How AI could impact geopolitics
17:07 Share

The rise of A I isn't just a tech story. IT could have profound implications for energy markets, for geopolitics. And to Better understand all this, i'm sitting down today with dairy coin, who is our global affairs, and ahead of the common socks gobal institute.

Jerry, talking about the program. I much to have you. So jered, we have had a lot of discussions on this program about the many ways at AI shaping markets, the many ways that I could shape industries and economies.

But your focused on what I could shape even beyond that. Why is this an issue for the U. S? So look.

when IT comes to A I, there's a lot of known, unknown, right will be open waited models over closed models that went out. Will the use cases ever justify the enormous then? Who will be the winners and losers? These are all very interesting questions.

None of them are urgent. There's a very urgent question right now, which is the one that I am focused on, which is A A I software has to run on A I hardware somewhere. Can the U. S.

Maintain a leadership position when IT comes to generate A I and the answer comes down to a larger question about whether or not the infrastructure can be built to keep a pace with the demand. And the punch line is the U. S.

Can maintain that leadership position. The infrastructure can keep up with demand, but not exclusively in the U. S.

To me, this is the most urgent, no, not known because of the U. S. Wants to continue to dominate in this A I journey and A I revolution.

It's gonna options outside of the U. S. To augment and supplement that.

Why can the U. S. Do this? What is the constraints? There is a just a funding issue or is the actual physical limitations.

So there is no shortage of funding in the space that the hyper scales are set to spend roughly one point one trillion dollars on the higher and in capex sociate with meeting A I demand. The issue is if you look at the eight thousand data centers worldwide, there's roughly three thousand in the united states. Those data centers have very low vaccine, so you have less than three percent vacancy rates.

So that's one problem, is we don't have enough space in existing data centers. Second problem, if you look at our existing data centers, a lot of them support cloud workloads, AI workloads or ultra identity requiring a concentrated source of power. And so it's a very different kind engineering problem.

So the question becomes, can you retrofit existing data centers to accommodate AI workload is either prohibitively expensive, not practical or just not feasible. So we have a need for differentiated data centers as a second problem. So we they can see problem.

We have a need for differentiated data centres. But let's say we figure all of that out A I workloads because they require a concentrated source of power. That power also has to be largely based load.

So intermet in power like solar and when don't work. So the U. S. Has plenty of baseload power.

If you think about natural gas, not be a short term solution, nuclear, which is a Better long term solution, in theory, the U. S. Has that baseload power.

But where that baseload power exists, transporting IT to new data centres to run these A I workloads becomes a political problem. So I think not my backyard. So if the U.

S. To britten, another rough thirty five plus online to meet this A I demand, uh, we don't enough differentiated data centers and we don't enough powered land to do IT. And so the U.

S. Is going to need some kind of an overflow option to deal with this hockey stick growth in power demand. And the issue is for decades, we got used to power demand that was either flat or Frankly declining.

And our energy grid is under enormous amount of pressure right now. Extreme weather conditions, you know, delayed maintenance lag time associated with getting the permanent and necessary requirements for upgrades that electric grade. And so this isn't we need to figure this out by the end of the decade, over the next twelve to eighteen months, the U.

S. Is gna. Have to make a very important decision about where all of the A I infrastructures is gonna get built and its not about nature, it's about geopolitics um it's nations that are going to be the ones who decide where these data centres are built. And so here's where the geopolitics has a tremendous and important nexus with what's happening technologically.

And so what will that look like and and what will be the implications for the U.

S. So I think the us. Has three options for and overflow for building out A I infrastructure. And none of them are really great option. So one option is you have to keep IT in, let's call the jeff re SONY anish, democratic world to think. Canada, the other countries, australia, great options.

France, uh, the problem is that we have a lot of the same political issues that we have here in the us, and there's a risk and going all in on that. So geopolitically, you don't worry about those countries staying on sides, but you worry that they have a lot of the same kind of complicated domestic political issues associated with basic power. Also, democratic countries don't have a track record of being able to do massive infrastructure at scale efficiently and quickly enough to meet.

So we should try that should certainly be some of the play free I infrastructure. But there's a risk that if you go all in on that type of country, none of IT hands up manifesting. Second option is you allow IT to persist in the global south, and you just kind of optimize for where there's cheap based of power that takes you to places like indonesia and male sia.

The problem there is, if you think about the whole geopolitical theis around this, the U. S. Wants to prevent china from getting AI capacity around these large language models, indonesia and malay.

Sia is very unlikely that if they are able to become hub straw, this demand, that capacity won't be allocated directly to china. So that's not a great option. Uh the third option, the middle ast and the middle east, probably more than any other countries, has probably the best attributes to accommodate all this.

