Building for the AI Era

My pleasure to have brilliant speakers to have this convers conversation. Mark, let me start and ask the first question. How we should support the massive need that AI and data center have?

Well, look, the the the need for data centers today is critical. It's really the tip of the spear of AI. And if you think about the innovation of the internet, you think about the innovation of of cloud computing, we're going through probably the biggest and most transformative time in our lifetimes in terms of a tectonic shift and how we're going to function as an economy. It's going to be a 7 trillion spend. It's going to be not a single decade.

It'll be a multi-deade approach that requires a lot of patience, a lot of investment. a lot of capital uh and most importantly uh a ton of power more power than I think we all had uh previously had understood and that creates a whole set of new challenges. So in my 30 years of building infrastructure it's probably the most unique and biggest opportunity we've ever seen.

Oh thank you so much Henrik. AI demand is growing fast. What are the biggest infrastructural bottlenecks that we need to fix first?

Thank you. I think some of the points were touched upon already. I think there's an interesting saying that power now equals knowledge. So clearly power is going to be an absolutely key component in terms of building the data centers. But you also need efficient and renewable powers within that.

You need the whole ecosystem uh the within the the unit the the electrical cables the switching etc. So all of those are putting certain strains on the supply chain. It also follows some of the challenges we've had about global supply chains with co but for us it's an incredibly exciting uh time. We think it's going to generate a huge economic boost over the next coming decades and akin to what's been mentioned quite common. You know when the railroads arrive globally it is transformative to our economies.

Even within standard chartered we're working about 300 different AI use cases. So we can see as a bank there's going to be a lot of change for us. Oh, thank you. Well, Peter, we start that conversation about AI bubble. So many talks currently.

What is your opinion? Please share.

So, I get the easy question I guess. Um, look, I I I think bubble, no bubble, it it depends to be honest. Um, in terms of the perspective that you have that that you look at. So are we talking about the valuations of sort of Nvidia for example who is trading at 50x and are they overvalued that's not what I'm referring to here uh when we talk about AI bubbles um I think in in this case maybe in on on this panel I think it's about uh if we talk about data centers uh and from an FAB point of view we are a lender to a data center then ultimately what we evaluate is well the very strong barriers to entry for any investor. We look at obviously strong contracts that you have with trillion dollar companies that are highly investment grade rated ultimately.

Um so so from that point of view from our point of view you know we we don't necessarily look at the bubble side of it. uh although the equity valuations are relevant for us but again there there are significant equity cushions there are equity investors that are investing ahead of us um so so yes from our point of view it's it's something we've been doing a lot of uh in not just in in the UAE but uh in in North America Europe and and Southeast Asia and we'll continue to sort of uh look to do I I I think uh sorry I I I I think what you have to do as a lender at least is do not the prudent part is also the portfolio approach so you know you are you talking about Meta are you talking about Oracle uh or others because that's where you you look at the a portfolio approach sorry Henrik

well I was going to say as you see markets evolve or expand uh dramatically you see a lot of capital move into one space and there can be different opinions on those evaluations uh we have the theories come through around bubbles or overstretching etc. Uh I I think we're in a very significant growth phase here backed by predominantly very strong corporations who are using their balance sheets to invest for the future. Whether valuations are overstretched now or not is a a function of debate and of course not everything's going to be a success nor is everything a success in everything that you invest in. So I think you need to be measured around this. The last six months we've heard a lot about private credit bubbles and AI bubbles.

I think we need to look at the topic in the broader space of investment and for the most part many of these projects are investment grade. So we seem to have also lost the narrative there a little bit.

That that's the key. You know we we priced a securization two weeks ago for Switch which is our tier five private cloud business based in Las Vegas. That deal was five times over subscribed. Does that feel like a bubble? No, that's not a bubble because it means investment grade paper,

right,

that's tethered to AI is performing quite well.

And insurance companies need that paper. Again, I I I think I take a perhaps a bit more sanguin approach, which is that, you know, we're building this infrastructure for the next two decades. And so, as we think about what sits in this room, infrastructure capital, we're really at the precipice of building this infrastructure, Marina, today. And look at what happened to capex this year. The estimate put out by McKenzie earlier this year was we were going to spend $225 billion in capex to build AI.

By the second quarter that had gone to 300 billion, a 75 billion increase in 90 days. Then that forecast went to 350 billion. We're going to end the year at $368 billion of infrastructure for AI. that does not include power, which we talked earlier, there's a a whole ecosystem of grid independent power projects that we're building. And so, as we go down this road in AI, two things have held up to be true.

One, the dinner bill is a lot more expensive than we thought it was going to be.

