Nov 2, 2026 · 46 min · 12 segments
In this episode of the Global Rewire podcast, we are joined by Columbia Business School professor Stijn Van Nieuwerbugh to examine the massive financial infrastructure supporting the artificial…
Stijn Van NieuwerburghGuest
Fabio NatalucciHost
Craig TorresHostYeah, so the first point of my paper is a simple point, which is that this may well become the largest CapEx boom in world history.
To get at that conclusion, I start from the pipeline of all announced data center projects that are supposed to come online at some point in the next 10 years, let's say.
severely haircutted with the idea that not all of these projects will be materialized because some of them will not get power.
Some of them will get held up by local communities who don't want data centers in their town.
And so I end up kicking out about 40% of that pipeline to arrive at an estimated build-out of about 180 gigawatts.
Just to put that number in perspective, we currently in the United...
To put that number in perspective, we have currently about 60 gigawatts operational today.
All told, that would cost about $9 trillion to build that 180 gigawatts of compute.
For the listeners that are not familiar with the economics of this, about one-third of a data center is the cost of the real estate and the power infrastructure.
Two-thirds is the cost of the compute, usually the server racks, the GPU server racks.
They cost a lot of money in large part because a lot of these systems are NVIDIA systems that have incredible margins.
So NVIDIA is making a lot of profit on these systems and that's driving up the cost of these data centers.
If you scale that by GDP, that's about 3.2% of GDP per year over this eight year build out.
And the comparison is with the canal boom in the 1830s, the railroad boom between 1870 and 1890, the electrification in the early 20th century, the highway buildout, interstate highway buildout in the 1950s and early 60s in the United States.
And then, of course, also the internet fiber optic telecom boom in the late 1990s, early 2000s.
The other interesting analogy is that all of those, with the exception of the highway build-up, resulted in some sort of financial crisis, either a bond and or bank market crisis or a large stock market drawdown in the case of the internet boom.
Yeah, so the first point of my paper is a simple point, which is that this may well become the largest CapEx boom in world history.
To get at that conclusion, I start from the pipeline of all announced data center projects that are supposed to come online at some point in the next 10 years, let's say.
severely haircutted with the idea that not all of these projects will be materialized because some of them will not get power.
Some of them will get held up by local communities who don't want data centers in their town.
And so I end up kicking out about 40% of that pipeline to arrive at an estimated build-out of about 180 gigawatts.
Just to put that number in perspective, we currently in the United...
To put that number in perspective, we have currently about 60 gigawatts operational today.
All told, that would cost about $9 trillion to build that 180 gigawatts of compute.
For the listeners that are not familiar with the economics of this, about one-third of a data center is the cost of the real estate and the power infrastructure.
Two-thirds is the cost of the compute, usually the server racks, the GPU server racks.
They cost a lot of money in large part because a lot of these systems are NVIDIA systems that have incredible margins.
So NVIDIA is making a lot of profit on these systems and that's driving up the cost of these data centers.
If you scale that by GDP, that's about 3.2% of GDP per year over this eight year build out.
And the comparison is with the canal boom in the 1830s, the railroad boom between 1870 and 1890, the electrification in the early 20th century, the highway buildout, interstate highway buildout in the 1950s and early 60s in the United States.
And then, of course, also the internet fiber optic telecom boom in the late 1990s, early 2000s.
The other interesting analogy is that all of those, with the exception of the highway build-up, resulted in some sort of financial crisis, either a bond and or bank market crisis or a large stock market drawdown in the case of the internet boom.
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.