About this transcript: This is a full AI-generated transcript of China Optical Stocks Drop on Reports of US Ban — The China Show — 8/5/2026 from Bloomberg Television, published August 5, 2026. The transcript contains 15,957 words with timestamps and was generated using Whisper AI.
"It's nine a.m. in Shanghai Shenzhen and here in Hong Kong you're watching The China Show. I'm Averill Hong with David Ingress. Good morning. We're counting down to the open of markets in greater China. Let's get to your top stories today. Risk assets rallied to a track a record close on Wall Street"
[00:00:00] Speaker 1: It's nine a.m. in Shanghai Shenzhen and here in Hong Kong you're watching The China Show. I'm Averill Hong with David Ingress. Good morning. We're counting down to the open of markets in greater China. Let's get to your top stories today. Risk assets rallied to a track a record close on Wall Street and also in Europe by the way. Oil prices plunging in hopes of a deal between the U.S. and Iran.
[00:00:24] Speaker 2: SpaceX falls in late trade as fears over its A.I. spending outweighed optimism over its better than expected debut earnings. Also ahead on the program here we're watching China's data center component makers amid reports of a potential U.S. ban over security concerns. And sources say CXMT is set to begin production of smartphone chips to compete with rivals Samsung and SK Hines. Dave, it's really all about tech today. I mean, you look anywhere in the stack, these developments coming in, whether it's earning SpaceX, AMD in a way, but also the potential ban from the U.S. on Chinese components for data centers.
[00:01:18] Speaker 1: And then the progress that the memory makers seem to be making, whether it's Samsung or Changshin memory. Yeah. And Changshin memory, of course, getting into some of the areas of the likes of Samsung. Right. So we'll see what happens with the stock because, you know, we've been trying to figure out really at what point it becomes China's number one sustainably hold that. But you make a very good point that the earnings were a reminder that you are getting progress, perhaps not enough to justify some valuations out there. You are getting some disappointment markets, but it's not really showing up in markets because overall, it seems like the momentum trade is back. And I guess a softer dollar is helping that. Yeah. AI is back. It's getting its mojo, its groove on. Well, I, AI, right? Have a look at where we are as far as markets go. And we'll get into all those themes in a moment. But yes, you know, you have a softer dollar. You have lower oil. You have higher equity, higher equity prices. We are at a fresh record for the likes of Oz. Singapore is coming in line that has been trading in and around records anyway. So we are tracking gains that we had overnight. Taiwan is coming online fairly parity when you look at the magnitude of gains in the early goings. TSMC is coming up on your screens very shortly. We're at least up 3 percent in that. So some of the big tech. We're looking at a lot of these names, whether that's TSMC or Media Tech in Taiwan. You have the two, of course, in Korea. You have Baba and Tencent, of course, coming online in Hong Kong. In a couple of minutes, we're tracking those things very, very closely. We're looking at oil. So commodities very quickly, right? Oil has been tracking three, four week lows on that. Copper, yes, it's back at 14,000. So we're looking at the mining complex in greater China specifically. We are already seeing that rally play out in Oz today. We talked about that market being at a record. Copper in Shanghai, up half of 1 percent. Right. The read through going into the open today, A50 futures coming up on your screen. We have a dim sum bond auction, by the way, to tell you about today, testing the appetite for that kind of debt. And we are trading on the stronger side of three and a half years on dollar China today, 674. And we touched on this, the reemergence of EMFX or Asian FX and the breakout we're seeing across some metrics coming up on your screens. So Asia dollar index on the left, on the right side of your screens is the EMFX complex. And hence, we are also looking at the fix today on the back of these breakout moves we're seeing across the FX space here.
[00:03:39] Speaker 2: Yeah, fix also because of what Besson has said, right, about how he thinks the UN is undervalued. But also in terms of the market reaction to this potential ban that we're seeing on Chinese components for data centers. We're seeing early reaction in the Japanese optical names as well. The Trump administration reportedly planning to ban some of these components used in data centers to protect U.S. infrastructure. And that's according to Reuters. But Bloomberg intelligence thinks such a ban has limited effects. So let's bring in our team, China correspondent Mimin Lo and tech analyst Sean Chen for us. So Mimin, let's start with you because when it comes to some of these components, you know, we're trying to figure out how likely it will be actually put in place in terms of the ban, but also what companies might be affected, if at all.
[00:04:26] Speaker 3: Yeah, so Reuters is reporting that the government is looking to ban these optical transceivers. So these are the things that allow data to transfer the speed of light through optical cable cables in the data center. And China absolutely dominates this field. If you look at the biggest company, Zhongzi InnoLight, it has more than a quarter of the global market share. And Sean might be able to talk about this, but they estimate that Chinese companies make up more than half of the global market share. So if a straight cut ban does come through, it could be pretty chaotic for the industry, right? It would take time to find alternatives. And the downstream clients like the cloud providers, like Amazon, like Alphabet, their costs may go up if they have to source for alternative suppliers. And the reasoning behind this is because these transceivers are seen as potential entry points to install malware that could disrupt the entire data service, data center services and steal American data as well.
[00:05:22] Speaker 1: Talk a little bit more about that. The why involved. What's the reason we're looking at that?
[00:05:26] Speaker 3: Yes. So the House panel committee has been investigating a few years ago, if you recall, what was known as assault typhoon hacking attack in 2024. And U.S. intelligence officials had accused Chinese actors of breaching the telecoms network and hacking into the phone records of key politicians, including then presidential candidate Donald Trump. So this report, according to sources, is going to show how the U.S. telcos were compromised because of their connection to data centers and equipments. And those data centers contain some links to these Chinese telcos. Now, a few years ago, remember, Huawei was at the center of these sanctions. U.S. telcos were not allowed to use Chinese telco providers or components from them. And billions of dollars had been spent to rip out these Huawei components. But then their links to the data centers were not so regulated. And that's why they are now starting to scrutinize that.
[00:06:20] Speaker 1: OK, Min Min, Lo, thank you so much. Of course, this is a group of stocks we're tracking very closely today on the back of news. And I think it's important to note, probably showed it earlier, double-digit gains yesterday, right? So there is some of these gains to be unwound, of course, if you're on the other side of that trade here. Let's bring in Sean, of course, here with us. Having a look at the fundamentals, we know, of course, these companies get the bulk of their revenues from outside mainland China. How will this likely materialize from a sort of corporate perspective, fundamental perspective?
[00:06:48] Speaker 4: Well, I think to answer this question, there is one more question to ask first, is what is a Chinese transceiver? Is it a transceiver made in China or is it a transceiver made by a Chinese company? Sure. How do you define it? If the ban, if it materializes, is origin-based, then companies like Enolite and eOptolink, they can actually shrug it off because most of their shipments to U.S. customers now originate from Southeast Asia, particularly Thailand. If the ban turns out to be entity-based, then it could be a big hit to these companies because over 90% of the revenue comes from overseas customers. And out of the 90%, a large chunk is from the U.S.
[00:07:27] Speaker 2: Not to mention also the likelihood of whether this ban is going to materialize because this would also be very disruptive for U.S. hyperscalers. So how do you weigh out what we might actually see?
