Under Pressure: Can data centers keep up with AI?

Download Transcript

Spencer Levy
In the U.S. And around the world, the growth of A.I., along with other drivers, has been the source of an important real estate story. It's about demand, development, and sometimes drama around data centers. By some estimates, there's literally a 0% vacancy rate in the space, creating urgency and challenges for users and developers. On this episode, we dive into the fray.

Andy Power
The demand trend is moving at the speed of light and we're living in the physical world. And to design, build, permit, electrify this infrastructure, can't keep up with it. 

Spencer Levy
That's Andy Power, the President and CEO of Digital Realty, a platform with nearly 6,000 customers on six continents, from large multinationals with routine co-location data needs, all the way to hyperscale cloud computing providers. With the emergence of A.I., the company has grown to an enterprise value north of $80 billion, with a nearly $17 billion construction pipeline. 

Pat Lynch 
We're bringing record new supply globally to the industry, but it's getting taken up just as fast and we're doing leases three years out. 

Spencer Levy
And that's Pat Lynch, an Executive Managing Director for CBRE's Data Center Solutions. Based in Denver, Pat spent half his 30-year career in digital infrastructure at CBRE, and he's now the global advisory leader for the company's data-center business. Coming up, digital realty, data centers, and a look at CBRE's powerful new report on the sector and its impact in the world. I'm Spencer Levy, and that's right now on The Weekly Take. 

Spencer Levy
Welcome to The Weekly Take and we are talking about maybe the hottest topic in real estate today, data centers. And who better to talk about it than Andy Power, the CEO of one of the largest owners of data centers in the world. Andy, what a pleasure to have you here on the show. 

Andy Power
Thanks, Spencer, delighted to be joining you. 

Spencer Levy
And our old friend, Pat Lynch. Welcome back to the show, Pat. 

Pat Lynch
It's a real honor to do this with you. And Andy, I can't thank you enough. I know how busy you must be. So to join us today means a world to me. So thank you so much. 

Andy Power
Absolutely. 

Spencer Levy
Well, we're delighted to have both of you. And Pat, I know that we just came out with the Global Data Center Trends Report, 2026. The trends were unbelievable. Frankly, in addition to strength in Europe and Asia, am I correct in saying there's no vacancy in the United States or almost negative vacancy? I mean, that's how much demand there is today for data centers. 

Pat Lynch
Yeah, it's essentially zero, Spencer, right? You've got a little bit of transition and I would make the argument that what little vacancy is actually in the report – it's accurate, of course – but it's probably older space that may be level three or the original version of digital or some of the successors of Purdue. So it's unbelievable. And what jumps out at me is a couple of other things, right, the continued growth in pricing. So we think about the expenses and it is getting more expensive to build the assets. I know we'll talk later about cost of power and cost of construction and equipment, but pricing continues to trend up in the mid to high teens and given the continued demand, I don't see that changing for a couple of years. And the other piece that jumped out is the amount of absorption. We're bringing record new supply globally to the industry, but it's getting taken up just as fast and we're doing leases three years out for product that's going to deliver. So I think it's unprecedented from a standpoint of continued demand pricing increases. It's a pretty exciting time. 

Spencer Levy
So Andy, as long as I've been following real estate, even though we wanna have no vacancy, that's not always the healthiest market, having 0% vacancy. We wanna have like five, six, seven percent. That's when I knew New York was healthy, okay? I grew up in New York, real estate. That was like our magic number. But 0% of vacancy means there's some people who can't get data center space today. And some of those people might be A.I. related, which is obviously the hottest of hot topics today. Are people being left out and how do we address it? 

Andy Power
It's just a phenomenon that the demand trend is moving at the speed of light and we're living in the physical world. And to design, build, permit, electrify this infrastructure can't keep up with it. So if we were able as a country or as a world to bring this infrastructure on faster, I think you would see it fill up faster, right? Now, customers have their own equipment that they're moving in. So there's always a chance that if we caught up that the supply chain for chips could have a weakness. But I just think basically technological growth innovation has been restrained. We see it in our business when the big customers are literally competing with each other for that next block of capacity in the markets where they need it. I'm sure there's some form of pecking order happening here where customers that can move quicker than others, customers that can scale. Customers that are not newbies to scaling their infrastructure with us are probably getting first-mover advantage. But at every twist here, you've seen a new unlock. Just at the beginning of this year, Claude’s release basically opened up this tremendous envelope of coding opportunities. Now, up until that point, everyone was talking GPUs, GPUs, GPUs and training for A.I.. Well, Claude basically said we need CPUs for coding, and we could do more coding, and we can do it faster, more efficient, and deliver more innovation. So I think that this is technology we're trying to support and we're doing our best. And probably the silver lining is the fact that the demand is outpacing us probably is gonna keep us from driving the quote proverbial car off the road as we scale this infrastructure so quickly. 

