The View Beyond: Has the AI correction bottomed out?

Hello everyone, welcome to the Jules Barre podcast. This is Richard Tying, the Chinese strategist and head of research Hong Kong for Bang Jules Barre. It's great to have Hong Hao with us again to discuss the outlook for the Chinese market. How is the renowned economist and he's also the managing partner and CIO of Lotus S. Ad Management? Hi Hao, thank you very much for your time speaking with us today. Hey Richard, thanks for having me. Well, not surprisingly, most of our discussion today will be on AI and you know the market was still debating how much further the AI rally could go just two months ago. But now we already talk about whether the AI stocks have fallen enough. And if we look at the SOX, the Philadelphia semiconductor index in the US, that had dropped nearly 30% from its June peak to its July low. And the Korean market almost had the chin-ass index and the staff 15 that's in China also dropped by around 30%. Now, we've seen some rebound in the AI stocks and my question is, do you think this correction has already come to an end or we probably still some time away from a full-blown recovery? I think we have reached an important low, especially for the cost of the index because the index in the middle of less than two months with plunge almost 50%. Many of the leading Korean companies seeing the share price probably half or sometimes even more than half. So to me, a typical bubble burst process would take the share price from peak to drop about half to two-third. So now because Cosby is roughly half the price of his peak, so we can say that it's a typical bubble burst in process and it has reached an important low. But having said that though, because the retail leverage is still very high. We've seen derivative trading is quite down. Institutional leverage has been cut back, but the retail leverage is still relatively high. I think from the peak level, the retail leverage is off by 15% ish. So it is still very high level to be comfortable with. So I would say that even though we have reached an important low and also many of the leading semi-conductor companies having a technical bounce, but given the retail leverage still so high, I think we're probably entering a volatile phase of consolidation with the retail leverage being the trigger or the enhancer of market volatility. So I would say that now with the semi-conductor sector, we have entered stage of divergence in performance. The good quality names would perform well. I think capital would be happily flowing back to those names. But I think there are also some names that is basically taking along in the just because they are in the sector, but they didn't have good performance, good earnings, or they don't have technological barrier to entry. This names probably would continue to underperform. So we're entering a stage of divergent performance. It's interesting you mentioned that now we will have differentiation between those that deliver real earnings compared to those that take AI as a concept or the medic trading opportunity. And I think it's also interesting you mentioned Korea at the beginning of your comments. The size of leverage ETF that has fallen by more than half, but I think you're right that the retail margin level has dropped much less. And I think it's quite funny. I've seen increasingly more investors in Asia now compare Korea today with the China Asia market in 2015. And if that comparison makes sense, we probably still have some time to go before reaching the bottom, which I think is consistent with your conclusion as well. The other thing that we might have to look back and conclude is that this AI theme has been around for two or three years already. And the AI theme is still the same AI theme. But when it comes to stock market, it's not necessarily the same bunch of stocks that outperform at every point of time. For example, at the beginning of the AI theme, the leaders were Max 7 and VIA stood up particularly because of his dominance in GPU. But then in the first half of this year, most of the focus were on memory and semiconductors. These are the two segments that took the lead. So question for you, let's say at one point of time when this correction finally comes to an end, do you think the money would just go back the memory stocks, go back to the semiconductor stocks to drive the next rally? Or it might actually be something else. Could it be hyperscalers? Could it be AI applications, software, etc. What's the take on that besides the divergence between earning stocks and poor earning stocks? I think there are many sub-setters within the AI semiconductor sectors. Even with the memory chip names, you have NAND flash and you have DRM and HBM. There are very different technology being applied in different parts of AI calculation. So for me, it would be tricky to say which sub-setter will lead the way to our perform. For me, there are two main sub-setters. One is the GPU and the CPU. So they use for training and deduction. But then the memory chip names is really for memory and for storing logical chains and for real live applications. So they are very essential to the AI story. For the other sub-setters, such as the light transmission sector and also with the advanced packaging sub-setter, I think many of these sub-setters, they would have one or two names that is leading the entire sector. So I would say that for now, I would just go with the respective leaders within