[00:07:38] Speaker 4: I think you're absolutely right that it's almost impossible for U.S. hyperscalers to find immediate alternatives to these transceivers because we believe Chinese suppliers, including eOptolink, Enolite, and Dongshan Precision, which acquired source photonics, they have more than half of the high-speed data com transceiver market share. So the hyperscalers, they are most likely going to push back the proposed ban. And because of the severe disruption, we believe the reported ban will at least be softened.
[00:08:11] Speaker 1: Where across the optical networking supply chain do we need to be looking at more closely now?
[00:08:17] Speaker 4: I think there are two groups of companies we can pay more attention to because we think under the simmering geopolitical tensions, they might be in a better position. The first group is the companies focusing on China's domestic substitution story. So these companies, their fortunes are not tied to U.S. hyperscalers or U.S. AI infrastructure build-out. For example, companies like Acelink and Huagong Technology. The second group of companies is those who also rely on U.S. hyperscalers, but they are not the downstream module assembler. Instead, they make upstream or midstream components, such as optical engines. So companies like these include Suzhou TFC Optical.
[00:08:56] Speaker 2: Is there a way that China might retaliate if the ban does indeed come into effect? Because there's also these rumblings that maybe this purported announcement might be a bargaining chip ahead of Trump and Xi meeting also.
[00:09:10] Speaker 4: Yeah, I do believe China has a strong card in the optical supply chain to play as well. That is Indian phosphide substrate. Because in addition to China's strong position in the downstream transceiver modules, which relies a lot on the U.S. hyperscalers, China is also strong in upstream Indian phosphide substrate, which is critical to make the laser chips. And China already implemented export control from early 2025, but has so far allowed exports to avoid disruption of the supply chain. If U.S. indeed implements the ban on Chinese transceiver makers, it won't be a surprise to see China tightening the upstream Indian phosphide export.
[00:09:54] Speaker 2: Sean, great to have you. Thank you so much. Thank you. Bloomberg Intelligence Tech Analyst Sean Chen. And of course, as we were alluding to a short while ago, when it comes to context being important and how we saw that strong bounds in the optical names just a day ago, we'll see how China fares. But also against the backdrop of Hong Kong China overnight leads, they look like a bit of a laggard into the strong tech rally. Let's bring in Bloomberg Markets reporter, Anthony Stephens. So what do you make of this?
[00:10:21] Speaker 5: Yeah, it increases the possibility that China is going to get robbed of today's technology rally, right? From a risk management function, even with the fundamental underpinnings, you have to cut some risk. This is a very volatile space. Margin balances have been coming off. The space is off around 30 to 40 percent from the highest. So people will have to manage this risk. It's it's the biggest part of the Chinese index and the Chinese index has led the HSTech on the way up. But recently that started to revert a little bit. Right. So from a long shot book management perspective as well, people are under pressure as the HSTech, which is a larger market cap index, has started to rally versus the China. So people are missing out on the margin in the in the rallies in Tencent and Alibaba. And on that front, there is a U.S. kind of policy that is helpful, right? They're not they've decided to not check open rate models for security, right? They're loving U.S. technology companies to use open rate models. So on one hand, the U.S. gives China something and on the other hand, they give them the takeaway. And that affects the long shot book positioning of investors in Chinese tech, right? The hardware space looks like a little bit more kind of threatened by the geopolitical environment. And it looks like the LLMs in the software space could have a little bit of room to run. That being said, the focus this week looks like very much back on Korea and Taiwan after the very strong guidance from SpaceX and AMD.
[00:11:45] Speaker 1: Well, speaking of, how are those those latter two things affecting trade today in the region?
[00:11:50] Speaker 5: So it's very interesting that those two stocks have actually come off since earnings, right? Like this, the sell on good earnings dynamic remains very strong in the U.S., given maybe that the markets are at all time highs. But in Asia, we are taking a much simpler reaction function, right? Like Taiwan is very strong, given that AMD is talking about a massive compute number. I think it's two trillion or something by 2030. They are estimating 40 percent annual growth in AI. So the read through from U.S. earnings into the region is very strong. SpaceX, similarly, like say 18 billion CapEx, yeah, sure, it's a little bit less than people expected. But 18 billion dollars in CapEx from a company that was not even listed last year is in a material needle moving for Taiwan and Korea. So for the rest of the region, you're seeing these huge gains across the AI supply chain as the U.S. sticks with their CapEx guns.
[00:12:42] Speaker 2: So momentum is back?
[00:12:43] Speaker 5: Momentum is back, defined as tech momentum, right? You know, it's not everything at the same time. We're seeing this, like, quality dispersion, especially in the Japan's earnings season. You're seeing the companies that are giving good earnings getting rewarded. But any miss, hugely punished. So momentum, you will get the momentum if you have the earnings growth behind it. So it's a little bit more subtle story than the first half of the year.
[00:13:06] Speaker 1: Okay. Anthony, fantastic. Great context as we go into the open today. Of course, it looks like we will be, here in Hong Kong, we'll be robbed, of course, of the feel-good vibes we're seeing across the region here today. Right. You know, we're covering these AI arrays today from almost every angle. I was going to say every angle, but we'd be selling you guys short there. Coming up, at least most of it, Goldman Sachs will be joining us to talk us through their latest targets, price targets on Jirpu and Minimax and why, perhaps, they think one over the other? Question mark. We'll be joining us at About the Open today. Also, in the next hour, Rhodium Group's Riva Guzon will be joining us to talk us through their tech report guide. You'll be surprised, in fact, what they found there as they stack U.S. and China and their respective capabilities. Counting down to the Open of Trade in Shanghai, in Shenzhen and here in Hong Kong. The opening bell, 16 minutes away. This is The China Show. Welcome back. The midpoint of the day is slightly more relevant than other days, given the rally we've seen in EMFX. And obviously, at the start of this week, we've had a yen intervention story. Today, the midpoint in the onshore, certainly, is 678.89 against the U.S. dollar. The estimates were for somewhat stronger, but that has been the case anyway. It's the delta. Just to talk about this, because in the offshore markets today, we have a dim sum bond auction. So just watch out for that as well. That's it. Anyway, why don't we take our heads from Earth to space?
[00:14:46] Speaker 2: Yeah, bringing things back more grounded, right, or the other way around, I should say. Yeah, big picture. Because we are, I mean, you just talked about how this yen fix is playing out, but we've also been seeing the dollar softness or a reversal of the rebound. Maybe that's also what's helping, to your point, right, the macro backdrop, helping some of these tech names in Asia as well. That's right. Although the exception seems to be SpaceX and how it fared in late trade, it closed lower. And this was after disclosing higher than expected spending on its AI business. That's despite its revenue exceeding estimates in the company's first earnings report after its IPO in June. Bloomberg tech anchor Ed Ludlow has the details.
[00:15:29] Speaker 6: SpaceX spent a staggering $18.4 billion in the quarter, and the vast majority of that, 86%, was on AI infrastructure. But renting out compute capacity or computing power to companies like Google and Anthropic is becoming a major business, and they gave lots of evidence of that. Second, like, where is the growth? SpaceX said they will hit $100 billion in annual revenue run rate by the end of this year. And Musk even went a step further, saying they have line of sight, like complete confidence, that SpaceX will hit $1 trillion of real revenue in 2030, pulling forward a target that they set out in their IPO prospectus. Third, Starship. The next test flight could happen later this month. The rocket's heat shield problem that they had in recent tests is essentially solved. And SpaceX expects to begin launching its next-generation Starlink V3 satellites and its first orbital AI data center satellites next year. In that case, with NVIDIA compute architecture. Finally, watch out. AT&T, Verizon, and T-Mobile. SpaceX says it's building a hybrid satellite and terrestrial mobile network, and believes it can win customers from those U.S. carriers by eliminating dead zones. And the stocks of those carriers reacted sharply during the course of SpaceX's first-ever quarterly earnings call. The story is one about growth, short, near, medium-term, and long-term, but a lot of momentum with SpaceX. This is Ed Ludlow for Bloomberg News in San Francisco.
