Ways AI Data Center Capacity Will Reach 90GW by 2027 & How AMD, WDC Benefit

back to the Watch List. I'm Nicole Petallides. Time now for our watch list panel. And we're discussing the latest, latest in centers and the build And wes been contributing to GDP. Lisa Martin business and technology analyst, Lisa Martin media Group and Ron Westfall, Vice heat networking and infrastructure hyper frame research. Glad to see you both with us. Goldman Sachs has been making some comments pertaining to this build Lisa, I'll start with you. How did you interpret it? W do you think it mea Great to be with you, Nicole. Thank you so much for having me. I think the GoldmanSf things. The data center boom. It's becoming bigger. We knowthd it's als becoming a lot more physical. On the size front, we're moving from a smaller to a our campuses that are measured not in square footage, but hundreds of megawatts eventually. Soon gigaw On the bright side, this is no longer just a GPU story, right? Data centers pull i around them. This entire ecosystem, right? compute, storage networking, also cooling power, etc. And the constraints, those are changing as well.and and capital for sure. Are t We've known that for a while, but weknow that on the constraint side, power grid connections, we're also seeing water. Andcan't forget the community kind of ance. All of those are imp really the speed of the buildout. And what we're seeing is that this is really be a massive industrial for investors, that both. A broadens the opportunities and the risks, the massive capital expenditures is something you're watching as well Ron, tell me morebout this. As you think about this, the communities areng back to a certain ext but these data centers are getting bigger and bigger, aren't they? Yeah, I think the momentum is already established and it will continue to Lisa's point. Iink s goingo result in, you know, these five gigawatts of capacitynto the end of this year. And that is being fuele primarily by just the hyperscaler capital expenditure l pr reach $750 billion by the end of thisear. So we're really looking at what is driving this. Well, part of it is the fact that the high density AI infrastructure is going to be required to continue the training for the fron models out there. And also inferencing is becoming more and prevalent, and that will actually continue to leverage what's going on at the data centers, but also more distribu throughout the edge. So that alsoives more opportunity f you know, enabling AI capabilities throughout the ecosystem. And I think it's also important to add that when it comes to the US data center market, specifically, that we're looking at a quarter of 12% over the next three years through 2030. And that, I think aligns with the Goldman Sachs estima And I think the reasons for that is despite some hiccups and despite, some say, l community concerns, there'll be plenty of offsets. We're seeing already that the data center bills are impacting election cycles, for example. And so I think this is something that will continue to play out where the growth is sof there are some, I would say minor opposition factors going into this and constraints, this is n going to stop the overall momentum, right? Because we couldee some slowing from community opposition or whatever. But you both seem to be on the camp there. We'll that growth that's inherent and continues here. When we think of naming comes to your mind in theat leadership and market? A le of td this is no longer just the GPU race, AMD and Western Digital. AMD makes this much more interesting because it isn't just a number two GPU ate story anymore. I think AMD's data center business hit close to 7 billion last quarter. And then within the last year, we've seen meta open AI, anthropic they've all committed up to, I think it's 14GW of AMD GPUs combined. We know that r efo becomeart of the economics here. But I think the customer question is how much can AI can we produce from every wat from every rack? And I don't think AMD has to win to replace Nvidia. Itneeds a bigger piece of a rapidly expandingmart front, AI doesn't just have a compute problem, right? We're talking data economy. And so of course it's also it's only going to continue, which meanssf under r underappreciated component of the ecosystem training. As Ron was talking about training inference, theyt. So compute heavy. But as we get deeper into inference agents, what's going to happen is this data flywheel, right? Continuously being created. And then so we're not just creating new dathink what's going to happen is historic rprise date pulled back in for its value. And that's a tailwind for Western Digital that I think is there. It's getting more and more exposure to the hyperscale cloud AI storage demand. And you know, the more data we have, it has to live somewhere in Western Digital, enter the picture and wenk about, you know, what we need with A OpenAI's annualized revenue that came in approximately 20llt has hit many of the names, including Western Digital. That was one that sort of went into that story. And Western Digital, which had run up to nearly $800. I pulled that up because I was looking at s those names because they've been beaten down already this week. You were lookinge same names here, Ron. These are the story AMD that you talked about gaining market share in Western Digital