Spencer Levy
One of the things I really noted was the incredible development that's happening in southern Europe and in South America and Latin America. And I know you have product there. How do you see developing outside or buying outside the United States today, Andy? 

Andy Power
What's interesting is data centers has kind of been thrust into the mainstream media, but there's so many layers behind what a data center is, where it's located and the use case, it's really serving. Everyone wants to talk about A.I., but we've got this tremendous digital transformation happening for enterprisers coming out of server closets, office buildings, closing down data centers, old telco facilities. Cloud is this called half a trillion dollar business growing 30 plus percent. I personally believe there's so much A.I. focus now that cloud growth is being restrained, i.e. It's not meeting what its full potential could be, given these big hyperscale customers have to decide whether they're growing to grow their cloud business or service the cookie monsters of the A.I. labs who are just gobbling up capacity. In those two markets, you just mentioned two things that happened that are really recent and relevant. Digital. In Europe in particular, has been called, putting together this string of critical nodes across the Mediterranean. It really goes back to where it started with Marseille, a company we had acquired and literally took a submarine, call it, repair Shop from World War II and turned that into, called the Connectivity Hub of Europe, subsea cables and networks coming in there. And now we've been replicating that with data centers coming online in Barcelona, in Rome, entering Milan, entering Lisbon, a subsea cable project in Crete. And you think of times right now when we're in this tremendous conflict and you've got straights being closed and you got a lot of volatility in the world, these technology companies are thinking about, we need different ways to communicate, get our innovation and technology spreading. You flip over to Brazil and what you're seeing there and at least in our portfolio, because we had a very good start to the year with our business in Brazil is the real globalization of A.I. from a customer base and a location base. Cloud globalized along the way with data sovereignty, took some time and then started to ramp it at faster paces outside the U.S. than it was in the U.S. A.I., the phenomenon of A.I. has not followed that same path to date. There's still just such a concentration of A.I. infrastructure happening in the United States and it's intuitive. You're bringing on infrastructure where you can do it the fastest and we're built for this kind of infrastructure growth. But now we're starting to see it in our portfolio. Both Brazil and Japan were two of the biggest markets outside of the U.S. In the last quarter. And you're seeing customers going there, yes, for cloud growth, but also A.I. growth. You're seeing that from the U.S. hyperscalers, but also some of the Asia customers as well. So you're see this infrastructure build out really starting to spread its wings on the geos. 

Spencer Levy 
So, Pat, let's get a little bit basic first for our listeners, even though this is not our first data center episode. Just walk through the math of how people lease space in the data center space, what the different types of data centers are. Let's start there. 

Pat Lynch
That could be a podcast in and of itself. 

Spencer Levy 
Yeah.

Pat Lynch
So let me, I think if you start with carrier facilities, right, very network centric, that's the interconnection hubs that digital developed and a couple of others that I think have driven and will continue to drive huge value to the industry. So that's kind of the main and main intersection as you move out, you've got colocation facilities. That tends to be in major markets. We still have a number of enterprise facilities that exist today, which I think is surprising to some people. And then if you move to what some of the recent expanses of these gigawatt, multi-gigawatt facilities, they tend to fall into more remote areas, oftentimes where they can find power. So you think West Texas, Wisconsin-Canadian border, Pennsylvania, et cetera. And the pricing structure really goes back to power. Right? Where can I find the power? And then in the case of a digital building, you know, there's a powered shell lease. There are turnkey leases. Turnkey means our clients can come into a digital facility depending on how the deal is structured. put up their racks, put their cabling, but the electrical and the cooling is all there, comes into the space. And their monthly rent is based on, call it, you know, $175 a KW per month plus the power that they utilize. Hadyou asked me this question two or three years ago, s hundred and seventy five – keep me honest Andy – but it would have been in the low hundreds or maybe even below a hundred for some clients. So maybe that is an example of the amount of rent growth that we're seeing in this space, but that you're basically renting power. And I think the other dynamic that's taken place – and, Andy, I will ask you to keep me honest on this – three or four years ago, the majority of our clients were taking racks that would I'd say in the 10 to 15 KW rack range. There's still a lot of that out there, But we now have clients that are building. 100, 200 KW per rack, which is a completely different design. Let me step back. It's a different design model, often requires liquid cooling, but the complexity of the industry has gotten significantly more challenging, which I think is why you see the largest companies like a Digital Reality and a few others succeeding. The local regional players are almost non-existent today. 