those sub-setters. I think that way, then at least you avoid putting your money into some of the names that doesn't have real earnings only have a story to tell. So I would say that basically invest your money with quality rather than stories. And from a geographical perspective, I think the AI discussion is still incomplete. If we only talk about US companies, only talk about Korean companies without considering China because the competition from the Chinese language model has actually become so intense. Right now, Kimi is a Chinese model that scores the best in terms of intelligence score according to artificial analysis and is only 5% behind clocked. And Kimi is 70% cheaper in terms of cost compared to his US peers. And we've been hearing stories about more companies even in the US switching over to the cheaper Chinese models for cost reasons. Now, indeed, this is reflected from the decline in the token expenditure index. But the obvious question that follows is whether this means Kappa is overall with that have to come down because if AI model companies do not earn as much or they do not have as much money to spend on investments. And if this happens, how would it affect the long term fundamental thesis of the entire AI supply chain? I think Kimi is doing well, scoring well, but I also heard that in many of these tests, they ask specific set of questions, mathematics, logical deduction and all that. And so it's possible to train your model to excel in this kind of tests. You know, it's a Chinese student doing exam cramp for the GALCAP for the University Entrance exam. So even though they can score very high at the exam, but then their capability is probably less than what their scores are actually showing. With the many examples of this, and also, you know, there's a sort of a whisper in the marketing that deep seek is going to launch as the new model G4 Pro. And from what I heard is that token is even cheaper than Kimi, but then the result is much better. So it's going to be a Western ROM killer. So I think it's basically two different way to approach AI development. One is the efficient lowest cost possible way to do it. And another way is the US way, which is as much as possible, scale up, claim your territory and blocking other new engines from entering. And also a very high amount of cat packs alone, you know, is a very large entry barrier for the newcomer. So it's very different approach and with different results already. So the Chinese AI models, they are claimed to be just as good as the US, but substantially cheaper. My way of understanding is that the cheaper token would induce an explosion in AI usage and applications, because you know, prices now is substantially cheaper or cheap enough for the personal users to start using AI on a massive scale. So I would say that it's actually good news. So lower token costs will actually induce substantially larger demand for AI. I think for the Western model though, because I'm pretty sure that the US would have technology embargo for US companies or US individuals to be using Chinese AI is similar to how a Chinese residents who wants to use American R&M, they will have to use a video and to bypass all the restrictions, but then it's substantially slowed down the calculation process. So they're, you know, processing minuses at this stage. I'm still seeing it as the two different ways of develop AI and use AI by saying to Chinese R&M and AI companies they're destined to achieve lowest cost possible and becoming a very important force of AI rolled out to the mess. So recently we've seen a very interesting statistic, you know, I've seen a number saying that the websites visited by AI robots is actually much more than number of websites reached by human. So basically AI is doing more research work for human at this stage. So I think we're at a critical juncture, I think this thing with the cheapest lowest cost possible AI tokens is going to accelerate the AI adoption in the human society. Right. And in July when all these AI stocks sold off, we were actually seeing quite a bit of rotation in the US market. in which banks and healthcare stocks were going up. And at the same time, despite all these financing challenges, equity investors seem to turn a little bit more constructive towards the US hyperscalers, because they think these guys may be a good, lagger plate. And I think one of those reasons is simply because Microsoft, Google, Amazon, they all report it very strong growth in the cloud businesses during the reporting season. I think what's interesting is that somehow, China also share a bit of this optimism, where the internet stocks rally, when the AI hardware stocks fell in July. So from the low in around June, the enhancing tech index is already up 15%. I think a lot of people do have that same question, whether this China internet rally can last even longer. What are your initial thoughts on this, how? - Yeah, I think the Chinese intern names are having a technical rebound, because it's been underperforming for almost a year now. And also the relatives and the performance has been so extreme that it all is well for a technical rebound. So I think by now, because the index, the Hengsen tech and also the Hengsen index itself has risen by well over 10%, more than 15%. And in this case, then I will say that the first phase of technical repairment is done. I think now we're getting into a stage, if people are looking for reasons for an extension of this