The memory supercycle and what it means for Asia Pacific equities

Welcome to Beyond the Benchmark, the EFG podcast with Moz Afzal.

Moz Afzal:

Hi everyone. So today I have with me Chris Chan from our Hong Kong office. Chris Chan is the lead portfolio manager for our Asian equity funds, as well as doing a lot of work for the CIO office within Asian equities itself. So Chris, welcome.

Chris Chan:

Great Moz good to be back. Thank you.

Moz Afzal:

So obviously we're right in the middle of the Middle East crisis. Today is the third of March just so that we can date this podcast. Obviously lots of volatility in the Middle East region with the US, Israeli and Iranian sort of tit for tat and lots of missiles and drones flying around everywhere. Obviously a pretty scary time, certainly for the region. And we certainly wish the best for everyone in the region as well as our colleagues, especially, and our clients of course. And then so I guess what is the impact certainly so far within Asia?

Chris Chan:

Yeah, so I think there's a couple of things here. Korea was closed and open today and it was down a pretty hefty 7%, of course, driven by the two big weights, Samsung and Hynix, which of course led the Kospi up so dramatically over the past kind of a year or 18 months. So I think depending on how this situation obviously transpires, I think any risk or concern when it comes to geopolitics or overall global growth, whether that's inflation induced or otherwise, you're naturally going to get a large amount of profit taking from of course those markets that have done well and of course done more so than Kospi, which has unquestionably been the best performing, at least major market of the past year. So it's only natural we would get profit taking. And of course, even outside of that, the tech cycle, which is prevailing, of course, in Taiwan and Korea, being a relatively high beta global cyclical proxy, again, is going to be taking a bit of a hit.

Outside of that, naturally the more defensive markets like India and Asean, not to say they won't offer weaker absolute performance, but on a relative basis for the Asia region being slightly more defensive, particularly on the export side being less export orientated compared to, of course, North Asia, then naturally these global growth concerns and geopolitical risks, they're slightly more insulated. So you've seen that in share price performance. For example, today, the Southeast Asian markets have certainly outperformed Korea and Taiwan. I think within China and Hong Kong, again, Hang Seng historically has always been much more open because of foreign flows, foreign positioning, so more susceptible to, again, global concerns and global selloffs. And on the flip side, the China A-share market historically actually has one of the lowest correlations to not just US markets, but global markets as a whole. And of course, there are other reasons outside of what's going on in the Middle East as to why we expect that divergence to potentially converge going forward.

Moz Afzal:

Yeah. So I think obviously a bit of profit taking, which is, given the strong performance we've had certainly on a year-to-date basis, certainly made some sense. Let's move on to maybe more medium and long-term consideration suddenly in Asia. So we've obviously got the Korea and Taiwan technology proxies, which have obviously performed very, very well. And if I look at Korea, for example, extension over 50-day moving averages would also suggest that there's a bit of a correction. But Korea's been clearly the darling over the last 12, 18 months. What do you think the reasons for that are?

Chris Chan:

Yeah, so there's a multitude of reasons behind it. I mean, let's clearly address the first one, which is of course, going back to the Samsung Hynix, the memory cycle. In a nutshell, essentially this memory supercycle has essentially been driven by a couple of things. But essentially when you look at what you call high bandwidth memory, which of course is the memory being used in AI servers along with Nvidia's chips and Googles and the rest of them. And so of course, with the explosion of demand we've seen over the past year in these area, what that essentially is meant is that on top of the underlying strong demand of the high bandwidth memory itself, of course, for AI purposes, it's also essentially crowded out the more conventional, what you call DRAM memory. That's the memory that goes into the more traditional elements, general servers, consumer electronics, smartphones and whatnot, because essentially to create essentially a HBM memory, you're essentially needing to use three times the capacity versus the DRAM.