[00:17:05] Speaker 2: Dave.
[00:17:05] Speaker 6: Yeah, what's up?
[00:17:06] Speaker 2: SpaceX is also... What's up, SpaceX. Yeah, I'm here. Not the stock. You know, it's some of its AI plans, I suppose. But it's also interesting in the context of China, right, because there were those rumblings that maybe they're trying to take the Chinese out of their supply chain in a way. And the reports about Tesla, although Musk kind of debunked it.
[00:17:26] Speaker 1: Yes, yes, that's a big picture.
[00:17:27] Speaker 2: And then people talking about how, you know, when it comes to Tesla and SpaceX, a merger is a matter of when, not if. So that was something else that we were watching out coming into today as well. But I guess the geopolitical backdrop when it comes to today's market reaction that we're expecting on the Chinese optical names, given this reported ban underway, right? And against that backdrop, also, some of these other, I guess, macro data points, services, rating dogs.
[00:17:54] Speaker 1: Rating dogs coming out, I think, today in about 20 minutes or so. So we'll see what happens with that. Obviously, we've talked about a lot of the stocks and sectors we're tracking from copper, optical, of course. And we're still waiting. In fact, I'm just zeroed in now on my terminal. Because we should be getting pricing in the pre-markets of Zhongji Inolite. And on the back of a really strong rally a day prior, keep in mind, $1,000 is what we're tracking today. It's looking like we could take that out. The IPO price in Hong Kong several days ago was about $900 something. And we've rallied certainly since that point in time. Low Turner, we've talked about that. You want strength, fine, Catholic Pacific. Okay, we now have pricing down 15% in Zhongji Inolite. Eight shares are trading about 15% lower. Just to also mention, do not be surprised if you see the A-share counterparts, which should be giving you pricing in a couple of minutes. Also to see similar declines on the back of that report coming out of Reuters that the U.S. is drafting some rules here to perhaps ban some of these components from U.S. data centers. We'll have more ahead, including a fuller preview of the trading session ahead. This is The China Show.
[00:19:04] Speaker 2: The China Show. reported U.S. ban to come on Chinese components for U.S. data centers. And we're already seeing this massive reversal from yesterday's rally in the likes of Zhongji Inolite, right? Optical name that was part of the beneficiaries that we were tracking just a day ago after NVIDIA announced its co-package optical CPO has entered mass production. So that was a driver but then you have these reported bans that could come and things are starting to change. But against the backdrop of all that, let's bring in our Bloomberg Markets reporter Anthony Stevens. So you've been watching that in terms of movers, market reaction, what else are you watching?
[00:20:16] Speaker 5: The fact that HSTech is opening flat is not bad. You know, it's participating a little bit. Given the overall geopolitical kind of headwinds, it's not a bad performance. It will be important to see how the Chinese semiconductor ecosystem performs as a whole. So we're talking about foundries and memory. So CXMT will be very interesting to watch whether it continues to rally alongside Samsung or Hynex. So does it start to trade as like a China play constraint from the U.S. market? Because it's important, right? We've heard these reports that U.S. companies are looking for CXMT chips, not only Apple but recently HP as well. So whether they can get CXMT chips but the AI servers can't get photoelectrics, that's going to be a very interesting dispersion. So that trade is very much alive. You know, what parts or what parts of the supply chain does the U.S. want from China and what does it want to curb? Because that's now an evolving space. It's a debate that didn't exist before. Right. And now we have this kind of relaxation of policy on Chinese open weight LLMs. And how does that play out? You know, the software space in the U.S. has been rallying very sharply because they have these cheap tools to work with. Now, that's another evolving debate. Taiwan and Korea, much simpler, just going up.
[00:21:31] Speaker 1: Yeah. And should we be surprised if we see a similar move down when Shenzhen comes online and the China X index in the similar names, the Asia counterparts.
[00:21:41] Speaker 5: Yeah. Look, I would not be surprised at all. Given the scale of risk management that needs to happen to adjust to something like this, right? The local retail is very long this space. You know, the government is trying to protect them from themselves with some of this kind of national team buying in ETFs and whatnot. But ahead of fundamental news like this, will they bother to step in today? That is an open question. The other thing, since you brought it up, right? So what's leading or what's keeping
[00:22:09] Speaker 1: Hang Seng tech decent today, Minimax is up four, trip was up about four percent. Unsurprising to you. Yes, because of the open weight LLM change, right?
[00:22:18] Speaker 5: We needed the U.S. Government to pick a lane. And therefore, now, they've gone with the side of the largest software industry to help American companies get cheaper AI to produce the software, right? Like, Palantir, for example, yesterday was up 30 percent, right? That's a very painful short position for long shots. So that'll be very interesting to watch. Anthony, fantastic.
[00:22:39] Speaker 1: Thank you so much. Great setup as we go into the opening bell today, just under four and a half minutes away here. So we're looking at that, obviously, copper-related stocks, Cathie Pacific's reporting earnings. The opening bell is coming up next. This is The China Show. Good Wednesday morning. Welcome back. You're watching The China Show. Before we get inundated with using the equity markets, we will when you see these moves in the markets. Just to mention, there is a rare dim sum bond auction in the offshore markets today. A few tenors, a couple of billion RMBs, so just watch out for that state test to get appetite for the curve.
[00:23:41] Speaker 2: I see what you did there. I don't know what I did. Pun intended? Or unintended? I don't know. Who knows? Let's live with the consequences. Yeah. Let's also take a look at how mainland markets are faring, right, at the open. I mean, we are setting up for perhaps a violent reaction. We're seeing the earlier indications already in Chinese optical stocks. But this is also coming against the backdrop of the memory names being in the spotlight, likes of Changsin Memory, for example. They seem to be making, seem to be making some headway when it comes to, you know, smartphone memory. And then, you know, this is against the backdrop also of what Samsung seems to be laying out in its plans to compete with SK Hynix, with Micron as well. So maybe AI hardware, maybe they have a shot, you know, the comeback. We're talking about how momentum seems to be creeping back up again. We're looking at, in terms of the mainland reaction, we're down. Perhaps no surprises there. Of course, we have been taking a look at some of these, you know, he talked about dim sum bonds. We're looking at the 10 year yield, sitting flat 1.7 there. Let's also take a look at how things are faring with that softness in Changsin Memory. Take a look at how Hang Seng is setting up for today against the backdrop of these geopolitical developments, right. So we have been seeing that early reaction already in terms of these optical names in China and the ones in Japan, right, the other side of the coin, I suppose, that are gaining. And at a time where, of course, investors also still, I guess, digesting what we got from HSBC just a day ago. If you can flip the board and take us there. Following on from that share buyback disappointment, that's another stock that is continuing to come under pressure, pulling back from the highs, I should say. On a day where it looks like for the finance names, those are the ones that are in focus, not just in Hong Kong, but also in the region. We see how that's propelled the likes of the ASX, even the Singapore benchmark. Yeah. So that's something that we're watching as well, given the tech rotation or out of that rotation.
