Latest / The Emerging Markets Podcast with Fexingo: Developing Economies, Growth, and Investment / How AI Energy Demand Is Reshaping Emerging Market Grids
Transcript
- Lucas: There's a story that's been quietly reshaping several corners of the emerging market universe, and it doesn't start with a trade war or a currency crisis. It starts with a power plug. Luna: You're talking about the electricity demand from AI data centers, aren't you? Lucas: Exactly. And it's not just a tech story — it's an infrastructure story, an energy story, and increasingly an emerging market story. The basic math goes like this: training a single large AI model now consumes roughly the same electricity as a small town uses in a year. And once that model goes into production, inference — actually running it — can eat even more. Luna: I've seen estimates that global data center electricity demand could double by 2030. A lot of that growth is happening in places you wouldn't expect. Lucas: Right. The obvious locations — Northern Virginia, Ireland, Singapore — are running out of grid capacity. So hyperscalers like Microsoft, Amazon, and Google are looking at emerging markets where land is cheap, regulations are lighter, and critically, where there's access to power that doesn't take years to interconnect. Luna: Malaysia has been a huge beneficiary. I read that they've approved over four and a half gigawatts of new data center capacity in the last two years — that's roughly the output of four nuclear reactors. Lucas: And that's a staggering number for a country that's not exactly known as a tech hub. Malaysia's edge is that it already has a fairly robust electrical grid, partly because of its manufacturing base, and it's got a business-friendly government that's willing to fast-track permits. But there's a tension here. The grid wasn't built for this kind of load. Luna: So you get the risk of blackouts, or at least the risk that data centers crowd out residential and industrial users. That's already happened in places like Singapore, which froze new data center construction for years. Lucas: Precisely. And that's where the investment angle gets interesting. If you look at the emerging market ETF sell-off we've seen this week — the iShares MSCI Emerging Markets ETF, ticker e e m, down more than six percent in five days — it's been broad, but some markets are holding up better than others. India's INDA ETF is actually up half a percent over that period. Luna: India is also building data centers at a furious pace. Reliance and Adani are both in the game. Lucas: Right. So the market is starting to differentiate between emerging markets that have the grid capacity to support ai driven growth and those that don't. It's not the only factor — currency weakness, geopolitical risk, commodity prices all play a role. But it's becoming a lens. Luna: Let's talk about another example: Chile. They've got abundant solar power in the Atacama Desert, and they're attracting AI data centers specifically because of cheap renewable energy. Lucas: Chile is a fascinating case. They have a natural advantage in solar — the Atacama has the highest solar irradiance on Earth. And the cost of solar has fallen so much that electricity there can be cheaper than almost anywhere else. Google opened a data center in Chile back in 2023, and now Amazon is building one too. But Chile's grid is long and thin — it runs over three thousand miles from north to south. Transmitting that solar power to where it's needed isn't trivial. Luna: And then there's Saudi Arabia. They're investing heavily in AI infrastructure as part of Vision 2030, but their grid reliability is a question mark. Rolling blackouts in summer are not uncommon. Lucas: Saudi Arabia has the capital and the political will, but they're essentially trying to build a digital economy on top of a grid that was designed for oil extraction and desalination. They're spending billions to upgrade, but it's going to take years. In the meantime, data center operators are having to invest in backup power — diesel generators, battery storage — which adds cost and complexity. Luna: This feels like a real bottleneck. You can have all the GPUs in the world, but if the lights go out, your model doesn't train. Lucas: Exactly. And that's why some investors are starting to look at emerging market utility companies as a sort of pick and shovel play on AI. The companies that own and operate the grids in these fast-growing regions could become strategic assets. But it's not a simple trade — many of those utilities are state-owned, or heavily regulated, and the returns might be capped. Luna: You know, it's interesting to think about how this ties back to the overall theme of the show. We talk a lot about which emerging markets are diverging and why. Maybe the next big differentiator is simply: who has the power? Lucas: I think that's exactly right. And it's one of those rare cases where a physical constraint — kilowatts — is actually shaping investment flows in a way that's measurable. We've seen countries like Vietnam and Indonesia losing out on some of this investment because their grids simply can't handle the load. Luna: