Will Southeast Asia’s digital rush come at a climate cost?

Happy New Year to you. Hello there. Well, trying to be ready for 2025, although I did catch a bug on the way back from vacation. And that's kind of slowing me down. A bit. I think I might have laryngitis. How about you? Did you get the best start to a new, new podcast season with the, you know, losing your voice. We can have that at the end of the year. I guess that's the perils of traveling during the holiday period. Yeah, I'm feeling it a little bit as well, but I

think we'll, we'll we'll get into it. We'll battle through. Yes, we will. Now, we're talking all things data, data. Do you say, which one do you say? Do you say because I I know I say it differently and I'm often interviewing people saying this word, uh, I could jar a bit, you know, that we're saying. I think it's widely accepted either way. OK, quiz, I know you've been, you've been waiting for months for. It feels like for a new quiz, today it's super simple.

It's going to focus on the energy use of AI processes, which are becoming very common of course. Now Leil Ling, have you created an AI image before? Now this isn't the question. This isn't the quiz question. That would have been a very easy question. No, I have not created an AI image before. I don't even know how I would start to do that. Have you? Maybe, yeah, yeah. Maybe that's something for 2025 for you to, to do. Now, creating a single AI image.

Is the same as running your refrigerator for how long? Wow, a day, a month? A year? Good grief. I have no idea. Alright, well you just get sit on that one. And we'll be at the end. OK, it's time for our main story this week, and let me take you inside a data center. It takes the press of a button for the secret room to be revealed. A solid wall slides open, revealing the innards

of a data center through an expansive glass pane. Picture rows and rows of black cages humming with sound and lights blinking inside, stretching into the distance. These are the network racks of the new industries driving a digital transformation. This is a high security zone, that's carefully temperature controlled, it's monitored by a team around the clock. Even staff members need permission from their clients to enter the data floor.

And these clients are the hyper scales, they're all here, the cloud providers, big data, digital services, game makers, artificial intelligence pioneers, operators that need vast computing resources, automation and efficiency. Now this facility I'm in is dubbed STT Bank. One, it stands out in a neighborhood in the Thai capital with futuristic cladding, a high security perimeter. It's pretty

conspicuous amongst the residential apartments and small factories. This is a data center operated by Singaporean outfit ST Tele Media Global data centers. But firms from all around the world are pouring billions of dollars into buildings just like this right across Southeast Asia, and they're forming part of a new digital infrastructure landscape across the region. The tech rush is here. Lee Ling, have you ever stepped inside a data center? What do you think, Jack?

hanging out in data centers on the weekend. Oh, OK. I mean, it's the first time for me actually being in one. I've seen the pictures of it, but never gone inside before. Well, yeah, I definitely haven't been in one, but I've seen, I think, a lot of them on television. And in spy movies, um, and I do know that these types of infrastructure are popping up all around us because I read your article and that they're doing so at a massive scale because they need to fulfill lots

of new tasks that our economies are increasingly requiring. So these are cases in building like hidden hives, right? They're secretive, anonymous, like you said, but increasingly vital for the digital economies. I do want to pick your brains about. Your research and doing the story though. So, so you can help. So we can help our listeners unpack the kind of growth that you're talking about and look at the impact this is also having on energy consumption, because

that's really the big question mark for us. There are multiple factors driving the growth of data centers, right? You touched on AI development and the need for cloud computing. Can you give us a rough idea of the size of this growth? Yeah, exactly right. It's pushing. 20% annual growth in demand for data centers in the ASEAN region all the way to 2028, so that's according to analysis from Maybank. And then if we look globally, all of this is happening at a rapid pace as well.