Andy Power
We had a phenomenon where things were pretty darn good for data centers for a while, right? Since the dot-com crisis, you had these multi-layers of demand for enterprises moving to purpose-built infrastructure, uh, these carrier hotels and the connectivity to them for the enterprise. It was about, call it, performance, efficiency, cost reduction. You had the cloud come on the scene. People are going to go to the first cloud dial up credit cards. And now that you know, in multiple clouds. And all along, most of the time, the supply could kind of keep up with it. Not every second away, but it was keeping up. And I don't think what you didn't really see oversupplied, but you saw this also happen when capital started saying, Hey, this is data center. This thing's looked like a good investment opportunity. We had the ZIRP era of interest rates called quickly going down. So capital rotating from, Hey I'm an opportunity fund. I shoot for 20 IRRs. I invest in data centers to the likes of real estate and infrastructure funds. And then what happened was these demand trends compounded upon each other, in my opinion. A.I. was supplemental to that, in addition to incremental demand, and the supply ran out of gas while interest rates accelerated. And all those things, call it, along with what Pat said, we're not selling small t-shirt sizes anymore. It's all the mediums became larger and larges became extra larges and everything keeps getting bigger, has really pushed this to a place where the rates have really accelerated. And what happens next is probably the question. Does history repeat itself or are we in a new paradigm? I believe based on the fact that we're just still getting going on some of these demand trends of build out, this is going to be multiple years of what's required. We're in an inflationary environment right now as we build because everyone's trying to build as fast or so. Basis are increasing. And lastly, at least for digital, which I'd say is everything Pat described around data centers we do, except for probably the more hinterland one-off projects for one customer, we're focused on markets that have like locational sensitivity. Those workloads cannot just up and move across the country. And in those markets – be it in Northern Virginia, Chicago, Dallas, a Frankfurt, an Amsterdam, a London, a Singapore, Tokyo – the supply constraints I don't see reversing due to things like the electron delivery, the permitting, and the people. 

Spencer Levy
So let me dig into that point for just a moment here, and I'll use a term that we hear in this space quite a bit, which is latency. And my understanding is that some industries require being as close to other data centers as possible because the speed of light – and thank you for Albert Einstein for defining that one for me – but the speed light actually matters when you're in trading. It actually matters in video games. It actually matters for crypto, but it might not matter. For certain technologies. And the one that I've heard is A.I., you could have a remote center, which does a lot of your work and it's maybe not as critical to be latent. How do we determine where do we put these and who the user is? 

Andy Power
 There's this concept I would call like data gravity, which–let's start with latency or location of the workload. And to me, that is a confluence of where, call it, data centers have grown and the snowball compounded as well as the architecture of the cloud. So when you are a consumer of cloud, when CBRE is accustomed to cloud, your CIO, your CTO is using Azure, or using AWS, using GCI. You are picking a region to put your data. You want it in the IAD, which is the airport code or acronym, meaning it's in the Northern Virginia pocket, right? Or you want it, in Santa Clara, or you want in Chicago, right. In those markets, the cloud then architected and AWS and the others pick this leg to their stool, this leg of the stool, and this leg of their stool and they have radius restrictions. And that is an incremental redundancy to the actual redundancy, resiliency of the data center, i.e. someone's jackhammering the road, they take down power, take down fiber optics, cloud zone goes down, somebody does the wrong thing in the data center – but you got the backup through the three legs of the stool. Right. Now think about–I just gave you an example of one cloud provider. Now we know there's several cloud providers just here in the U.S. let alone other multinational players. Then you have Enterprises who don't put all their eggs in the cloud and want to put hybrid infrastructure and off more often than not, they're putting it in some physical proximity to where their cloud is and call it these markets that build upon each other again and again. That's why in my view, there's a massive addressable market we're serving and then is growing at a great clip that is going to have to stay in these hub markets. Step over to A.I. I agree with you today when people are running anywhere for any type of data center capacity, they're essentially saying, for A.I. training, I can put this anywhere, just get to it as fast as possible. These early days of A.I. inference, I don't think is what the world's gonna look like down the road. I think A.I. inference today is a glorified new version of a Google search. And Google improved its own product, basically. When you search now, you get A.I. at the top of your feed, right? But when I think when you get to a world where you have actual workloads that have private data have real critical uptime requirements, I think that the major markets maybe will be expanding, but the major market’s in proximity to those things I mentioned of that data center of data gravity and can be incrementally and incrementally important. I don't understand why, and you can go back just a few months, one of the big cloud providers went down for 12 hours. And all these businesses couldn't process transactions, couldn't do workflows. Why is that gonna be not equally or not more important for A.I.? When you're trying to cure cancer, can you really accept an outage? Can you set up the downtime, right? When you’re trying to use robotics in a factory, in an unsafe environment, when you're literally using this technology to save lives, I just don't see how we're not going to need that proximity to connectivity for a piece of this A.I. inference, especially when it comes to corporate data sets in enterprise use cases. 