So for the same amount of overall capacity, it's essentially leaving a lot less space to produce DRAM. So that's created this shortage and this squeeze, and that's actually coming at a time on the supply side where demand actually, even outside of high bandwidth memory, i.e. Conventional DRAM has actually accelerated, not through of course, smartphone and whatnot, which remains relatively tepid, but actually through the shift from training to inference AI. And I think this is an interesting point because back at the end of, or at least the start, sorry, of 2025, I was certainly speaking to a lot of the Taiwanese supply chain and they were saying, "Well, actually the Chinese have started to experiment with general service, traditional general servers that we use for normal enterprise functions." They were starting to use the same general service for inference or an AI inference. And that's actually gained a lot of traction over the past year, and now that's being more used globally.

So this is certainly an area where people maybe two years ago didn't think that they could use general servers in such capacity. Now they certainly are. So that means, again, this more conventional memory that are used in general servers like DRAM, they're also getting that AI tailwind as well as the HBM. So this has created this high demand shortage of supply. Of course, we've seen that. I think memory pricing has gone up four times in just like the last three or four months. Now I think more importantly, looking forward as to for anyone thinking, "Okay, have I missed it? Where's cycle growth from here?" The kind of two indicators that I look for is, of course, the high frequency data in spot pricing, particularly when there's a divergence between spot pricing and contract pricing, which is a slightly more longer term. Certainly this year from a direction point of view, I think it's pretty much assured that the demand supply will remain very in favour of price increases.

We're always looking at the weekly and the monthly data. I would note, albeit it's very volatile, but actually looking at the end Feb data for spot pricing and memory, it was pretty flat actually month on month, again, after being up four times. So you do have to be a little bit cautious. It could be a one-off, but certainly if a second month comes in when spot pricing slows down, then obviously this being momentum play, you've got to be cautious of that. The second thing briefly is on inventory. And I think this is important because right now you're actually selling channel inventory or upstream or downstream, very, very tight inventory, obviously in line with high demand. And I think final point, of course, is valuation clearly on a price to book, as you'd imagine, it's essentially at historic highs. That said, price earnings ratio is actually near trough.

That does give me a little bit of confidence because it's essentially still, it's saying the market is implying that you're at peak earnings. I think if it was a higher PE multiple, then there'll be less margin upside. As you go into 2027, you get risk to supply from China and elsewhere, so you've got to be a little bit more cautious on.

Moz Afzal:

Yeah. I guess it certainly is a commodity. I think there are two or three other, maybe just stepping back a little bit, just thinking about memory and how important it is as I think you mentioned as we move to the inference part of the AI story, the NEFA memory just goes up rather, rather dramatically. And so that's certainly, if you like, the backdrop to why these prices have continued to be very strong. Probably another few considerations maybe, and maybe you can comment, is that essentially, other than the Chinese are now coming on stream or say want for another few more years, but saying being a monopoly for three players for ... Can you just comment a little bit on, because of this shortage, it's having knock on impact to say handset costs, memory is a big chunk of handset cost components, and you're hearing a lot more noise about pricing just getting a little bit unrealistic and margins being hit, the phone companies or the companies that make phones, and even laptops and iPads and PCs and so on and so forth. What are your thoughts on that? Does it mean that because pricing is so strong that we see a pause in those other areas?

Chris Chan:

Yeah, absolutely. And that's of course why globally, not just in Taiwan and Korea, the smartphone, particularly smartphone where HBM or sorry, DRAM memory and NAND as well, which of course has also gone up, they constitute typically about 30%, maybe 35% of the BMO, essentially the bill of materials to produce a phone. So it's a pretty significant portion. And so ultimately, if you are a phone manufacturer and suddenly you're seeing one of the biggest portions of your cost going up dramatically, you kind of have two options. You can either say, "Okay, well, assuming I can get the memory, I will pay that amount and then pass that on in terms of higher phone pricing to the retail." But of course then you might keep your margin because you're passing it on, but of course ultimately volumes are probably going to be weaker. Or you can say, "Okay, well, I'm not willing to have lower volumes, but I'm going to take the margin hit." Or a third option, you might just say, "Well, actually I'm just going to produce less because I'm not willing to do that."