Computational needs are doubling every 100 days. It's all thanks to AI. Now people are already using AI a lot in their daily lives, and we know that these are really intensive processes that require a lot of computer power. 88% of knowledge workers in ASEAN are already using AI. That's based on a report from LinkedIn and Microsoft, I'm guessing you're not one of them. I'm guessing you are. Yes, I am using AI, um, a little bit. What do you use it for? I

mostly use it for researching. There are pretty useful tools, useful AI tools that can help you find information really quickly. So for that purpose, it's been great. I know that there are a lot more things that can be done, but, you know, slowly, slowly in the journalism world to embrace these types of things. And of course, AI is being built into a lot of the things we do with our computers and phones anyway, that you might not even really make a deliberate choice to use it.

And the other interesting thing as well. That's happening is this deep demographic shift that we've seen in recent years, the rise of the middle class, a younger tech savvy workforce is emerging as well. And in recent years, I think it was some 125,000 new internet users have been coming online daily in the region, and that's a staggering number really, isn't it? Yeah, 125,000 new people to the internet every day. That's, I couldn't believe that, seeing, seeing that statistic across Asia.

And of course, with all of this comes the steep competition from some of the world's biggest and most influential global tech companies that want a piece of this action. Now, our region is broadly still underserved by data, so there's lots of growth potential, and that's why we're starting to see some of these huge investment announcements. So let's just look at some of them from the past year.

Google, a billion dollars in Thailand, $2 billion in Malaysia, the promise to create tens of thousands of jobs right throughout Southeast Asia, Microsoft's pretty similar. Billions in Indonesia, Malaysia, it's also building its first data center in Thailand. Amazon Web Services, they've pledged some $9 billion for cloud services and infrastructure in Singapore alone over the next 5 years and,

and more around the region. And then you throw in the likes of, Nvidia, China's GDS, lots of smaller operators, locally backed or developed companies as well. It's a hot market out there for, for these corporations trying to make their mark in the space. But Jack, this wouldn't be climate conversations without having a conversation about the climate, right? I know it's the quiz question for later, but we know that the energy demands

of using generative AI are huge. Building out and then running all of these data centers is going to be very energy intensive. Isn't it? Companies and governments are still facing a dilemma around their carbon footprint and how to actually develop the technology and the infrastructure without consuming so much fossil fuel energy. Yeah, definitely. And, and of course this is an area that we know

is getting far more scrutiny. Now as economies slowly decarbonize or try to decarbonize in response to, The challenge of climate change. Data centers right now are pushing the needle in the wrong direction. It seems pretty undeniable that that's the case. Energy use within these data centers is sky high. It's estimated to already account for 1 to 1.5% of global energy consumption, which is a lot, given the growth as well that's happening in the, in this space that we

talked about earlier. If you just look at some of the companies, Microsoft, for example, big investor in chat GPT maker OpenAI, it's already seen its carbon emissions rise by 30% since 2020 due to its investments in global data centers. Then over to Google, it's also seen a 50% increase in 2023 emissions compared to 2019, and that's also largely

attributed to data infrastructure. So you have companies like these that have net zero goals, but all of a sudden are nowhere close to achieving them because of this demand for data and that's driven by AI. And we've talked about some of the numbers, but let's just look at how to train a generative AI model, for example, that can be the equivalent of powering 100 homes for a year, and training something like chat GPT 4. Was estimated to have used about 500 times that, so

these are really intensive processes. It's a huge amount of energy and some big tech companies are already looking towards alternative energy like nuclear power to replace some of the fossil fuel based energy that's been powering a lot of these data centers. Google, for example, announced last October that it's entered a deal with the California company Cairo's Power to buy 500 megawatts of nuclear energy by 20.

35 from what's known as small modular reactors. Our correspondent in California, Iris Spitzer, recently did a story about these so-called SMRRs. They are a fraction of the size of conventional reactors and can be assembled at a factory and transported elsewhere for installation. So not like, you know, those big columns that we see that mark these nuclear plants that we're so familiar with already, right? So these are smaller experts say they can.