And then two, the tethering to power and the ability to build grid independent power solutions is becoming not only an extra, but I would share with you that that is a survival skill. If you want to play an AI infrastructure, you better know how to build power. You better know how to work with regulators. And most importantly, you better know how to put offtake in a base load to move yourself ahead in the queue. There's so many things happening in relation to this infrastructure.

But if you listen to Jensen, Elon, Masasan, they've all said the same thing in the last 90 days.

Power is the AI currency.

And that is an absolute truth today. I think you also touched on a really point and that is there are new financing techniques which are used now which didn't exist maybe 20 or 30 years ago. So the capital structure for the most part is placed with the appropriate risk appetite. So securitization is an excellent example where you are trunching the capital structure and you have a waterfall effect so you can buy into the capital structure that is appropriate for your risk tolerance that is opening up huge new avenues of capital and facilitating the ability to finance these mega projects

and and that's I think look there's a need of I think $7 trillion or so of of capex over the next five years or so

and that needs to be financed It needs to be invested by by everyone who's involved in the ecosystem and and the capital markets point is important because or the financing point is important because you have everyone participating uh from insurance companies, pension funds, uh in sovereign sovereign wealth funds, lenders, banks, R&Bs market, CNBS market, PP private person market 144. So literally the entire capital market system is sort of participating in this. So yeah.

Oh, thank you for this dialogue. Mark, can the JCC really become AI global hub? And what still needs to be built for that to happen?

Well, look, we we've been um a partner to the region. We've been a partner to the UAE for over a decade. Um there's been no more prominent investor in AI today than the UAE. And I say the UAE Inc. more broadly.

But if you think about the pension funds, the sovereign wealth funds and the various uh entities that the UAA has set up, um they've approached it with an ecosystem approach. And I think when you think about creating a hub for AI, part of the game is investing, part of the game is building, and then part of the game is also building an ecosystem where you can um foster the right environment to create AI. And the UAE has hit on all three of those. the GCC is hitting on all three of those. Um the challenge is ultimately you have to follow the compute and so so far the hyperscalers have sort of voted with their megawws and so today there's about 800 megawatts of hypers scale compute here in the GCC.

A big chunk of that sits in Bahrain where Amazon's major node is for the region. Microsoft has set up shop here in the UAE. other hyperscalers have followed and so so far you've seen a lot of compute being generated here in the UAE. Saudi is a lagard economy which is coming strong in terms of building compute capacity and from there you have Bahrain and Kuwait and and Qatar also at about 100 megawatts in total compute capacity. This is probably at the end of the day a market that gets somewhere to about 3 and a half to 5 gawatt from an infrastructure perspective and from a power perspective.

What I like about the region is there's an abundance of power. There certainly is a der of capital which is good. You need capital and um the region has not been afraid to recruit talent. You need talent whether you're building infrastructure, whether you're building large language models, you're moving to inferencing or you're actually moving to applications which is really where the UAE has excelled. Um you have to create that entire ecosystem where AI can flourish and the UAE has been an incredible model for that.

Yeah, I I think to to add to Mark's point as and as a as the as FAB, the largest bank in the UAE, we and and being part of that ecosystem of of supporting the growth of AI investment, not just here in the Middle East, but but outside with all our partners uh whether it's the sovereign wealth funds here that are some of the largest investors in the world uh or us as a lender to some of the largest investors in the world uh in globally. It's it's very much sort of core and front of uh of of everyone's mind of the top leadership in the UAE to the extent of obviously the talent is is coming here. We're seeing it uh here in the room and and ADFG uh finance week. I think there we go. All right.

And I think um from a from a talent point of view, the UAE set up um Muhammad Zia's University of AI to foster the next generation of talent. uh so it it's really much a a focus and I think for for the next decade very much uh I I would expect that the UEE to be um another hub uh because you know we have land we have we have power we have capital and we have good uh investors and and relationships with the likes of Nvidia and the US so they can get the chips as well so so I think so

and it's not only um the AI data center space I would say is across all infrastructure verticals. The UEE and Abu Dhabi is an absolute leader. If you look in logistics, transportation, this has been decades of foresight and investing. Infrastructure more recently has become an incredibly popular asset class. But you know 10 years ago, infrastructure was a little bit quieter in the financing space and people were a lot more focused on leverage finance or other more exciting opportunities.

the last three to five years, the returns from the sector, the scale of the projects and investing in has been a really unique opportunity, but you need to be a long-term investor in these projects because there they are going to there's a build period and then there's a monetization period and the foresight to do that is really important. But putting that infrastructure in place powers decades going forwards. We can conversely look at the areas that have not invested in infrastructure and the struggles that they're currently facing. Um, how are investment strategists should change to keep up with AI's rapid growth?