[00:25:44] Speaker 1: Well, speaking of Singapore, just to mention, isn't DBS coming out with earnings very soon? OK, just put that on the radar, on the back of this broader financials.
[00:25:51] Speaker 2: Tomorrow, we'll get that. And, you know, they've been doing well there. So I think so. This is the earnings test that we'll watch out for. OK. And then taking a closer look at what we're seeing in Zhongji, InnoLight, just to recap things for you. Close your eyes if you're long. Oh, OK. So 10% down, A shares, H shares, both in negative territory. And then, you know, you're seeing that reversal from the rally that we saw a day ago. So actually, maybe not that bad.
[00:26:23] Speaker 1: I mean, we're coming off lows already quite quickly.
[00:26:25] Speaker 2: Yeah. And this was after yesterday's rally, right? Right. So put that into context. And the likes of Sean Sean, right, already talking about how, you know, how will they implement this ban if this also disrupts things for U.S. hyperscalers.
[00:26:39] Speaker 1: Where do you get the replacement, right? Because your top two in the world are effectively on your screens right now. So being able to replace that in the supply chain is easier said than done. Yeah. But, I mean, we'll really have to.
[00:26:52] Speaker 2: There's no alternative, I suppose.
[00:26:54] Speaker 1: Yeah.
[00:26:54] Speaker 2: Let's quickly take a look at some of the other movers as well. We're watching copper stocks. We saw the commodity rally. You're seeing them in the green, the likes of Jiangxi, these Hong Kong listed names for the board again. Let's take a look at the likes of STO after it was proed for safety management failures, following on for multiple workplace safety incidents. So you're seeing that reaction there as well, Dave.
[00:27:21] Speaker 1: Yes. This is the logistics. Yes. We are down on that stock specifically today. Well, let's focus in on what is moving higher. I think last week checked the reason why Hang Sintech was doing quite favorably amidst the rally, I mean, relatively speaking, not so much, of course, but okay, given the news flow, is the stock bottom of your screens, one of the ones leading the gains here in Jirpoo. Minimax is also coming up very shortly. Just keep an eye on what the share price is. A thousand of that. Minimax also should be coming up on your screens very shortly. 240, let's call it that. The reason why we bring that up is because Goldman Sachs has recently just put out their revised price targets for both those stocks. They do expect competition among these Chinese AI developers to further intensify, increasing, I guess, the emphasis here on achieving best performance to cost balance. And they've also, as we were just pointing out, they've not just raised the price target, changed the price target, they've raised their revenue expectations there for both companies. Joining us here on set is Ronald Kung, head of Asia Internet Research at Goldman Sachs. Good morning and thank you for coming in. Good morning. Okay, lay out the thesis on the thesis in both the changes you've made the price target and why you've why you've done that.
[00:28:30] Speaker 7: So we put out our L.M. primer and followed on on each of the weekly to developments from cost efficiencies. For example, last year in deep seek to recently, the whole focus is on model intelligence and the move towards frontier performance and increase in pricing power of Chinese models from extremely cheap models. So now there is some pricing power. But we think from now on, we're talking about performance to price frontier, the performance price ratio, which one has the best performance versus what the underlying cost is. And as we see on a weekly basis, more and more models to be launched, we expect the 3 trillion sized class parameter models, there'll be at least five or six of those in two months time, two to three months time from China. And so while the frontier performance continue to be pushed upwards, the cost efficiencies of those become more and more important. And therefore, we introduced a competitive positioning score looking not just at the frontier performance, but what are the cost efficiencies, particularly the margins for inference. That is kind of the core essence of the whole business model is to generate margins of inference and eventually cover training costs, eventually cover the overall cost, be a profitable business longer term, but also the financial strength. That is key to how long duration of this competition could last and who eventually consolidates to the top AI model players from China, similar to what we see in the kind of developed markets in China. We think there's still a process to go through for these players to eventually move up to consolidate towards the top and best position.
[00:29:59] Speaker 2: So financial strength important that would determine who the eventual winners are. But talk to us about what you're seeing for the moment in terms of cost to performance basis, how you assess, you know, the stack, who's ranking high among the AI model developers.
[00:30:15] Speaker 7: Yeah, so we saw the first scaling up to the 3 trillion size parameters for K3. And now we're talking about some of the new latest developments in the past week was the DeepSeq V4 Flash, which is a 200 ish billion parameter model, but much, much smaller. But with post training, with all of that supervised fine tuning, we call SFT, the performance has reached closer and closer to front end top three and an open way to call into arena dot AI. And we focus on arena AI because of that voting. It's a it's a blind blind box and voting from actual users. And therefore, we also look at on the video side, Minimax H3 has has now is open weight. So these open weight models with much smaller size cost efficiency seems to have moved up in performance as well. But of course, they're much smaller models, so we can't expect it to perform like a 10x larger size model in terms of the full knowledge base. But the the process and the actual performance in coding has improved and coding is generally, in our view, a yes, no answer. And therefore, the improvements of that coding front has been very significant for the Chinese models.
[00:31:17] Speaker 1: I mean, they're coming out with new models. It might sound like I'm exaggerating, but almost on a daily basis, daily basis now. But I guess my question is, how different really are they from one another? And do we have enough information to determine early who might be best positioned to bring among the Chinese models, reign supreme amongst that group in, let's say, in the next 12 months?
[00:31:40] Speaker 7: So we also saw the Qoen 3.8 Max. Yes. And that has been also quite high in the ranks. But as you said, the rankings are moving on a daily basis. And therefore, what we're seeing is a harness layer is now coming in. Who could conquer or gather biggest traffic for the harness layer, the agent layer, where if you go into the harness like a work buddy or the coder work, which is now called Qoen work, then you could switch models at whatever time you want. But all your work flow and agentic functions that you've set up are still in that harness layer. So we're seeing that amongst the Internet mega caps, actually, that they're focusing on the harness layer. Individual AI companies, independents, are also launching their Z code, Minimax code, Kimi code, and all of that different code equivalent, which is that agent harness layer. You could switch models at whatever time. But that harness layer is one way for us to get more positive data feedback loop to improve your own model, which is actual usage of coding to improve your own model that drives the further improvement on model performance.
[00:32:41] Speaker 2: So the drive helps revenue. But what about monetization? I mean, where do the profits come?
[00:32:48] Speaker 7: Yes. So we talk about the open weight. So it's not 100% open source. And what we're seeing is some of these commercial terms with commercial terms licenses are now growing for these AI open weight models. So if you generate more than $20 million US dollars a year using these models, then you have to discuss a term on the monetization sharing with the model owners. And we're seeing that. So we've lifted our year-end AR numbers to $13 billion US dollars amongst AI models by year-end this year. Previously, $10 billion just a month ago when we published the primer. And so that increases mainly on this very strong adoption of token numbers. We've seen that from some third-party platforms like Open Router. The token numbers are growing. The API channels is where, to your question, that they're monetizing on every token, pricing, input, output, and kind of KB cash, all of that blends together. But that AR is ramping up. But not all of the models go through their own API channel. So, therefore, this community license is opening up a new venue for US hyperscalers, neoclouds, or to house these models and a new revenue sharing kind of stream of revenues that could drive these Chinese AI model revenues.
