Jul 21, 2026 · 2 hr 16 min · 10 segments
In this episode of Purpose Unlocked, Leslie and Jennifer are joined by Nathaniel Gabriel and Jay Dobson of the UBS Chief Investment Office, along with Bridget Grant of Moore & Van Allen, for a timely…
Nathaniel GabelGuest
Bridget GrantGuestJennifer WinstellHostLeslie KinchlowHostUm, [clears throat] as I mentioned, we're gonna talk about data centers and what is driving this unprecedented demand for data centers today.
Um, I thought we'd start with you, Nathaniel, and, um, we have some questions teed up for Bridget as well, but let's, let's get the conversation started.
So, you know, obviously there have been data centers for decades, um, but what we're seeing today really began with the launch of ChatGPT almost on the dot in November of, of 2022.
Um, before that, for, for basically three or four years, we were building around $8 billion to $10 billion worth of U.S. data center capacity per year.
Um, but after ChatGPT came out, uh, you can look at the graph of data center construction values, it, it just turns into a hockey stick, uh, and it surges well above $30 billion.
So, uh, you know, that, that GPT moment, uh, really kicked off a global arms race to advance these, these AI models, um, you know, definitely for commercial use, but also increasingly for national security purposes.
Uh, and we've seen all these big tech companies, the hyperscalers pledge hundreds of billions of dollars of annual spending on, on data centers to train AI models, but also to respond to user inquiries.
Uh, and so, you know, those giant AI training facilities are probably what we think of first.
Um, you know, Jay can do a better job of, uh, explaining a gigawatt, but it's essentially like, uh, powering a million U.S. homes.
Uh, and these are typically built in, in the remote locations where power is cheap and available, and, and speed and latency are less of a concern.
So this is what helps you when you make a query of the AI model, uh, and those are built closer to metro areas in, in often these co-location environments where managing the speed of answering your query is more important.
So total demand, I'd say kind of hard to pin down, but we're tracking some 200 gigawatts of data center project announcements.
[chuckles] We, we need to adjust for all kinds of bottlenecks, um, but we think maybe 75 to 95 gigawatts are possible over the next few years.
But to really put this in perspective how much we can actually build, we only built six gigawatts in 2024 and only seven in 2025, so very large pipeline of, of demand here.
Um, I think most people would be surprised at how few data centers are actually being built in the United States on an annual basis.
Um, [clears throat] as I mentioned, we're gonna talk about data centers and what is driving this unprecedented demand for data centers today.
Um, I thought we'd start with you, Nathaniel, and, um, we have some questions teed up for Bridget as well, but let's, let's get the conversation started.
So, you know, obviously there have been data centers for decades, um, but what we're seeing today really began with the launch of ChatGPT almost on the dot in November of, of 2022.
Um, before that, for, for basically three or four years, we were building around $8 billion to $10 billion worth of U.S. data center capacity per year.
Um, but after ChatGPT came out, uh, you can look at the graph of data center construction values, it, it just turns into a hockey stick, uh, and it surges well above $30 billion.
So, uh, you know, that, that GPT moment, uh, really kicked off a global arms race to advance these, these AI models, um, you know, definitely for commercial use, but also increasingly for national security purposes.
Uh, and we've seen all these big tech companies, the hyperscalers pledge hundreds of billions of dollars of annual spending on, on data centers to train AI models, but also to respond to user inquiries.
Uh, and so, you know, those giant AI training facilities are probably what we think of first.
Um, you know, Jay can do a better job of, uh, explaining a gigawatt, but it's essentially like, uh, powering a million U.S. homes.
Uh, and these are typically built in, in the remote locations where power is cheap and available, and, and speed and latency are less of a concern.
So this is what helps you when you make a query of the AI model, uh, and those are built closer to metro areas in, in often these co-location environments where managing the speed of answering your query is more important.
So total demand, I'd say kind of hard to pin down, but we're tracking some 200 gigawatts of data center project announcements.
[chuckles] We, we need to adjust for all kinds of bottlenecks, um, but we think maybe 75 to 95 gigawatts are possible over the next few years.
But to really put this in perspective how much we can actually build, we only built six gigawatts in 2024 and only seven in 2025, so very large pipeline of, of demand here.
Um, I think most people would be surprised at how few data centers are actually being built in the United States on an annual basis.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.