And I actually think that obviously ties in with, can they actually get enough supply regardless of whether they're willing to pay the price? And there is sounds when you look at the Chinese handset makers that there are sounds, probably the third option might be more likely that actually even if they wanted to, they can't get enough supplies. So they are just actually producing less, whether they can pass that on or not. And you've already seen that. It's coming at a bad time. You have to remember China, of course, being the biggest smartphone market is the swing factor globally, and this actually held up smartphone volumes relatively well because you had these incentive or trade-in incentive schemes in China, which essentially helped smartphones amongst other product categories last year. Now, the problem is going into this year, bad timing with the pricing, but smartphone sales was already trailing off quite significantly because you essentially had both the exploration of the trade-in incentives, but also you had a lot of front pulled in demand of smartphones last year.

Moz Afzal:

Yeah. So I think the point being is that I guess memories now become a bottleneck for many different industries. And I think certainly something to keep a close watch out for. As I said, traditionally, manufacturers do overorder and then that leads to a glut very quickly. So I think that's the key thing that certainly we are watching. Staying on Korea, so outside of obviously the technology areas, sectors, what else is going on? Because I think if I think about over the last, say, certainly 12 or 18 months ago when the rallying career started, there were kind of other forces of work there.

Chris Chan:

Absolutely. Yeah. There's been very exciting times even outside of the memory cycle. And there's a couple of things both in terms of the sectors and thematics at play, but also maybe arguing more importantly is the undertone of the retail investor, which is getting a lot of headlines right now. Because I think if you rewind to the start of 2025, it was very much the start of the Kospi outperformance was driven by this corporate governance reform. That's really been going on for certainly two years now as they essentially tried to copycat the Japanese playbook and frankly, relatively successfully, they've really from a top down governmental level introduced guidelines and encouraged and in many cases forced a lot of the listed companies to improve their corporate governance, improve their payout ratios, use their cash better, cancel treasury shares. Now, the second phase I would say really has been talked more about the last three to six months, and that's really the interaction of the retail investor.

And this is where it gets quite interesting because again, from a top down point of view, you've had this relatively new president. And so he's actually done a couple of things. He's essentially trying to propose, not confirmed yet, but in the works to essentially give investors lower capital gains tax if they sell overseas and by overseas ultimately means US and US tech, which historically the retail investor has focused on because they hadn't wanted to invest in domestic. So lower capital gains tax, if they bring that back and invest domestically, so that's one element. Then the second element is, again, a lot of people, just like in Asia in general, put a lot of their savings into housing. He wants to control obviously housing costs and sold, particularly has always been an issue in terms of affordability. So they try to take a double whammy of policy where essentially he's offering incentives and introducing excessive tax on housing, particularly those that have multiple ownerships, and then giving benefits if they, again, invest that in the domestic, particularly the ETF market. So by all accounts, my understanding is you have seen quite a increase in house listings in Seoul off the back of this. And so you can imagine that is going to, with the housing and the overseas, that is naturally leading to quite a significant portion of retail flows.

Moz Afzal:

So let's move on to China. Actually, before we do, let's finish Taiwan first, because clearly Taiwan Semi is the big driver to the market and has been for some time. And again, similar story to the memory story under investing in GPU manufacturing, and that certainly had quite a big impact on pricing and the fact that demand is way outstripping supply. Now they've announced big CapEx spending plans now to try and alleviate this shortage. Any thoughts on that? So there's a very similar story to the memory story we just talked about is underinvesting CapEx and this big demand through AI comes along and then you've got this mismatch. What are your thoughts around what's going on with GPU manufacturing and I guess AI manufacturing in general and the wider Taiwanese tech community?