[00:34:00] Speaker 1: Do you see, this is for the independent AI labs, do you see a path to profitability? Does it matter that they're currently perhaps not profitable from a stock analysis perspective?
[00:34:12] Speaker 7: So, we've got to look into the unieconomics and at what level of gross margins will cover enough of your token numbers on a daily basis, margins for your inference that covers training cost and EBIT. In our estimates, more like 2028 to 2029, depending on which of the frontier level or more kind of agentic, which is currently pricing, very, very low prices. But that will drive further and further adoption and a path closer to AGI. Therefore, the agentic AI is where our government research globally and our US teams are expecting the 24x in token numbers from now to 2030. That increase in tokens will likely bring a much bigger pie of tokens and then the revenue pool and the gross profits that could cover the business models of these companies. But we're also watchful for regulations, all the different things that could impact the swing of the revenue forecast that we have for the Chinese AI model.
[00:35:07] Speaker 2: It sounds like also, as you alluded to, workplace applications seem to be the next realm of competition. Where does this leave the older internet names? I mean, how do things stack up then?
[00:35:21] Speaker 7: Yes, so amongst the mega caps, ByteDance, Alibaba, Tencent, they have been focusing on various aspects of Frontier. Either the Quen 3.8 Max that we talked about, or Tencent, Huan Yuan 3, HY3 is actually a much smaller model, but has very low cost and a much lower pricing setup. And then we think ByteDance is a multi-modal focused, which means not just the text model, Doubao, and a consumer-facing app that is number one in China, but also have a very strong multi-modal video generation called SeaDream, SeaDance and SeaDream for Image. So SeaDance is where recently the H3 from Minimax is coming up with an open-weight approach. So if we looked at the deep-seek moment, which is the open-source approach for text models, then the recent launch of Minimax H3 actually bringing an open-weight really useful and comprehensive model up to the open-weight. So, yes, competition will increase, but our mega caps are first focused on keeping themselves in the model layer, but also the harness that we talked about, and we haven't talked about yet, is cloud and data centers, which is our top sub-sector in China, Internet.
[00:36:26] Speaker 1: Is that, okay, so my question is, what's the best way to play China AI now?
[00:36:30] Speaker 7: Across, we think, we believe the growth in tokens and the capex uplift that we're anticipating for the second half of this year as domestic chip supply increases. Cloud and data centers are our top preferred sub-sector, and that's where the hyperscale is, which are the Internet mega caps, our best position with strong cash flows, the ability to fund the capex and grow a cloud business, where we're estimating a 45% growth for Alibaba cloud business in the June quarter that we anticipate. So these are the key numbers that are actually driving the growth is driven by the token numbers that we talked about across the model stack and the agentic-driven tokens.
[00:37:10] Speaker 1: Fantastic, yeah, and some of the names there, GTS, VNet, Alibaba, and Kingsoft Cloud. The price target, just to mention, on Jerpu is 1610, Minimax is 800. That was Ronald Kung, head of Asia Internet Research at Goldman Sachs. Right, ahead on shows, you're not going to want to miss this one, okay? You're going to be quite surprised to see, actually, some of the results here. So Riva Bujon from Rhodium will be joining us in about 30 minutes from now. So they've actually done an analysis on several metrics, how the U.S. stacks up against China or vice versa across, not just chipped, but the entire, everything that needs to go into an AI supply chain. They've stacked that. The report card is out. We'll talk about that at those times on your screens. Lots more ahead. This is Bloomberg. Rating dog. Not our favourite craft beer. That is the latest private survey that's coming out today. So we're looking at composite and services. We're looking at 50.4 and services, above 50. That's coming in, though, a few points shorter than estimates there. On the composite side, also an expansion at 50.8. I should note, while in expansion, both are a drop from the previous month and the other one, of course, missing estimates there.
[00:38:45] Speaker 2: And this follows on from the manufacturing PMIs that we've been seeing softening as well into contraction. Let's talk about memory, though, because we've been trying to dissect among the top three so far. Is there chunks in memory that's, you know, really giving them a run for their money? CXMT is said to be poised to produce smartphone memory chips on par with the industry's most advanced designs. Our chief North Asia correspondent, Steve Engel, is with us. Steve, what do we know?
[00:39:15] Speaker 8: Yeah, this would be a very interesting development, obviously, because CXMT really coming on to the global stage as far as, you know, awareness and with their big listing and the like. Only recently, of course, there's that big triopoly between SK Hynix and Samsung Electronics and Micron that really dominate the memory chip space. And obviously, the industry is faced with shortages with all the developments in AI. So on the one hand, if CXMT is coming out with the advanced sixth generation, low power, double data rate, DDR chips that would be on par with the most advanced chips from those three big companies, that would be a significant development. It's something that some even in Silicon Valley have called for because of the shortage, right? But again, CXMT is on the Pentagon blacklist. So there's all kinds of different export restrictions that could come into play here. But again, this is the kind of memory chip that would allow, you know, lower power, faster efficiency, power efficiency. I should say, faster speeds that allows you to do AI apps locally, so on your smartphone, on your laptops. So this is according to a source right now. So we'll have to see if this can come to fruition. It would be a big development for Beijing's pursuit of self-sufficiency in these kind of indigenous technology developments amid those export restrictions.
[00:40:43] Speaker 1: And not that challenges have stood in the way for the most part, they've found a way to solutions for that. But what remain the biggest challenges here that they need to sort of dissect?
[00:40:52] Speaker 8: Well, to produce one is fine. Can you do it at scale? They can probably do it at at a at a cheaper price point. But can you reach the scalability and the efficiency of the longstanding players? That would probably be the biggest question. And can they keep pace with the rapid development and the pace of development? Because SK Hynex, for one, has their most announced in March, I believe, their most recent sixth generation low power, LPDDR6. That's the next generation. Also, Samsung Electronics has talked about they have a similar thing in the pipeline. And SK Hynex is talking about just from one generation to the next, 33 percent faster speeds and also 20 percent more power efficiency than the LPDDR5X. So, again, it's can you build it at scale and can you keep up with the efficiencies that the existing players have?
[00:41:46] Speaker 1: Well, Steve, thank you so much. Stephen Engel there. Our chief North Asia correspondent, market reaction, if any, on your screens. We're flat on CX-70. We're up in Korea. But that's perhaps down to most other tailwinds. We're feeling across the risk asset space here in the Asia-Pacific today. A look at markets on that note. Lots more ahead. We'll take a short break. This is Bloomberg.
[00:42:04] Speaker 2: Here are some of the corporate stories we're following. Paramount posted a surprise surge in second quarter profits as cost cuts tied to its Skydance merger continued to pay off. Adjusted EBITDA rose 27 percent to $1.1 billion while streaming profits jumped. And Paramount Plus subscriber growth topped estimates. The company said it still expects its proposed $110 billion acquisition of Warner Brothers Discovery to close despite mounting legal challenges.
[00:42:55] Speaker 9: We remain highly confident that this transaction will close. And we're preparing for basically a successful combination once it does. If you take a step back and just look at exactly where we are today, we received approvals from basically, you know, 65 regulators and 65 countries around the world, you know, including the United States federal government, Canada, European Union, China, and many more.