Spencer Levy
Let's look at the other factors at play here of location decisions, right? So when we do location decisions in all forms of real estate, we have the same basic criteria. Where's the best labor? Where's the best infrastructure? And in the case of data centers, you're looking for the best water, the best power, things like that, right. Pat, how do you see the location decisions? In addition to Andy's point of view, where the major markets are big and going to get bigger, are there other places – not just because of resources, but because there are differences in law? Like when you go cross-border, there's difference in data and security. How do you see it, Pat? 

Pat Lynch
I truly believe, and I believe this because it's coming from our clients, right?  It's the financial service clients, it's the credit card clients, in some cases universities and healthcare. What we put in these large training models remotely is very important. But in order for us to actually utilize that technology and get a return on that investment, by the way, that is going to have to get to our phones, to our glasses, to our homes, to automobiles, to commercial drones, and that is not unique to the U.S. I just say that the U.S. is ahead of the rest of the world and getting some of these large models built. Building those large models in other parts of the word for the reasons you're talking about, data sovereignty, et cetera, is even more challenging. So there's a lot to the question you ask, but the piece that I have concerns about is our ability to add 50 to 150 megawatts for a facility in Manhattan or L.A. or San Francisco or Denver or Washington, D.C. We're going to need power to support the facilities that Andy and his competitive subsets are gonna build. And I don't know that as a utility industry, and maybe we'll have time today to talk a little bit about the pushback, but all of that infrastructure that Andy's company is building, and there's a fiber infrastructure and there is a wireless infrastructure that's gonna be critical as well. That part actually is a significant concern of mine, Spencer, because it is going to be necessary to get the full utilization of A.I and I don't know that as industries that we've kind of picked up on that yet. 

Andy Power
I look at what's transpired to date and the major data center markets have been tried and true for several years. At the same time, certain applications by hyperscale customers have not fallen in those markets. It's well published. Some of the more B2C applications from the folks in the social media verticals or obviously Google as a multifaceted business has cloud, but also has a very big consumer business. They also put data centers in places that have lower costs of land, labor, water, electricity. And these are the most sophisticated buyers on the planet for the last two decades. If they could have moved those data centers that they clustered in Northern Virginia or Frankfurt or Paris or Singapore and put them all where they had data centers that was just cheaper, they would have done that. Like, there's no question about that, right? So I think these architects have mattered. We're seeing a little bit of this history repeat itself on these early days of A.I., where it's very, in my opinion, B2C application. Meaning business to consumer, and the consumer can just tolerate the latency when you're using your Chat or your Claude and saying what restaurant should I go to tonight or compare this or build a spreadsheet and you just wait. But I see what going to what Pat says, you're gonna see more of this stuff migrate to proximate markets that have these data gravity phenomenons. Now, what could change, you're going to see more markets like that. I mean, we in data center land, Pat, even in the U.S. you've been talking, and what about four-ish, five-ish markets for a long time, and then, oh, Phoenix became a big, big market. Oh, New Albany, Ohio became a market, right? But percentage-wise, it wasn't a big movement for a long time in cloud, like new cloud markets really opening up. We're seeing the Atlanta market at 13 megawatts of absorption for the longest time, and now it's on pace to be one of the largest markets in the country. We had our largest lease in the history of digital realty last quarter, it was in Charlotte. So I think you could see more other markets be relevant. But I'm from least how we're steering the business of Digital reticent of just chasing off to a location that has electrons to build a one-off almost islanded one customer infrastructure given the longevity of our investment essentially 

Spencer Levy
What I love about what you said here is you got right to the bottom line. You say, oh, we're the biggest or one of the biggest owners of data centers, but we're very disciplined about where we're putting them, or where we are investing. We could follow the train to stranded power in old nuclear power plants and really tertiary locations, but they may not have the long-term durability. They may have short-term value, but in the long term the durability is going to be in these clusters. 