Vietnam has been trying to attract data centers for years. But their grid suffers from chronic under-investment — we saw that during the heatwaves in 2023 and 2024 when they had rolling blackouts. No hyperscaler wants to build a billion-dollar facility there without guarantees. Lucas: So the winners are likely to be countries that combine grid capacity, political stability, and a regulatory environment that's pro-business but not reckless. Malaysia fits that bill. Chile fits it too, though with geographic challenges. And Saudi Arabia is a wild card because of the scale of their ambition. Luna: And it's worth noting that this isn't just about big tech. Local companies in these markets are also starting to use AI more intensively. So the demand is coming from both multinationals and domestic firms. Lucas: Good point. And that creates a virtuous cycle — more data centers lead to better connectivity, which attracts more tech investment, which drives economic growth. But only if the grid can keep up. Luna: Speaking of keeping up — and this is a bit of a tangent, but it connects — we've been able to have this conversation without any ads, and that's something we're proud of. If you find this kind of deep dive useful, you can support the show at buy me a coffee dot com slash fexingo. Lucas: Yeah, it's a small way to help us stay independent and keep the focus on the substance, not the sponsors. Appreciate anyone who throws a few bucks our way. Luna: Alright, back to the grid. We've talked about Malaysia, Chile, Saudi Arabia — but what about Africa? There's been some interesting movement. Lucas: Africa is a huge opportunity, but also a huge challenge. The continent has some of the cheapest renewable energy potential in the world — solar in the Sahara, hydro in the Congo — but the grid infrastructure is often nonexistent or unreliable. A few billion-dollar projects are in the works, like the morocco uk power cable, which could eventually supply clean electricity to data centers in Europe. But for now, most African countries are still in the planning phase. Luna: And yet, companies like Amazon and Microsoft are building small edge data centers in places like Nairobi and Lagos to serve local demand. They're not the hyperscale facilities, but they're important for latency. Lucas: Exactly. Edge data centers are smaller, consume less power, and can run on diesel generators or battery banks. They're a way to test the waters without betting the farm on a grid that might not deliver. So we're seeing a two-tier market: hyperscale facilities in the grid-reliable countries, and edge nodes everywhere else. Luna: That's a useful framework. So for an investor looking at emerging markets, you might want to look at which countries are getting the hyperscale investments — those are likely to have stronger long-term growth stories. Lucas: Right. And you can track that by following the announcements from the big cloud providers. They publish their data center regions. In the last two years, we've seen new regions announced in Malaysia, Chile, Saudi Arabia, and even in Israel and South Africa. Each one represents a multi billion dollar bet on that country's infrastructure. Luna: What about the other side of the coin — the countries that are missing out? We mentioned Vietnam. Any others? Lucas: Indonesia is a notable example. They have the population, the economic growth, and the digital adoption, but their grid is heavily dependent on coal and suffers from frequent outages. The government has been trying to push renewable energy, but it's slow. As a result, most hyperscalers are setting up in Singapore or Malaysia and serving Indonesia from there. Luna: That's a missed opportunity. Especially since Indonesia has huge geothermal potential — it's on the Ring of Fire, after all. Lucas: Exactly. But geothermal requires upfront capital and long development timelines. The hyperscalers want power now. So the countries that have already invested in grid modernization are the ones that are getting the business. Luna: And that creates a feedback loop: more data centers bring more tax revenue and economic activity, which can be reinvested in the grid. But it's a chicken and egg problem for the laggards. Lucas: Absolutely. And that's why I think this is one of the most important under-the-radar trends in emerging markets right now. It's not just about AI — it's about which countries can build the infrastructure to support the next generation of technology. The winners will be those that solve the power problem. Luna: And the losers will be those that don't. It's a simple enough lens, but it has real implications for portfolio allocation. Lucas: Right. So if you're looking at the broad emerging market sell-off this June, it's worth asking: is this market getting hammered because of a temporary risk-off mood, or is it because the structural story is weakening? Grid capacity is one of those structural factors that can separate the two. Luna: Great episode. I feel like I have a new way to think about EM investing now. Lucas: Glad to hear it. That does it for this episode of the Emerging Markets Podcast. We'll be back tomorrow with another angle.