative premise of AI and the fear factor, the fearmongering. To Michael's point, if you look at AI as an efficiency play, which I think we all would.

Dave Vellante

>> Sure.

Diane Bryant

>> You could take the low-level workloads or low-level work that you do and offload it to AI, it's clear. If it's an efficiency play, you think about Jevons paradox. Was does Jevons paradox do? Anytime there's efficiency, there's greater demand for that, whatever that capacity is that just received the efficiency. So when you make an engineer more productive, because you give him GenAI, he can now do more, and that more is your top line. That's your revenue. So you've released a limiter to driving your revenue, driving the top line. So you're not going to let that engineer go. You're now going to benefit from the fact that that engineer can do more, and I can accelerate my company. So it's beautiful. It's Jevons paradox. It's beautiful. And to Michael's point, it's going to unleash a lot of companies that now can be more efficient and deliver more in the same amount of time. The negative side of it is, yes, if you are in a function, a line function or a staff function, if you're HR, if you're legal, it is true that your job could be replaced or that organization will come down. So in that way, enterprise is going to get a benefit on the OpEx side. If you're a support function, a line function, you should worry. But if you are a function that is literally the core of the company... Even marketing, if you're in enterprise marketing, you could reduce the headcount out of marketing. But if you're a marketing company, now you've been able to take on more clients for the same amount of capacity. So it's very different. It has the opportunity to drive the top line, not just the bottom line.

Dave Vellante

>> Machines have always replaced humans, right?

Diane Bryant

>> Yeah.

Dave Vellante

>> First time-

Diane Bryant

>> Do you remember the nineties? I remember we had a man that walked around with a basket and delivered mail. Guess what? That man's gone. We got PC's. We got email. That man lost his job. People lose jobs.

Dave Vellante

>> People used to hang billboards. Okay. The first time ever it's cognitive function, so that's a little different, but-

Diane Bryant

>> But some people will lose their jobs. So we'll see as a GDP as a market, there will be acceleration because you've made your employees more efficient.

Dave Vellante

>> And more opportunity for all.

Diane Bryant

>> And more opportunity.

Dave Vellante

>> You mentioned before, and I agree with you, it's going to take the better part of a decade for this to play out, especially in the enterprise. You hear people talking about this is the year of agentic. It's not so easy. As a former CIO, you know. It's not so easy to get your data house in order, apply it. And then in the context of AI, take advantage of it to get competitive of differentiation.

Diane Bryant

>> Right.

Dave Vellante

>> Why is it so hard for the CIO to move at the speed of, say, NVIDIA or OpenAI?

Diane Bryant

>> Well, exactly what I was just pointing to, which is if you're SAP, deploying new AI features into that application, into that ERP, you're going to drive your revenue. It is your mainstream path to revenue acceleration. So you are going to do that fast and hard. Then if you're the CIO and you are looking around saying, well, how do I drive revenue for the company? I'm one of those staff positions that's going to get reduced, you're not in the core space, it's just harder. You don't have the whole company looking at you like if you were the business unit that's driving the product saying increase revenue, drive, drive, drive. You're the guy sitting back in IT going, I could lose my job. I don't know.

Dave Vellante

>> So it's a risk management. It's-

Diane Bryant

>> Well, it's a priority for the company. You're looking at an IT organization. That's a staff function. They don't have the expertise. They don't have the talent. You've got every organization in the company yelling at you, finances. They want this. HR says they need this. The engineers say they want this. It's a lot. And so it's a very slow slog to get IT to prioritize, make a selection, bring in the talent, bring in the capabilities, clean the data. It's a big effort. It's prioritization -

Dave Vellante

>> So that is-...

Diane Bryant

>> can drive top line.

Dave Vellante

>> That is the role of the CIO is prioritization, the business case, aligning with the business.

Diane Bryant

>> Figuring out if it could actually change the revenue trajectory of the company. Or else, why? It's just a burn.

Dave Vellante

>> And I guess understanding the degree of difficulty because there are probably some quick wins that you can get, but at the same time, if those quick wins don't add up to transformation-

Diane Bryant

>> Your quick wins come from SaaS. Your quick wins come from Workday integrating AI, from SAP and some Salesforce integrate, those are quick wins. All of a sudden, poof, I'm subscribing to SAP, and poof, now I have the capability to use a new feature. That's where the adoption comes, and those are the big winners. Then you have all the startups. If I want to take a little more risk and grab a Brightflag and fix my legal side. And then the long slog is enterprise, your own enterprise apps, your own enterprise data, but it's inevitable. With all that capital being spent, it's inevitable that AI will penetrate enterprise. It'll just take a little bit.