Chris Chan:

Yeah. So I really find Taiwan really interesting because I think the narrative in Taiwan is still very much underappreciated. And that seems strange because everyone knows Taiwan is the land of tech, everyone knows TSMC, but I don't feel that a lot of investors really understand what's going on on a structural basis. And so what I mean by this is if you roll back the years, frankly, until AI came along a few years ago, the US had and continues to have the reputation of leading the world globally in terms of IP design, ultimately coming out with the bestest, the fastest, the most powerful chips. And that was always the case. And that fitted in with the TSMCs, the foundries, essentially going to more advanced nodes, trying to fit more transistors into a single chip to ultimately make it more powerful on designs by the US. And then Taiwan essentially around that really did the kind of, you could argue that it was lower value add manufacturing components around it. So that was always a case for ultimately decades.

I actually think that's changing quite dramatically now. And I think you're seeing the bargaining power or the shifts in value add from actually IP design to areas that really were low value add, but actually now increasingly value add in the packaging particularly, and we'll come onto that and the component side. And it really, what's underpinned this is because there's this thing called Moore's Law, which I won't bore you with the details, but ultimately means there's probably a limitation in terms of the physical size a chip and the die can get. And so what this means is that it's now no longer a case of, okay, IP design, more powerful chip, send it to the foundry. Now, because of these physical limitations, they're now having to shift to what you essentially call chiplet architecture, which is essentially making ultimately smaller separate chips and putting them together to make this essentially super chip.

Now what that means is that with all the bundling together of these so- called chiplets and then having to put this, what we talked about before, high bandwidth memory and stack it on top, which is again, very different architecture to pre AI days, it means that you've had this fundamental shift away from actually just core IP design to actually the advanced packaging that ultimately puts all of this together, the value add and the importance of actually putting all these chiplets and memory together is actually where the increasing value add has gone. So putting it all together in terms of that supply chain advantage and the fact that that shift from IP design to the advanced packaging and all the components that Taiwan used to do, cooling, power, networking, that is the bread and butter and they're becoming more important in the AI realm. Actually, the bargain power and the supply chain has shifted and US are actually becoming increasingly more dependent on Taiwan. So I think this is a underappreciated structural factor behind Taiwan's outperformance.

Moz Afzal:

And I think also, I guess it reinforces what we've been arguing, certainly the market has been arguing over the last couple of months in particular, is this shift away from software to hardware in terms of the AI race. So for the last few minutes, let's focus on China. We've obviously got the Trump Xi Summit coming up and clearly China has a big vested interest in what's going on with the Straits of Hormuz, oil prices and with Iran and of course a piece in the region. What are kind of quick takes with respect to the summit and then maybe drilling down a little bit more in terms of the Chinese economy as we move forward?

Chris Chan:

Yeah. So when you look at the outperformance of China over the past year, it's really certainly in the last six months at least, it's really come despite the fact that the underlying economy has still been relatively weak. You see that in consumer, you see that in property. We talked about these consumer incentives rolling off. So that's really compounded the weakness in the consumer side. But of course, what held the China stock market up, particularly the Hang Seng, was the internet outperformance because everyone was appreciating the AI innovation. They're seeing this newsflow about these AI models competing with the US. And of course that's very much the case. But actually, of course, what's happened over the past two months is actually the Hang Seng tech index is now technically in a bear market. I think net positioning in the HS tech is at a six year low.

 

And this is really because of a couple of things. Of course, at the core is we can talk about the AI improvement all we want, but a lot of these businesses still have core businesses related to the underlying economy, e-commerce, of course, being one. And you're still seeing weakness there, which is leading to relatively negative earnings revisions. The second thing is you are getting a slight incremental negative on the regulation side, I would say slight. You're naturally getting policy support on the semiconductor side, but you've had various anti-monopoly inquiries on the travel OTAs. Also, you've had anti-involution, which is essentially trying to reign in excessive price cuts or price competition in certain areas of the internet platform. So that's made investors step back a little bit, but ultimately you've just seen a lot more competition. You've course had the same as you're seeing in the US, concerns about how are the inset platforms, are they actually going to be a threat? In terms of AI, again, travel OTAs, gaming companies, is that a threat rather than opportunity.