rally. I saw the reason or the trigger for the next phase of rally can come from many things. For example, policy, support, earnings, report, and perhaps new development in AI, maybe Alibaba, Bebo to roll out even better, more cost addition model, et cetera, et cetera. I think many of these things can spur the rally to go a bit further from here. But I think for now, the initial easy money has been made and the indices here in China, they all have a very decent technical rebound. One point I often bring up talking about in the next sector is that the call revenue drivers are e-commerce, advertising in some traditional businesses, and these are highly correlated with the Chinese economic side fold. So my question to you, how is the Chinese economy was relatively strong in the first quarter and that weaker in the second quarter? What do you think the economy will do in the second half? I think most recently the proliberal urge to accelerate some implementation of the existing fiscal policies, but the amount of policy stimulus actually did not increase. So along that thought, what's your expectation in terms of the outlook of economic policies in China? I would say that we are not getting any signals from the recent political meeting in July. Basically it was a reassertion of the existing policy statement, which is keeping the fiscal policy proactive and keeping the monetary policy more literally loose. There's no change in wording from the last discussion. But there's a tested acknowledgement of challenges we're facing from the external uncertainties, especially from the Iron War. Because of the Iron War, China has purchased substantially less amount of oil by about four million barrels a day. And as such, the Chinese oil refinery industry and also many of the upstream industries are being affected negatively by the war. And by now, even though there are little bits of hope here and there, but in generalizing in people are expecting the disruption in the oil industry to continue well into 2027. Our price would be more prone to rise than to fall. Unless we have a satisfactory conclusion of the war and also reopening of the hormones straight. So I think with this set of external uncertainties and also fairly little hints on new policy support, I would say that the trend would continue into the third quarter. So I would say on a quarter on a quarter basis, more likely than not, we're going to see a very similar growth number in the third quarter, in the low force rather than high force. So many of us are looking to October, which is the time for the fifth plenum to start. Normally, during this time, we will hear a new policy discussion, new direction, maybe, for how best to manage the Chinese economy going forward. So remember in 2024, just before the plenum, there was a flurry of very high level meetings that then the central bank came out to give very strong policy guidance for loose monetary policy stance. And also the Sasi and the CBRC, the NDRC, they all come out and say they're going to support the Chinese economy in the Chinese market. And therefore, in late September 2024, it actually initiated a very strong rally in the Chinese market and also in the Hong Kong market. So I think many of us are hoping for such a very strong policy signal coming out in late September, early October, to coincide with the fifth plenum. So I think right now, the Chinese economy is lowing, but it's not struggling. I think some of the sectors are doing exceedingly well. For example, just now we discussed semiconductor, sector, the AI sector, human oil robot, sector, and many high tech manufacturing sector, and the export sector, they are doing exceedingly well. But I think for anything that is properly related, consumption related sectors, they are failing very poorly. So I would say that it remains to be seen, whether such divergence between different industries can continue. But I think right now, the divergence has been so great that it backs for intervention, especially consumption. If you continue to push for high tech industries now, that would just good initiative. But in so doing, we are basically increasing the importance of investment in the Chinese economy. And if investment is of the top priority, then naturally, so consumption would be a lesser part of the Chinese economy. So right now, Chinese consumption is about 58%, less than 60%, which is low by any standard. It really backs for intervention from the top, whether lowering techs, whether to improve social benefits and all that. We've discussed many of these potential initiatives in our past podcasts. But so far, we're just seeing manufacturing going from strengths to strengths. Expose to continue to rise. But then consumption is still very weak. Right. And my question that follows is, first year, we are basically following one investment theme, which is AI versus non-AI. You mentioned about your outlook for the Chinese economy, at least in the third quarter, probably similarly slow compared to second quarter. So does that mean that the non-AI would remain to be underperforming? Or do you think there are some other thematics or some other factors that drive rotation within the market? Obviously, in the past months or so, since our last episode, we're seeing a rotation towards the value of the economy sectors starting. In the last two episodes, we discussed how the impending semiconductor bubble burst or significant correction is coming. And it did come. So I think