They have the cheap access to energy. They have abundant amount of land. We all know they can build massive infrastructure at scale very quickly.

They can build all this near the coast line, very useful for technology around look with cooling uh of the chips so they don't get overheated. They have the sovereign ambition. They have an enormous amount of capital that they can deploy flexibly and as they see fit.

And so in the short term, the the, the least makes a lot of sense, by the way, a lot of middle astern countries have also proven a willingness under some pressure from the U. S, to rip out a lot of the chinese made hardware to accommodate U. S.

Interest and get those necessary chip allocations to be able to do. The problem in the middle is fast forward, let's call, ten years from now, all of this AI capacity persist in the middle ast. And how do you know that they stay on sites? I there's a longer term geopolitical question of saudi abia cutter U A V R O geopolitical sringa tes.

They have enough economic where with all that on many issues, they've been willing to kind of swing on the issue by issue basis. The economic interest drive a lot of their geopolitical thinking. And so if you think about the future that A I capacity is very valuable, china, still their number one trading partners.

So how do you ensure that the geopolitics net out in the way that you want them to, if you're the U. S. Government, a decade from now? So none of these options are kind of a pana a to the geopolitical set of issues.

And the question of where the AI infrastructures can be built, if guess it's going to be some combination of the first option, which is keep IT in the democratic world and bias towards the middle ast. And I think, by the way, under the new administration, I would suspect that there is gonna a larger gravitation pull towards the middle ast and watch very carefully depending on how things play out with the war in the middle st. And a rearchitects ture of the middle east, if you end up with a scenario where, uh, this a number of things that save radio wants, there's another number of things that U.

S. Administration is willing to give in exchange for Normalization of relations with israel and A R architecture of the medals. I would be very surprised if an allocation of chips necessary to build out A I infrastructure wasn't part of the equation.

I guess my question to you is, are these companies, these U. S. Companies leading this? Are they on board?

So I think we're sort of early days on this. These countries in the midwest have a lot of leverage to be able to get countries to set up shop on the ground. I think the inflection point is a shift from doing so in a performative way or in a reactionary way.

I think what these countries would actually like is for the U. S. Companies to see the value of setting up shop. They are independent of the things they want from these countries. And I think we're not at the beginning of this, but that hasn't really reached its full maturity yet um but it's trending in that direction. But I think it's gonna A A long process and .

should but look, we ve already see and I think in some other industries that process moving forward.

But these things take time, right? If you think about the big push towards india as an example, to diverse y supply chains away from move india, even in the world's largest democracy, with a rich history of business that's been integrated yeah with the U. S.

In a lot of ways for for many, many decades. It's a long process there. IT doesn't happen overnight.

What I would say is there is right now a very pronounced we have we need dynamic between companies in the U. S. And the the three gulf countries of saudi arabia, cutter, N.

U. A. And it's not just about capital. It's about capital. It's about energy. It's about a regular or ecosystem that these leaders control, about the ability to do massive infrastructure quickly in its scale, about abundance of land. But it's also about an evolution in both the human capital, in the mindset of these countries that all sudden change the thesis.

And so how does china fit into all of this? How is that positioning itself in this, the global AI infrastructure race, which IT does not .

want to lose on for sure? For starters, if I did described the us. Competition before general AI, I would have said the U.

S. Had a three decade first move advantage. And I kind of ended in a tie because IT was an metric competition between the U.

S. In china. Most of those s metris disproportion unai. Benefit china.

And the view was that those aside, try or america's problem, not ours. Then, general A, I gave the U. S. A lifeline in the sense that china had an appeal battle, all when I came to large language models, because they didn't have the compute power with GPU, because of export controls.

They didn't want to train models at internet scale, the chinese fire wall and the way they govern their models was incredibly restrictive. And you weren't exactly seeing large flocks of entrepreneurs moving to beijing to set up the next large mega cat company for all the things that we've seen happen. Domesti C2Companies tha t hav e hav e exp erienced tha t jou rney.

Uh, that being said, necessity drives innovation. And right now, one of things we're seeing china do is they're trying to push themselves to have an efficiency renaissance. They're trying to capture efficiencies from energy perspective and a performance within the models.

Right now, it's an impressive R N D story will see how that materializes. They're still behind. There's a lot of debate about how close they are to cashing up.

I think we don't really know yet, but they certainly have the human capital and have the resources to throw at IT. The second thing you're seeing them do is lean into some of this sort of race around data centers around the world. So china has initiative that they call eastern data, western computing.