[00:43:23] Speaker 2: Blackstone is said to be exploring a second mega debt package to finance Anthropics use of Google AI chips. Sources say one initial proposal was for at least $36 billion of debt. If finalized, the deal would exceed the $35 billion of debt lined up by Apollo and Blackstone some two months ago to fund Anthropics use of Google's custom chips, Bloomberg has learned that Anthropics has struck a $10 billion deal for computing capacity from Nvidia-backed startup Volta Infra. Volta earlier said it secured a deal with an unnamed AI lab, and the agreement will run for six years. This is Anthropics latest effort to secure the computing power needed to meet its growing demand for AI products. And AI models from OpenAI as well as Anthropics have been involved in previously unreported cyber security incidents. The UK's AI Security Institute says during an evaluation, Mythos 5 and GPT 5.6 sole models engaged in what it called sustained potentially harmful activity against people and organizations. OpenAI and Anthropics are now working with the Institute to investigate the incidents. And AMD shares dropped in late trade after the chip maker gave an underwhelming sales outlook. Its third quarter revenue will come in around $13 billion, missing some of the most bullish estimates. CEO Lisa Su told analysts the company expects its data center revenue to more than double in 2027, with growth that could come in at more than 100%. And it's interesting, we're seeing the aftermarket performance in AMD. It's negative, but the CapEx and how they've leaned into it, along with the likes of SpaceX, that seems to be propelling tech stocks in the region.
[00:45:14] Speaker 1: I mean, it really is, right? Someone's cost is someone else's revenue, right? And I think it's there we go on your screens, right? Perhaps that's the further down the supply chain is how the markets are playing it. These hardware names are, as you can see, seeing some decent upside. Hang saying tech, up 6 cents to 1%. Some of that, of course, goes into that other benchmark you see bottom of your screens. Part of that also is we are still getting a rally across some of the AI labs, AI houses like Jerpu and Minimax. Interesting, the price action in the last 30 minutes is we are still down, but we're coming of lows quite quickly. We're now at about half the losses in the opening minutes for some of these optical stocks. They're trimming that. Solar stocks on the way up last week checked. We'll talk more about that. Plus, of course, the broader markets, correct myself, we're now down in that. Lots more ahead. This is Bloomberg.
[00:46:01] Speaker ?: We're now down in the middle of the last 30 minutes. We're now down in the middle of the last 30 minutes. We're now down in the middle of the last 30 minutes. We're now down in the middle of the last 30 minutes.
[00:46:19] Speaker 2: Welcome back to The China Show. Here's a look at the Hang Seng half hour into the session. Of course, David, we've been watching these confluence of tech factors when it comes to the headway that the memory maker seems to be making, but also the tailwind from the SpaceX AMD capital expenditure that seems to be driving maybe more of that AI momentum in trade today. That's coming up. And we're seeing, though, moving off those session lows, the Hang Seng. The drag today that we're seeing is those financial related stocks, including the likes of HSBC. Of course, just a day ago, we did see them, you know, disappointing on the share buyback. That's among the stocks that we'll be watching in the session today.
[00:47:06] Speaker 1: Yeah, I mean, the stock done very well, right? I think we're down 1.4 percent on 5HK. We're just getting some lines coming through. This perhaps looks to be related to notes. So they've offered to buy back notes of up to looking at 5 billion. So they've announced a tender offer for four series of notes, I think, just to talk about, of course, the use of capital in as far as the capital structure is concerned. We touched on the bid we're seeing across the supply chain today, right? So the Nikkei is showing up quite nicely. Cosby is up 4.6. We talked about Lenovo and some of the hardware names here in Hong Kong. Taiwan up 3.3 percent coming up on your screens. And of course, the next board is going to show you some of the individual names across these markets. And Oz, of course, in case you missed that to the point we just made on financials being doing very well. Financials and commodities is the bread and butter of the ASX 200. And we did hit a fresh record early in the session. Very nice rally across the tech space, as you can see on your screen. Software dollar helps. And software dollar is one that we have currently, whether that's you looking at the Asia dollar index on the Bloomberg terminal or the MSCI gauge, both also showing up. This breakout in either, of course, which is Asian FX or EMFX as an overlap between the two. There we go. Very nicely on your screen. So pick your poison between between the two here.
[00:48:27] Speaker 2: Yep. And also watching these dim sum bonds, right? You've been highlighting how that sale is coming online. So exciting.
[00:48:35] Speaker 1: I'm very excited for this today.
[00:48:36] Speaker 2: Yeah. You've got appetite for that?
[00:48:38] Speaker 1: I do. In fact, in fact, for those in the Hong Kong office and the Bloomberg today, we really had a very nice spread for breakfast.
[00:48:46] Speaker 2: I missed it.
[00:48:47] Speaker 1: I missed the breakfast. You did. You were very busy. I tried to get Min Min involved, but she wasn't having any of the pumpkin congee that was served at the dim sum room today. But very nice. You had radish cake.
[00:48:56] Speaker 2: Is that what I was smelling? Okay. That makes sense now.
[00:48:59] Speaker 1: Yes. Yes. Unfortunately. Well, you have another go because we get, of course, that offer in about this hour. Varied amounts for varied tenors, including an onshore, by the way, which is a 73 billion on your screen. 30-year bonds. That's on deck today. Okay. Should we talk about oil? Or bonds, maybe?
[00:49:16] Speaker 2: Or everything else? We can continue talking about bonds. Yeah. Treasuries and everything else in the mix. Yeah. David Savage joins us now from the MLive team, of course. So, David, what do you see dim sum aside in Treasury, specifically? I mean, we came into this week. There was a bit of softness in terms of Treasury auctions on the short end. Next week, we have longer-dated tenors that are up for sale. But then, you know, oil aside, I guess investors' concerns about Kevin Walsh. That's kind of been, you know, sidelined in a way. Yeah.
[00:49:50] Speaker 10: I think it's been interesting to see that, you know, the reaction in the bond market has been, you know, somewhat more muted to the declines in oil prices. It's obviously helped, particularly at the long end. But I think if you look at bond volatility, it's still very elevated. It's nowhere near sort of lows we saw about a month ago. And I think that shows bond investors, A, are still, you know, perhaps more skeptical than equity investors on, you know, whether we see a normalization of energy flows through the straight-up moves. But also, I think it points to the fact that there's other drivers for higher bond yields. You know, you noted, you know, some issues in terms of credibility at the Fed. But also, I think it's just this AI/CAPEX story, I think, continues to drive on the bond side. Bond yields higher. It's, you know, as AI/CAPEX spending continues to rise, it insulates economic growth. But it's also going to keep pushing up yields. So I think that that's going to continue to push up bond yields, even in the absence of another spike in oil prices. But I think that's definitely a risk.
[00:50:50] Speaker 1: I mean, just to mention, I'm glad you brought up oil. We're just getting a news headline coming throughout of Axios right now that the U.S. is nearing a deal, Hormuz deal to be more specific and sort of more accurate, aiming for a Wednesday announcement. I think we are seeing a slight downtick in oil prices as we speak. And I think the equity market is responding. If it is, of course, this headline, it's responding too.
[00:51:14] Speaker 2: Brent's sitting, you know, nicely below the 80 handles. Yes, yes, it is.