Andy Powe
You hit the nail on the head. I mean, I have no fears of building a data center, leasing it, operationalize it, supporting it in those more hinterland locations. To me, if it's only going to be able to support in that location, an A.I. workload or a workload that does not have any locational sensitivity, what happens when our lease comes due in 15 years, right? I really feel you're putting yourself in a place where the customer could shop all 50 states. And that's essentially why we, where we focus on markets with robust demand, but diverse demand and really locational and latency sensitivity to the workloads we're supporting. 

Spencer Levy
So Pat, let's turn back to the Global Data Center Trends, 2026. We touched on Europe. We certainly spent a lot of time on the U.S. Didn't spend as much time on Asia. Tell us what's going on over there. 

Pat Lynch
Yeah, I think it's relative to the rest of the globe, it's still an emerging region. Japan, that's a significant market opportunity for us. Lots of demand in Singapore, zero space, no vacancy. So that's created a strong market in Malaysia where many large developers and large clients are landing. We view India as hugely opportunistic as well. The three M's of Mumbai, Malaysia and Melbourne are all drawing a lot of attention, predominantly driven – early stage, remember large training models – driven by the availability of a large capacity of power. So Asia is a fascinating market, significant upside there, driven by of course significant demand that needs a place to land. I think the other piece globally that's impacting it is the, you know, where are we gonna allow data centers to be built? There's a component of pushback that is growing. 

Spencer Levy
Candidly, there's a real split of opinion here, because at the federal level that it's all full steam ahead at the state and local level it's fair to say the pushback is significant. What's your reaction? How do you interact with folks that may have a different point of view? 

Andy Power
At Digital Realty, we're trying to take this to a personal level and having our people out in front of folks that are operating in those communities and make sure our story, when it comes to electricity costs, when it comes to use of water, when is comes to jobs, when it comes to taxes, that on each of those elements, we think we're in the right place to be the right place for this infrastructure. We get out in from of folks in town halls. We open the kimono and say, hey, come tour our data system. Let me tell you, these are mission critical workloads. Your emails, your streaming, your workflows, your buying your coffee, your first responders, the scientific research, all that stuff doesn't happen without our critical infrastructure. When it comes to electricity, it's a pass through to our customers who all stood up and said, we wanna pay more than our fair share. We at Digital Realty have been investing in substations and other equipment to keep that grid more affordable, spreading those fixed costs. And I can tell you, in those hot summer nights, like we're kind of facing in the Northeast right now, we go back to our backup generation and give that grid a buffer or break to make it more resilient, which also is a cost savings to the constituents. When it comes to water, probably my favorite, Digital Reality’s 300 plus data centers around the world, we use less water than 18 – one-eight – California golf courses. There's 15,000 golf courses in the United States alone. I'm not trying to pick on the golf courses, but for anyone worried about water I think there's an asset class ahead of us in line to be worried about that. When it comes to jobs, tour around a data center market see the amount of skilled trades – electricians, engineers, great jobs that are scaling infrastructure – that's going to take decades to build out. This isn't we build this in a month or a year and we're gone type labor. Then when we operate these buildings, these are great jobs. We're hiring from former military because if you could work a submarine, you're a great job for you is to be able to work the critical infrastructure of a data center. And for every job inside of one of our buildings, the statute of four and a half jobs are happening in the general economy. And don't trust me on this. Call up the county supervisor in a Loudoun County or another market that has been navigating this for many, many years. This can be done in the right fashion. And I believe Digital Realty has been a part of that along the way. 

Spencer Levy
Well, let's talk about what comes next. We're talking about the existing clusters. We talked about certain clusters or maybe locations far afield. We didn't even address the potential for space data centers. So Pat or Andy, how do you address what comes next? 

