The Business of Data Centres: Capital, AI, and APAC'S Growth | In Conversation with Serene Nah

The higher connectivity hub metros will essentially be higher rent for some of these most timely workloads.

Yeah, this is interesting because there was a, I think Singapore's deputy finance minister was in an interview a couple of months ago and he was talking about different use cases and they said, oh, we don't want people to store their cat pictures or whatever.

Cat pictures? Okay.

in singapore but it does seem that there's a shift in policy in singapore i wonder if we'll see this in places like sydney or london or you know where the real estate is valuable and the access to the market is valuable and so they go well actually i i know debt center operators don't like this i i do want to dictate what the workloads are i don't want you to train a model that's being used in another country because my power is cheap i don't want you to store a bunch of photos i this is singapore we've got commanding markets so i want to be using financially critical or the kind of workloads that drive economic activity.

That seems to be an undertone from some of the CFA stuff, but do you think that would be more widely adopted?

Well, I think there are some market forces and also macro forces that will govern some of this.

which is at the heart of data centers, we sell space and power, right? And the amount of power that's available in each metro and the adjacent metros, just remembering the adjacent metros are important as well, will dictate what you can build and what you can offer.

You know, I like to think that there's not enough power in the world to supply all the AI needs we want in the world, which then means, you know, we got to pick and choose depending on where you need them to be.

Or space.

In space, that's true.

And also, like we said, let's not forget about cloud and digital transformation.

That's still growing, right? Especially in Southeast Asia, that's just at the beginning.

So you're going to find different workloads find its own natural spot and also have its own pricing.

Then when you come to the AI workloads, you might come across where, you know, you mentioned Sydney, but let's not forget outside of Sydney, there are actually many, many other metros that could potentially be lower cost because power is a bit more abundant or renewable energy is a bit more abundant.

And therefore there is a place for that.

But where else the more latency sensitive, critical workloads, government workloads, financial services might need to be in Sydney or hyperscale workloads might need to be in Sydney.

I think same with Singapore, which is we have a monitorium and we have, you know, we're focused on the carbon ceiling that we have.

And therefore, as part of what we've shared with the Singapore government, both as Digital Realty and also as my vice chair of APDCA, I said we really need to focus on the higher value workloads, which in this case would be cloud services or cloud services.

um needs for enterprises and also the government and financial services because of data sovereignty requirements in singapore and also ai inference and therefore as a result of that you kind of have a bifurcation because that's the best use of power and that will then enable not only more enterprises to be in singapore but also create an ecosystem of data center talent data center vendors and also the employees at the customer base that needs to be trained for this kind of higher value workloads.

There by itself means that our adjacent metros will be able to house maybe the training workloads or the less latency sensitive workloads or the less data sovereign workloads to complement this.

And so the connectivity to Singapore is going to be very important.

And how do you manage that? Probably more with your AP DCA hat on.

How do you manage that communication? Because you're talking about the high value workloads being in Singapore, which I think makes sense for a variety of different reasons, as you mentioned.

But then when we're talking about some of the So the training, for example, being in surrounding markets, do you see a bit of a conflict there? And like, how do we, how do we as an industry manage that so that we're not maybe upsetting certain communities or certain countries, certain metros that are deemed to have the lower value workload?

Yeah, I think it's important to not ascribe or maybe not to dive too deeply into what is lower valued.