[00:51:18] Speaker 1: And equity markets. Yeah, we are seeing a little bit of an uptick there. That brings me to the second point here. Tyx, Kospi, the Nikkei. Every time Chinese markets get some momentum going, we get reminded of the old school winners, like the ones I just mentioned. How do we look at that? Are the old winners still the new winners right now?
[00:51:38] Speaker 10: Yeah, it's definitely been the story, certainly the last sort of four sessions. We've seen that the Sox index post its best four days since March 2020. So it's definitely been sort of a reversal of what we saw basically through July, where we did see, you know, rotation out of those first half winners. I do think, you know, the Chinese competition story, I think that's going to continue to sort of plague and complicate sort of the, you know, the infrastructure rally that was so broad based across, you know, Asia's big equity benchmarks in the first half. So I think, you know, the real risk of commoditization, both at the AI intelligence layer and the infrastructure layer, I think that's going to continue to be, you know, a narrative that's going to continue to complicate the rally. So, yes, we're seeing some good gains today. But I think, you know, another part, you know, where we're seeing the weakness in the China X, I think that underscores that this is going to be a more complicated trade in the second half than it was in the first half.
[00:52:37] Speaker 1: David Savage, thank you so much. Our end life strategist there. I think just to add to the complication that David just alluded to there, of course, the optical, not illusion, it's certainly something tangible that we're trading on today. And the news coming out of Reuters that the White House is reportedly planning to ban some of the components, Chinese made components, whether that's Chinese made Chinese company made or is it Chinese or is it origin based? We still don't know, of course, the details on that, but effectively looking to draft rules to ban such going into U.S. data centers. That's according to Reuters. Min, Min, lower China correspondent is here to talk us through what specific components will be banned and how do we define which companies will be affected?
[00:53:16] Speaker 3: Yeah, it looks like the list of things that are getting banned is just going longer and longer, right? Previously, we were talking about robots and then it was power inverters. Before that, there were routers that were being banned as well. And now it's these optical fibers, the optical modules that are being used in data centers. And again, the pretext is that there is national security risk if malware are being installed, and then they could steal American data and disrupt these data center services. So in terms of companies, of course, don't be in a light is the number one, you know, the absolute biggest market share holder right now. And you to link is the number two. And together, all of these big Chinese players make up more than half of the global market share. So again, it depends on whether the house. It depends on how strictly that law is going to be implemented, right? If it bans all products coming from these companies, then the impact will be huge. But if it only bans products coming out of China geographically, many of these companies are prepared. They have already in the last few years started shifting production out of China to Southeast Asia, for example. So that could blunt some of that impact there.
[00:54:25] Speaker 2: And the justification for this reported ban, it seems to be tied to a salt typhoon hacking campaign several years back.
[00:54:32] Speaker 3: Yes, that was back in 2024, where U.S. intelligence officials were accusing Chinese actors of hacking into the phones or phone records of certain high profile politicians, including Donald Trump, who was the vice presidential candidate at the time. And this report from the House Panel Committee is expected to show that the U.S. telcos were compromised because of their connection to data centers that use some sort of Chinese components. This is according to what sources are telling us. And remember, a few years ago, there had been a lot of scrutiny into U.S. telcos. They banned Huawei. They banned Chinese components. But the links to the data centers were not so scrutinized. And now that is under the spotlight right now.
[00:55:14] Speaker 2: Min Min, thank you so much. Our Bloomberg China correspondent, Min Min Lo. And we continue this conversation on the AI stack. After the break, Rhodium Group's Riva Gujon joins us to break down their tech report card. As the U.S.-China AI race heats up. That's next. This is Bloomberg. All right. As promised, we're continuing that conversation on the AI stack. And we have the likes of Rhodium Group also. This new report dissecting the AI competition and where China stands compared with the U.S. and other G7 countries. They think Beijing is competing on a different axis with a focus on AI deployment as it becomes a default supplier of physical AI. And China's chokehold on critical raw materials. Well, this also means that it has the ability to throttle the tech ambitions of the U.S. and its partners.
[00:56:23] Speaker 1: Yeah. Looking at some of the numbers, the quantitative analysis on the back of the qualitative, and as you can see the scorecard on your screens, it also says that the U.S. cannot, is unable to match China's scale alone. So take a look at this. They're higher the number, the greater the dominance. Four being the perfect score, zero being you're completely dependent on somebody else. In their own assessment, China leads the U.S. in sectors like critical raw materials, energy, and AI models. But where it does fall short is in chips. But as you can see, that is in itself also very, very close. Let's bring in the co-author of that report, Reva Gujon, director at the Rhodium Group, joining us now. Reva, good morning from the Asia-Pacific. Thank you. Fantastic set of analysis and report and numbers here. Well, we took you through the findings. We took our viewers through the findings. Layat, what's important in your view here? Now we have the numbers. Yeah, exactly.
[00:57:18] Speaker 11: I mean, there are a lot of definitions of the tech stack out there. You've got Jensen's five-layer cake. The USAI action plan focuses on chips, models, and apps. But I would argue that's too myopic of a view. And we really need to assume a more holistic understanding of the tech stack. So we're not just flattering areas of existing strengths, but also taking a hard look at where China's playing to its asymmetric strengths. So you can see kind of where that baseline falls out today. And no surprise, of course, where China has some key advantages on critical inputs. It's pretty much at parity at the model layer with open weight models competing with Frontier Labs in the U.S. But some overlooked areas as well, connectivity infrastructure where AI and 6G come together. That's where China's obviously building on a big telecom footprint, and it plans to build on that, particularly in the global south. And then, of course, it has a very powerful manufacturing base to position itself to dominate in physical AI applications when we're looking at advanced robotics and other applications. And so this is, you know, a way to just break this down of where we stand today, but also to look into the future a bit, right? Because if we look into the 2030s and we understand how these AI systems are going to operate, there's some inherent security vulnerabilities there that's going to drive home the question of what constitutes trustworthiness in AI systems, who qualifies for that, who gets to feed into defense industrial-based structures, you know, where that's really going to matter. And that's where we come back to this broader argument on what does it take to achieve economic scale that can compete with China, both on the supply side and production capacity, as well as on the demand side in terms of the total addressable market.
[00:59:10] Speaker 2: Riva, I'm glad you raised the security angle. I mean, we've been trying to understand as well when it comes to the likes of open AI anthropic, these recent cyber failings. I mean, the extent in which that also maybe makes more of a case for perhaps the Chinese AI models that have been coming out. Do they have a better hold on security in a way?
[00:59:35] Speaker 11: Well, so we are in this phase of highly powerful models now, not just in the cyber domain, right? But we could see force multiplying effects in bio and other threat capabilities. And so that's a forcing function on both sides, both in Beijing and Washington in figuring out how to delineate safeguards, where to draw the line on where you gate frontier models. This is all very actively under debate. But again, if you look into the 2030s, we're talking about the development of hyperconnected AI systems, right? In the 6G era. So sensors and edge devices and models that are all wired into critical infrastructure. And so what that means is, you know, the surface area for any kind of adversarial attack vastly expands. And so cyber and data exposure become an inherent feature of the stack. And that's important, right? Because again, then who are you going to trust to feed into those supplier networks? Where that in some ways, too, restrictions carve out a market opportunity. And I think that's where you're starting to see some U.S. actions lean. Just earlier in the segment, you were talking about the FCC action on optical transceivers going into data centers. But there's and then robotics, drones, all kinds of different targets. That's part of this picture.
