Josephine Teo: evaluation

Transcript context

…He's squeezing two questions in one. Right. Alessio, a couple of years ago, we articulated for the government a cloud-first strategy, which therefore means that we accept that there are benefits of putting some of our workloads on the cloud. For one thing, it means that you don't have to have all the capacity available to you on a dedicated basis all the time. We acknowledge the need for flexibility. We acknowledge the need to be able to expand more quickly when the workload needs increase. But when we say a cloud-first strategy, it also means that there will be certain things that are perhaps not suitable to put on the cloud. And for those, you need to have a different set of infrastructure to support. So having a hybrid approach where some of the workloads, even for government, can go to the cloud, and then some of the workloads have to remain on-prem. I think that is a question of the mix. To the extent that you are able to identify the systems that are suitable to go to the cloud, then the need to have the workloads run on your on-prem systems is more circumscribed as a result. And potentially, you can devote better resources to safeguarding this smaller bucket rather than to try and spread your resources to protecting the whole, because you are also relying on security architecture of cloud service providers. So this hybrid approach, I think, has defined how we think about government workloads. In some sense, how we will think about AI workloads is not going to be entirely different. This is looking at the question from the government standpoint. But more broadly, if you think about Singapore as a whole, equally, not all the AI workloads can be hosted in Singapore. The analogy I like to make sometimes is, if you think about manufacturing, some of the earlier activities that were carried out in Singapore at some point in time became not feasible to continue. And then they have to be redistributed elsewhere. You're always going to be part of this supply chain. There is a global supply chain. There is a regional supply chain. And if everyone occupies a point in that supply chain that is optimal for their own circumstances, that plays to their advantage, then in fact, the whole system gains. That's also how we will think of it. Not all the AI workloads, no matter how much we expand our data center capacity, will be possible to host. Now, the only way we can host all the AI workloads is if we are totally unambitious. There's so little AI workload that you can host everything in Singapore. That has to be the case, right? I mean, if there's more AI workloads, it has to be distributed elsewhere. Does all of it require the latency, the very tight latency margins that you can tolerate and absolutely have to have them in Singapore? Some of it actually can be distributed, we'll have to see. But a reasonable guess would be that there is always going to be scope for redistribution. And in that sense, we look at the whole development in our region in a positive way. uted, we'll have to see. But a reasonable guess would be that there is always going to be scope for redistribution. And in that sense, we look at the whole development in our region in a positive way. There is just more scope to be able to host these activities. For Southeast Asia?…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.