Dave Vellante

>> I want to ask you about vertical integration. Think of the history of tech. IBM was obviously highly integrated, vertically integrated. You had the peripherals around it, and then in your world, Intel, Microsoft, Oracle, Seagate, I mean, the industry became highly fragmented, and then individual competition occurred on those lines, although you had a virtual integration with Microsoft and you were vertically integrated with foundry. And hyperscalers are pretty vertically integrated, for the most part, even though they buy one of everything from everybody. They're building their own silicon. NVIDIA, in a way, is pretty highly vertically integrated, although they don't own foundry. AMD finally got rid of Jerry Sanders, only real men have foundries. Okay, and that worked with TSM. What are your thoughts on vertical integration, when it makes sense, when it doesn't make sense?

Diane Bryant

>> So you're spot on. In the old days, the Intel mantra was you build an open platform. You invite everybody to build on top of your platform. So we have the processor. We invite Microsoft to put the OS on. We invite Seagate to put the hard drive, and we invite Micron to put the memory on. It was an open platform. And from that openness, you got HP, Dell, IBM. They were all able to x86 service. There was no lock. You're absolutely right though. The world has changed. That model is gone, and now the hyperscalers is becoming a very integrated stack with them building their own GPUs. It is control, control, control. NVIDIA is the same. They had a GPU. They have the software platform, CUDA. They're building their own systems now. Next step is the cloud. From an industry perspective, it is completely flipped. It's no longer kumbaya, let's build an open platform for the world to innovate on. It is let me build my stack. And how do you do that? Scale. That comes through consolidation. That's why we have the top 10, the magnificent eight now, with Broadcom being number eight. The magnificent eight, the largest market cap companies in the world, they have built their proprietary stack. You can throw Apple in there too. So that is the way of the world today. There's no more play nice.

Dave Vellante

>> Is Dell an exception that proves the rule? In other words, you can actually take a relatively low-margin business and if you have enough scale, you can actually thrive with that sort of open model. Purchasing GPUs from NVIDIA or CPUs from Intel and AMD, software from whomever, bringing it through your distribution channel?

Diane Bryant

>> I think you have to look at who's in control. That's huge.

Dave Vellante

>> I think that's the key.

Diane Bryant

>> So when I was running the data center business, Amazon, Andy Jassy, I love him, so I hope he's... Andy would graciously call me and say, this is the processor I want. Build it. Sell it to me. And if you do, this is the server builder I want, say Dell, give it to him. Dell, you put that in there and you give it back to me. It's pure control. And the industry, we were all puppets to that hyperscaler. Whatever they said to do, that's what we did into their locked system. And, of course, they want proprietary. When Andy would say, Diane, I want a special CPU. Nobody else can have it. It's like you got your special... It's scale. You got it there. They've got the scale. And so you're right, Dell's done very well. I did very well because I partnered with AWS. I gave them a plan. I partnered with Google. I partnered with Microsoft. You can't afford to lose those big five, right?

Dave Vellante

>> Right, right.

Diane Bryant

>> So, of course, you do what you have to do in order to win them. Dell's been very good at winning the business of these hyperscalers, and hence he's doing very well.

Dave Vellante

>> So let's close on how you see the enterprise AI and the adoption playing out. It's early days.

Diane Bryant

>> It's early days.

Dave Vellante

>> I think you said 20% today, and those are probably the big banks.

Diane Bryant

>> Yeah, 20% of generative AI capacity, use, downloads, use today, 20% is enterprise, 80% is consumer. And so to get enterprise now to adopt, it is change management, which is such a boring term. I hate it, but how do you get people to change? You've been doing the job the way you do it for 10 years, and I'm telling you, no, use this generative AI tool? I mean-

Dave Vellante

>> That's not how I do it....

Diane Bryant

>> you know it's not how I do it. And so when trying to drive change inside of a company, you always have to get the tipping point employees, those employees that are eager to jump in and try out generative AI and do their job differently. If you can get the tipping point to do it because they believe in it, they believe there's upside to the company, they believe they can be successful, they're not going to get fired because they're using GenAI because now they're not needed, their job isn't needed, they believe in it. You get those folks and either the rest follow or they leave. It's self-selection at that point. So it will happen. I mean, there's huge buy in. Like I said, the efficiency, especially if you're in the business unit line and you're the one that can unleash more revenue, top-line growth for your company, it is all upside for you. You shouldn't be afraid of generative AI.

Dave Vellante

>> Lean in, lean in-

Diane Bryant

>> Lean in....

Dave Vellante

>> into enterprise AI.

Diane Bryant

>> Lean in. Lean in.

Dave Vellante

>> Diane Bryant, thank you so much for coming into our studio. I'd love to have you back-

Diane Bryant

>> Thank you. Thank you....

Dave Vellante

>> in future episodes. So thank you.

Diane Bryant

>> Lots of fun. Great talking to you.

Dave Vellante

>> Okay, thank you for watching. This is Dave Vellante, John Furrier as well, for, "Data Center of the Future: AI factories." TheCUBE plus NYSE Wired from the New York Stock Exchange. We'll be right back right after this short break.