and some of the concerns there. You know, will we slow or Voting for growth. And I think there's in addition to what Lisa has already insightfully shared is the fact that AMD, for example, acquired World Labs for $8.2 billion. At acquisition alone, I anticip will, quite simply, make the company sm that it will be able to have better understanding of the spatial intelligence an frontier AI model devnt requirements to enable them to, quite simply, further their sale of the portfolio. And that portfolio, but certainly their CPU side. And I think that's going to be important in looking at the overall AMD pict and whylready they have projected that they're going to hit some major numbers over next several quarters. And I think in an to that, that they brought in ZT systems. And I think that's going to be ant in terms of having the again, the intelligence necessary, the design, I would say, capabilities to ens that their portfolio will adapt to the evolving needs of frontier AI models, and particularly on the Western digital side. I think it's not necessarily unio them. It's really looking at the overally space, and that certai a segment that's going to grow dramatically. I see Western Digital as having about 12% depending on how you estimate the m But when you're looking at Samsung, SK Hynix as well as micron, all of these players are going to continue think, have very impressive financial results. at includes, for example, you know, hitting margins that are up in the 80% range. So Ith, and I'll characterize it that will hav know, a very, I would say near- I would say impact. But when you're looking at, you know, the overall long term picture,ie growing borrowing, you know, some meta events. So this, I think is going to be really good news for the overall ecosystem. think about Agentic A too. I was just g at Palantir from Goldman Sachs. They actually talked very positively about Palantir, but just it to the big picture of the AI agents and what might happen there. Lisa, as you look at that part of the story too, right? We're looking at the data centers and we're thinking about some of tig names that we've mentioned. I mean, Nvidia has been off the charts lately too, with just all the deal making that it's been making with all these different names. I mean, even we have it in the news. So what do you think about Agentic AI and inference and all that? Like I was saying before, they a training gets all the headlines, right. But as we look at this emergencend not even the emergence, it's the growth of AI. It is conference I go to, it is one of the leading stories. And so as companies get lean more into inference and agents,s I talked about, it's the d story. It's how much data can we produce, how much canhe physical infrastructure support Because many companies and I hear this across industries, Ronnie, youpe leaning into agents treating them as they should, kin like interns or employees to make sure that what they are identifying is what the workloads that we should be automating, what are the workloads that will help augment people. And Agentic AI has has a front seat at the table. And, and everyone that I'm talking to is leaning in. Yeah. And your f big picture, Ron, I wanted to see what you project now and maybe in the years ahead here. Definitely now I think is to be integral rea to the overall workflows of organizations out there. And that's cer becoming clear on the enterprise side. Yes, there is still a l tire kicking going on in terms of, okay, how soon can we entrust agents to really, you know, carry through these requirements with guardrails built in? And I thinks going to be the key thing. I think what we're seeing actually is the ecosyst benefiting from a increasing emphasis on how to makeic safe, not just for the organizations, b consumers and for society as a whole. And that's going to be good newlly, for the additional players that we've already touched on. For example, we touched on the hyperscalers, but the AI networking players, Cisco, for example, comes to mind. They're leading, I would say, with the genetic ops port enhancements that are purpose built to address this very cha. You know, how can an organization have the visibility, the observability, the security, all of these factors built innto a, I would say, platform that enables them to have the confidence to adopt agent thus enable force multipliers for the existin workforce, but also, you know, continue to invest in them and avoid, you know, some of the concerns about silos. I organizations don't repeat the errors that have occurreh traditional network build outshere you have silos. And it's difficult for, say, one agent to talk to an agent. And so this is where I see MCP implementations as well as emerging agent to agent implementations making a huge difference. And Ink that's going to fuel and drive more agent tech implementations and there continue driving. The AI data center build we've been talking about. Yeah. You talked about the AI infrastructure for training and widespread inference workflows. There's a lot here, but it's this CapEx spend that is real and it continues. You could say whether how muc is or if it's slowing or, you know,f it's doubling, etc. The pace is one thing, but the fact is it is there and it continues to grow. Thank you both, Lisa Martin and Ron Westfal great conversation about some of