be up to 8% more efficient than traditional reactors as well. It's not just Google, um, Amazon's also announced in October deals for nuclear energy projects, including from small nuclear reactors as part of the company's transition to carbon-free energy. And Microsoft is also paying to restart one of the Three Mile Island reactors in the US. The Three Mile Island, of course, is famous for having suffered a partial meltdown

in 1979, and it has been shuttered ever since. So it is kind of an interesting type of investment by big tech, but there are of course also concerns about the safety of nuclear power and nuclear. Reactor. So it's still in the in the early stages and definitely not something we're likely to see anytime soon here in Asia, despite some discussion around nuclear power amid

the increasing pressure to transition to cleaner energy in the region. Now, at present, half of Asia's electricity still comes from coal. Major investments will need to keep happening over the next decade to change that landscape. Uh, renewable energy capacity will need to more than triple by 2035 in order for countries to meet their key decarbonization targets, uh, according to EMBA, an energy think tank.

But in the meantime, the power needs of data centers globally is forcing a little rethink on those net zero targets and keeping dirty plants alive. Yeah. I think a key part of that discussion is that. If companies do have access to clean energy, it will drive more growth in the sector, and companies will come and invest more in your country if they have fewer

concerns about the carbon impacts of their operations there. So talking to operators who are setting up or already running data centers in Thailand, there's a sense of like, of difficulty that exists because they have no choice but to use the power sources that are there. And if Thailand itself as a country, for example, doesn't, Install huge amounts of renewable energy, well, these countries, these companies then bear the burden of having to use higher

polluting energy sources. So I think that's difficult. Obviously, having the companies invest more in your country can lead to various economic advantages. AI is transforming so many of these sectors from education to health, logistics, agriculture, your training, a new workforce, you're adding wealth to your economy. I spoke to several regional leaders from the tech world. They're aware of this issue here. And right now, the green solutions rollout is just happening a little bit more

slowly than this wild clamoring for digitalization. And before we wrap up too, of course, we haven't touched all on regulations, and this is a big part of the picture, how we can create sustainable frameworks for this new technology to ensure that it is equitable and transparent. So if you are interested to learn more about what countries in this region are doing in that space, do check out my infocus story on CNA.Asia.

Quiz time. Lee Ling, you've been busy talking, but to repeat your question, creating a single AI image is the same as running your refrigerator for about how long? I'm gonna go with 1 month. This is you just being anti-AI again. It's like, it's about 30 minutes. Oh, OK. All right. But I think to put it into perspective, you can create one of these images in seconds. You could sit there on your computer for 5 minutes and

create 50 images if you wanted. So that's where you need to probably pay attention to. Um, how much energy you're using just by playing with these fun little games, or these fun little image creators for your podcast logo, for example. Cool input, input there. Thank you for that. I feel enlightened. That was illuminating. Thank you, Jack. OK. Now, it's weather time, brought to you by Lee Ling Time. And thank you for that. And we continue to expect both rainy days and bits of dry weather

across the region. Surprise, surprise. Uh, no, not really. Now, the National Environment Agency in Singapore says prevailing northeast monsoon conditions, I expect. To continue in the first fortnight of the new year with winds blowing mainly from the northwest or northeast. Now what that means is that in Singapore, we can expect moderate to heavy thundery showers on most days in the

first two weeks of the year. Now in the around the middle of the month, strengthening of high pressure systems over the northern Asian continent. May bring a wind surge over the South China Sea. And what that means, Jack, is wetter conditions in the region. In Thailand, where you're at, temperatures will be in the low to mid-30s through the middle of the month. Mostly sunny, with reasonable humidity levels, not like here in Singapore. It

seems where we'll see a lot more rain and muggy weather. Hm, low to mid-30s. I've heard that story before in terms of the weather in Thailand. Although we've had some really cool mornings and nights, so that's been pleasant. We hope the air pollution stays away for a little bit longer. Maybe you can update us on the prospects for that next week. OK, that's it for this episode of Climate Conversations. Thanks as always for joining us early in 2025. See you

back here, same time, same place. Next week, I'm Lein Tan. And I'm Jack Board. Thanks as always to the team that put together this podcast, Sai Yayin, Tiffany Ang, Janai Jahari, and Christina Roberts.