But all roads often lead to ByteDance, which is interesting. They came out with a very impressive and well-documented sea dance, which is essentially the Hollywood star video AI platform, which is by all accounts the best globally right now. It's the most used AI model or chat app currently. So that's scaring a lot of investors of other clearly listed companies like Alibaba and Tencent. And they're having to spend more money, particularly through Chinese New Year on promotions to get users in China to start using their AI models. So there's a lot more competition out there, particularly listed versus ByteDance, but the underlying momentum is still very good in terms of comparing the recent AI models from the China and the US. China's still very, very close. Their API usage or volume is now surpassed the US.

And I think where China's really excelling versus the US is they're being a bit more aggressive on the Agentic side and the enterprise side, and that's partly because they can do these agentic high token usage procedures because they have lower token costs. So they are getting a lot of traction there, which I think is positive for the overall industry. But yes, right now there's a lot of competition and frankly, low monetization in the space.

Moz Afzal:

So we have the next five-year plan coming up again in the next few days. Any thoughts about what's being talked about with respect to that?

Chris Chan:

Yeah, we don't expect too much policy response. I think despite the property market continuing, particularly in tier three and tier four cities, the pricing continues to trickle down. It's clear that the government and top level officials are, I wouldn't say comfortable, but not too concerned. I think it goes back to the base case. They want to control the tail risks in terms of property developers not going bankrupt and losing that confidence, but likewise, they're still okay to manage the deflation in areas like that. So we don't expect any significant policy announcements. We think it's just a continuation of what we already know, heavy support, heavy investment on the more advanced technology, humanoids, whatnot, and also still encouraging the consumer side because they know that that's obviously important. But I mean, in terms of the advanced tech, I'm sure a lot of people have seen this on their social media or on their news sites, but I would really encourage people to look at the recent China New Year gala, which through the robotics are getting a lot of news flow.

But I think if you want to look at the video, what actually is most important, compare 2026 Gala robot show to 2025, and you've probably seen this already, but it really shows just in one year how huge the improvement has been made in humanoids. I think China is unquestionably leading the US in at least the physical functionality of that. And that just reminds us as investors that the rate of innovation, particularly in China in these areas is extreme and there's a lot of opportunities to take advantage of in the supply chain here.

Moz Afzal:

Yeah, no, absolutely. I think that certainly is the case. And I guess we still can have more of the same, this sort of two-tier China where you've got obviously advanced technology continued to move at a pace as you just described. Same time, very slow on the domestic economy and turning that around out. I think that's probably the real dichotomy with respect to China.

Chris Chan:

Yeah, absolutely. Yeah. It's hard to see a quick fix for the domestic consumer and the domestic economy, but for sure that they're making great strides in the innovative side. So yes, it's a very much a double-edged sword.

Moz Afzal:

Yeah, no, absolutely. Well, Chris, listen, thank you very much for taking the time to speak to us.

Chris Chan:

Great. Appreciate it. Thank you, Moz.

Moz Afzal:

Thanks. So with that, we'll wrap up and thank you for listening, and we'll be back to you again soon.

 

This podcast is provided for general information only and assumes a certain level of knowledge of financial markets. It is provided for informational purposes only and should not be considered as an offer, investment recommendation, or solicitation to deal in any of the investments or products mentioned herein, and does not constitute investment research. The views in this podcast are those of the contributors at the time of publication and do not necessarily reflect those of EFG International or New Capital. The companies discussed in this podcast have been selected for illustrative purposes only, or to demonstrate our investment management style, and not as an investment recommendation or indication of their future performance. The value of investments and the income from them can go down as well as up, and investors may get back less than the amount invested. Past performance is not a guide to future returns, return projections, or estimates, and provides no guarantee of future results.