towards the end of July, with the Korean cospy down 12, 13% in one day. And that sort of give it a very castace of the market at the end of July. So we can say that the bubble or the significant correction has reached a very important role. And at the same time, we're seeing money is actually falling back to the value names. So even though we've seen significant volatility or applicable volatility in the semiconductor sector, for example, cospy experience the fastest plunge in history because of the semiconductor volatility. But then for the rest of the world, and especially in Hong Kong and in China, we actually seeing the index relatively stable, especially in Hong Kong. And also, many of the value names are being revisited by investors. And you can actually call Hensent. And Alibaba and many of the internet platform names they are sort of old cabish because they are basically operating a very large platform. But then they don't have enough AI exposure. So many of these names actually make a comeback in the months of July. And we think that the technical rebound, the first phase of technical rebound, is largely done. So I think in the next phase, we've probably gone and continue to see money falling back to the value sector. So one reason being the sectors that have been under the for so long, it begs for some attention because they got in so cheap. And secondly, I think because of the continued volatility in the tech hardware semiconductor names, they would make people who want to seek risk haven and they probably move some of the money into the value sector. So I think after the first phase of technical rebound, going to the next phase, we're basically looking for new factors that has not been priced into the Asia Plac right now. For example, policy initiatives, conclusion of the war, et cetera, et cetera. So many of these potential good news would have the power to extend the rotation further from here. Right. And from equity, let's now switch here to discuss Chinese bonds. Because in July, there is a listing of the futures of R&B Chinese government bonds in Hong Kong. Obviously new offshore derivative product that can be a hedging tool for international investors. How am I right to see this as another move to increase the international participation in R&B assets? What kind of implications do you think will have on the market? I think that it will make the Chinese such a bond market more open. more inviting for the foreign investors. Basically what happened is that back then in the Chinese industry, foreign market, many of the traders are domestic large commercial banks. They are a bunch of buy-and-hold investors. The strategy has served them well because in the last couple of years, in the Chinese long-born new, it has been continuing to trend down and making new or time low after low for the past couple of years. So if you're a buy-and-hold, foreign investors, you would have done exceedingly well over the past couple of years. Now we have reached a critical juncture where Bonu is at all time low, but then the risk of a reversion cannot be underestimated. You basically need a tour to sort of hedge the interest rate risk. I think for domestic traders, they have the treasury futures to play with, but that futures is mostly traded in domestic market. For foreign investors, if they have a very large Chinese-born exposure and has done well for the past couple of years, they probably want to explore new channels to manage interest rate risk. Now the Chinese-born market is probably the second largest in the world, but it's worth trillions of yuan and with very good liquidity. I would say that the introduction of such treasury futures for foreign investors, they will make them take a very serious look at the Chinese treasury-born market as a imestable as a class. Right. Now let's finish this podcast with gold. Clearly a breakout of gold prices from the downward channel since the beginning of the year. We've been discussing gold for quite some time now, and this seems to be quite an interesting juncture. How do you think the moment of recovery has finally arrived? If you remember in the past two episodes, we've discussed gold can have at least a decent technical bounce at a level that is slightly less than 4,000 and gold started to rebound at around 3900-ish, which is a very good and strong technical support level. I think by now gold has risen 500 US dollars already in a very short time as I'm in the middle of a week or so. So it is showing that the interest in gold is still very hot, and silver is having a similar rebound as well, and now it's bouncing towards an important moving average. In my view, I would say that the rebound will continue, but whether it's a trend reversal remains to be seen. I think a technical rebound of this magnitude and this strength is good enough for traders. And so as in right now, the silver futures have 66, as in gold futures at 4,400-ish. I think the rebound has to continue. We're probably going to see silver closer to 80 and also gold closer to 4,800. Let's hope that it happened, and it would give a much needed release for those people who got in at the end of February, and now they're probably still experiencing some substantial losses. Indeed. All right, that's pretty much all we have to discuss today. Thank you very much. How are you sharing? Ladies and gentlemen, thank you for listening, and stay tuned for our next podcast. Goodbye and speak soon. Thanks, Richard. [BLANK_AUDIO]