Pat Lynch
I'll go first, and I alluded to it a little bit earlier, Spencer. First of all, I think it's wildly exciting. Is it going to be data centers in space? Three years ago one of the big hyperscalers was talking about building data centers in the ocean. That didn't quite yet evolve. So I wish I was smart enough to forecast what's going to happen there. What I do believe passionately is we're still early stages in this edge model, whatever that ends up being. And now, everything we do in our day-to-day lives involves – several things we do – involves electricity and this future is talked about A.I. in that capacity right what ultimately what we're going to do with it is going to be way beyond at least my ability to comprehend but it's going to be cool and it's gonna require a lot of infrastructure not just these mega facilities in the middle of Texas but in major cities the fiber optics, the rooftops, you know, technology to our houses is going to look very different. So to me, to have spent my career, going off 35 years now in this space to see it evolve is just unbelievable. I'm very fortunate and blessed to have had that. And I'm, I'm excited to see where this thing goes next. 

Andy Power
So maybe to hit your space question for a second. And here at Digital, we're already supporting the leading global satellite communication customers because they're anchoring back to our data centers every minute of the day. When I think what comes next, I think of more and better as two words. And I think it's more workloads, more applications, more technology, more customers. We're just living in an era where two companies you've never heard of a few years ago we're gonna be going public in the trillion-dollar range. And I think that speech to technologies is allowing this incredible innovation and growth happen before us. I think that means this Jevon’s paradox of one thing opens up the ability to do more and grow even faster is clearly a continuing theme. And you're gonna need the physical infrastructure, the data centers to provide that. Probably more data centers and more markets. We can do it the right way and in the right places in the right framework. I think you're gonna see more institutional capital following about data centers. It's kind of funny. I'm part of a 20-plus year public data center REIT and we're still the new kids on the block, the niche asset class that folks are still trying to get their head around. We’ve just gone through a pandemic that challenged the necessity of shopping centers and office buildings. People are scratching the heads who actually wants to own a data center down the road and not understanding that that technology is what could stop the next pandemic from happening and is what is upending the retail segment and changing the way we live, work and socialize. I think all this stuff is more and better ways of living, right? We're going to be living healthier, safer lives, longer lives. And listen, I know there's all the doomerism around this, but I'm just pretty darn glass half full on the productivity we all are gonna be able to have, enjoy and the mundane things technology will help us with. So I think data centers are really at this heart of the tremendous growth and in a tremendous way. 

Spencer Levy
Yeah, and I would just echo that comment, but I'd also just say, look–what was the term you used there? I hadn't heard that term before: When new technology opens the door to do new things?

Andy Power
So Jevons paradox is the concept.

Spencer Levy 
Jevons Paradox, okay, well, you know, first time we've used it on the show and I'm going to dig into that. But I like that because people are afraid of new technology because of all of the disruption it causes – but it actually opens more doors than it closes and this is not just space. You could have said oh no, there'll never be a space data center No, you didn't say that what you said was it's complementary and in fact creates more demand for what we got here. Pat, how do you see it.

Pat Lynch 
Well, first of all, as a frequent listener, I think that's the first time anybody's brought up something that you didn't know about. So kudos to Andy on that one. No, I think, Spencer, we have to be careful with any kind of new emerging global technology. But the good, in my passionate opinion, is gonna far outweigh, and there's just so many things that is gonna come from this that I think will be outstanding. I've now been fortunate to be a grandfather, as of about 17 months ago.

Spencer Levy
Congratulations

Pat Lynch 
I'm super excited to see what technology will mean for Olivia James Lynch when she's 17 or 18. It excites me way more than it makes me nervous. 

Spencer Levy
So on behalf of The Weekly Take, what a great conversation today with Andy Power, the CEO and President of Digital Realty, one of the largest owners of data centers in the world. Andy, thank you so much for coming out today. 

Andy Power
Thanks so much for having me guys. 

Spencer Levy
And our old friend, Pat Lynch, executive managing director, Data Center Solutions, did a great job talking about the space. 

Pat Lynch 
Thank you. 

Spencer Levy
And our new report, the Global Data Center Trends Report, 2026. Check it out.

Spencer Levy
To find that report, you can check out our website. We'll post a link at CBRE.com/TheWeeklyTake, and it's also available at the CBRE Insights page as well. Once again, it's called the Global Data Center Trends Report for 2026. Our guests also mentioned all the jobs being created by the data center boom. CBRE is proud to support an innovative program with Meta that trains technicians to install fiber and other equipment at data centers, creating solid careers for U.S. workers in an area where there's an acute shortage of skilled labor. The program's called Level Up, and we're working to bring you more on a future show. In the meantime, there's info available at CBRE.com/LevelUp. Thanks for joining us. I'm Spencer Levy. Be smart. Be safe. Be well.