Hendrick,

uh, I think investors are very pragmatic. They are going to where they see strong returns based off strong credit ratings where they will see that their capital will be deployed efficiently and be secure and be safe. And I think in the recent changes whilst we've seen some movements in credit spreads for top rated companies I think overall the investment demand in the data II space continues to be incredibly strong. So I see this as a shortterm maybe setback or a little bit of repricing that's going through which is natural given the size of the transactions and the amount of capital that's being deployed. But I don't think on the whole investors are moving away.

If if investors have risk concerns, they may come up with slight tweaks on structures or other opportunities. But fundamentally, this is a generational opportunity to invest in something that's going to be very dramatic.

Investors like good deals, right? I think you know if if you have good products, um nothing resonates with an LP today better than a good co-invest. And so thinking through private credit, it's having the right SMA structure with your clients to make sure that they can see everything that you're investing in. I think as a global investor, um, you know, we we manage about 120 billion of asset center management. And the the great thing about having a global business is we can give investors what they want,

right?

And so having a tailored co-investment program and giving them the products they like are essential. I think the other thing is I I'm not a believer that the the private credit bubble has busted. I mean, what we see in our private credit business is really strong appetite. There is a slight repricing of risk. The good deals go fast.

The harder deals take longer to syndicate. But the two thematics that investors are are still really excited about is one is power. you know, power today is is a different game and to the extent that you understand how to build that grid independent solution that ultimately is tethered to what you said earlier, a long-term investment grade commitment. So, if I have a 17-year commitment with Microsoft and that 17-year commitment is tethered to uh LG or it's tethered to solar or it's tethered to some unique powered solution that is tethered to my data center, the returns on power are actually better than the returns on the data center,

which is interesting because that was not the case three or four years ago. We're still building data centers. We're still investing in it. It's it's it's part of our bread and butter. But investors, sophisticated investors are saying, how can we take advantage of the power arbitrage and how can you tether that to an investment grade lease?

And so um I think you know power is a really interesting frontier. The rules are being rewritten and whenever there's an industry that goes through a complete restructuring what happens is returns are mispriced and you can create alpha. I love sectors where we can go in and differentiate ourselves and we're doing that. We're doing that in power and we're also doing that in private credit. So private credit for me is still really important but power is also really important.

The PowerPoint um that Mark's making I think I read somewhere for the for the 7 trillion sort of cap capex that's required over the next 5 years and and the power is sort of part of that but what it actually means is just to put things in context it means the the new power source you're going to find is the equivalent of the entire of Japan that's today. So in in a in a given year. So that's the size of power and the investment need that that that's required to develop the power need.

I think the private credit point is really interesting because I think it's partly how you define private credit. Arguably you could say any loan, any bilateral loan could be considered private credit and financial institutions have done that for several centuries. So I think it's a lot of focus on a few specific cases that have happened. uh there may be some stretch valuations but a lot of as I said earlier a lot of the private credit space that's helping in the AI data center financing is investment grade so we're not talking low single B type transactions here or issues in supply chain finance which have happened or uh other challenges so we're going through an economic cycle like we always do investments there'll be changes to those economic cycles and there'll be appropriate valuations for that but we we don't think there's a uh an overheated situation, but as investors clearly need to pick your right opportunities and your right spots and work with the best managers in the asset class.

Peter, so how do you believe banks, investors and infrastructure providers work better together to build a strong AI ecosystem? Look, I think the it goes back to the point discussed earlier probably, but the amount of capital that we talking about the trillions of dollars that we refer to, it's not going to come from one form or one pocket of uh of the market. Uh it's really going to come from equity investors. Uh and and that's GPS. It's going to come from sovereign wealth funds, going to come from pension funds.

uh and when it comes to that the debt side is going to come from not just the banks but the capital markets and the entirety of the capital markets that I mentioned before and all of us working together as an example on a on a recent transaction actually a digital bridge transaction there were banks part of that transaction which was American banks, European banks, Middle Eastern banks, Japanese bank all of us really coming together to try and make sure we had all the liquidity that we needed. This is a $ 38 billion fund, debt financing, uh so a a a big financing and to try and make sure that we could just cover that and that was by the way well covered. So so the liquidity is there, but all of us have to work together to make sure that we solve the problem. Um, and and that's how we do it ultimately. And there there's there's plenty of opportunities, but it's partnering with the right managers, the right operators who's got the experience to be able to deliver a data center on time and and and connectivity to the grid and land.

Thank you so much for sharing all your expertise, opinions. It was wonderful conversation.

Thank you.