And by way, they they account for about a third of the clean energy investments around the world. They're investing six point one billion dollars in creating these different data center hubs around the world, leaning into the fact that there are the number one trading partner still with a lot of countries, they have master investments in nuclear power, which is something they're uniquely able to do. And so is this interesting story of contradictions.

I think last year, they produced a record amount of coal, while the same time producing uh, record amount of clean energy. Investments around the world today have the ability to kind of wear different hats in the journey. They certainly have plenty of resources, but this is accessible for that.

They afford to fall behind this. They've been a situation where you've had to catch up before and they've successfully caught up before a lot of different technologies. But there has never been a technology more important than this.

Why would they look to middle ast to partner with the middle ast and why would the middle is find them a compelling partner?

Says a very good question. So um I mentioned before that so ua aba cutter N U A R H what I would describe as geopolitical swing states, meaning may have things about their economies that differently them that in the competition between the U. S, put them in this unique position where both countries need them to get ahead and that gives an enormous amount of leverage to kind play.

Both sides act more geopolitically independent or quite independent IT turns out that when IT comes to A I infrastructure because the U S. Has so much unattainable wer um because of the allocation of GPU that that might be the limits of how much they can lean into that economic advantage to embrace this geopolitical swing state status because they can't build out the A I infrastructure for this new chapter of artificial intelligence without getting that allocation from the U S, which requires approval from the U S. Government right.

So in the short and medium term it's a binary choice. They can you lean into china on A, I, or they can lean into the U. S.

N. A. I am talking about ji is a very broad, the AI that requires the most advanced chips into binary choice. And right now you're seeing you to these countries, lean into the U. S.

How much does the fact that Donald trump is going to be our president again shortly weight into that binary choice? Is that matter a lot in terms of talking about some of policy shift he seems .

to want to pursuit. So when IT comes to U. S. In china, my view is this is the issue that has the most bipartisan consensus, is the most important geopolitical issue. It's the foundational geopolitical theater.

That is why I believe the geopolitics are more uncertain than they have been in at least three decades. When dollar trump was president the first time, he was the first one to change the policy. The policy for multiple administrations, democrat and republican, had been bring china into liberal international order, and they will confirm, and everything will work out.

Trump got the U. S. Off of that policy and pivoted towards a tough protections of posture on china. I think that america's allies and partners were surprised when job came in and didn't just continue trumps policies, but was actually even more protectionist than his processor.

I think whether you got Harris or or getting trump, it's going to be the case that each administration would be tougher than the previous administration. I think that will continue a pace. I think the character of that protectionism doesn't particularly change from one administration to the other.

I think IT just gets more intensified, right? So the character ter used to be that the republicans like terror democrates like export controls, think very neat out as they both like both and um we pay more attention to terrace because trump is very vocal about them. But a lot of people have a lot of opinions about what's gona happen when the forty seven president takes office.

I think that it's hard to imagine that you're not going to have an expansion of terrifying that touch this particular issue. I think the thing that I would watch for is, are there additional terms s that we see that inadvertently impact the AI supply in in ways that were not obvious right here you're talking about that they're refining and processing of certain critical minerals that go into the wafers and know these things when administration comes in, they do a lot very quickly. The transition team isn't the same thing as a fully staff administration. They announced things on day one, and then the second, third order affects reveal themselves. But until we see with that first back of executive order are impossible.

Jared, thanks so much for joining us again. Always interesting conversion with you. Thank you. This episode of gold sax exchanges was recorded on monday, november twenty four. I'm your host to Alice, Cindy and things for listening.

The opinions and views expressed in this program may not necessarily reflect the institutional views of goon sax or its affiliate. This program should not be copied, distributed, published or reproduced, in whole or in part, or disclosed by any recipient to any other person without the express written consent of goon sax.

Each name of a third party organization mentioned in this program is the property of the company to what to relates, is used to strictly for informational and identification purposes only, and is not used to imply any ownership or license rights between any such company and common sex. The content of this program does not constitute recommendation from any golden sex entity to the recipient, and is provided for informational purposes only. X is not providing any financial, economic, legal, investment, accounting or tax address through program or to its recipient.

Certain information contained in this program constitutes for looking statements, and there is no guarantee that these results will be achieved. Goma sex has no obligation to provide updates or changes to the information in this program. Past performance not guarantee future results, which may vary. Neither golden sex nor any of its civilians makes any representation or warranty, express or implied as to the accuracy or complete ness of the statements or any information contained in program and any liability, therefore including a respective direct, indirect or consequently loss or damage, is expressly disclaim ed.