[01:01:01] Speaker 1: Yeah, well, I was going to get to whether the policymakers are focused, I would say not on the right, not on the right thing, but too limited in the scope that they're trying to, you know, create some distance between hardware and the supply chain that goes into that. So I guess my question is, does it seem like there is a coherent strategy on the part of U.S. policymakers? And if there is one, should that include others or can the U.S. go at it alone?
[01:01:33] Speaker 11: Coherent is not the word I would use here. It's kind of a mad dash effort, right? When you see on the critical raw materials side where to diversify as fast as you can, where China holds very significant leverage, of course, still trying to build out the energy infrastructure, but still a lot of constraints there. And then, of course, you know, as we talked about, there are other areas of the stack that need focus. So this needs a more holistic approach. The U.S. cannot go at it alone. It does have a lot of strength at key layers of the stack. But you look at semiconductors, obviously, that's not a U.S. go it alone strategy. Those are very complex global supply chains that involve key tech partners. But again, when we look at, you know, connected devices, what's the manufacturing base you need to meaningfully diversify away from China? Obviously, the U.S. cannot do that alone. And then again, in total demand as well. You need key markets. That's where markets like India play a role. And this is part of that coalition effort where, you know, it's there for the taking. If the U.S. applied a more coherent strategy, we're not there today. Certainly, it's a pretty politically fractious climate still. But this is what's at stake as we head into the 2030s.
[01:02:47] Speaker 2: Reva, as you alluded to, it is complicated among the G7s, the sort of cooperation in AI. Do you expect more divergence then within the G7, you know, the allies there? If not, then how do you think they can compete with China in terms of scale?
[01:03:09] Speaker 11: Yeah. So on the one hand, there is common ground and the urgency to diversify, particularly in critical inputs. There's common ground in recognizing that there are very credible cyber and data vulnerabilities that need to be addressed. Now, the tools and the timeline to do that, that's what's varying considerably. And you don't have the style of U.S. leadership, right, to kind of steer everyone along and helping define what are those economic security standards with the kind of political goodwill you would need to create that coalition. So on the one hand, you're looking for de facto convergence, right, where different countries are doing things and swimming in the same direction, but they're doing that more or less independently, which still has important effects collectively at the end of the day. But you also have tech sovereignty that is a very strong current in this matrix. And so as each country, each block is looking at what can they do as they look at each layer of the stack, where they can lean in and build a more credible tech sovereignty play, but also where critical dependencies need to be accepted within that coalition and where redundancy and waste and wasted time also can compromise the objective of reaching commercial scale. And that's where I think it's really important to get companies understand this very well. And tying an economic security agenda to a market scale argument is going to be an important feature moving forward. We'll see if we get there.
[01:04:47] Speaker 1: And just in terms of picking your battles, what specific fronts are worth fighting for. We know, of course, China leads in some cases, in most cases because of just years of industrial policies. What I'm trying to get to is have, you know, has the boat, you know, just gone too far from the dock at this point in time on certain fronts? Where would you focus the attention if you were to advise, of course, policymakers?
[01:05:09] Speaker 11: Yeah, well, I mean, there's on the one hand, there's the defense of existing strengths, right? So obviously where the U.S. and its partners have strengths and in the lead and in semiconductors, for example, in total aggregate compute, there are still gaps there on, you know, so there are no restrictions, for example, on remote access to high performance compute. There are still some outstanding measures there in terms of the semiconductor manufacturing equipment controls and how far those could go. So there's sort of a foundational layer there still that applies. But then again, looking forward right now, the big question is around potential model restrictions. But it's a very thorny debate because when you look at the scoreboard and that model leaderboard in particular, Chinese open weight models are clearly competing at the frontier. And so you can't really go for decisively in restrictions before you have viable alternatives in the open market. And so that's part of the dilemma here that's feeding into the debate. But in device ecosystems, critical inputs, you can't solve for one problem, for example, on a critical mineral that are refined material on the one end of supply chain, but then create bigger dependencies downstream for an intermediate component or at the device level itself. So that's that's really where that holistic approach comes into play and understanding how this all fits together, where different countries are positioning themselves who actually have a lot of political will and want to position themselves in this tech stack competition. Have that be part of these trade arrangements that it's striking with with the U.S. But it it it's going to take moving beyond the ongoing leverage plays the ongoing trade frictions to to really realize the opportunity at hand and also again what's what's at stake.
[01:07:07] Speaker 2: So again, what's what's what's at stake. Reva, it's been great to have you on the show. Thank you so much for this really nuanced look on the A.I. race between the U.S. and China. Reva Goujeon is director at Rodium Group. This is Bloomberg. All right. We're set to get earnings from Cathay Pacific later. The carrier expected to report a surge in profit for the first half of this year. Bloomberg's Asia Aviation and Transport reporter Danny Lee joins us now. We're going to be a So, Danny, we are expecting that jump. What do you think the drivers are? How much momentum is there for the second half? Well, it'll be really interesting to see the breakdown. I mean, Cathay has already flagged. It's going to see lower losses from
[01:08:03] Speaker 12: subsidiaries and associates. In particular, Hong Kong Express, its budget subsidiary, lost half a billion dollars last year. So it's flagging an improvement in its performance, although still like to be loss making. And then from the likes of associates like China had lower share of losses or may have made a contribution, at least in the first half. So there is improvement there. And also that bigger jump is also accentuated by this 1.4 billion gain on its China state, which was diluted. Overall, though, it should see a strong performance because of the 17.4 percent jump in passenger numbers at its main brand. Now, a real boost given what has been going on for all the disruption through the Middle East and even though we've had really surging fuel costs. Yeah. How has it been weathering those fuel costs? And given how there's been a lack of capacity increases among some of its rivals, what are you seeing going forward? Well, Cathay has been very quick to put up fuel surcharges. That's been a real pain point in a sense for passengers who normally travel. However, Cathay's been putting up surcharges significantly, but it hasn't seen, according to analysts, at least, much disruption in terms of weakening demand. So passengers are willing to pay. And there has been strong demand across the board because passengers do not want to necessarily travel through the Middle East. So it has seen a huge flow on effect. And I think it's just overall, given it is, you know, hedged, you know, for its fuel costs, just how much it has been able to weather some of those fuel costs cost jumps overall. Clearly, we saw in recent weeks, Singapore Airlines has had a bit of a hit as well from the war. So overall, you know, Cathay's in a much better position compared to some of its peers in Asia
[01:09:45] Speaker 2: and in Europe. Yeah, I mean, we've been seeing some of these carriers in Asia vindicated for the hedging, right, given what we saw out of the Middle East and the impact that Danny, great to have you. Thank you. Bloomberg's Asia aviation and transport reporter Danny Lee. And of course, we are seeing that tech rally in the region today. Very clear. The outstanding one from Kospi. This is Bloomberg. It's 11:29 a.m. in Tokyo. Japanese markets going to break in a minute. Of course, we've been watching the yen. It's kind of flat today. I mean, if there was intervention, I mean, as we know from the most available data following Thursday, Friday, it looks like, you know, it looks like they spent on something big. Yeah, right. Looks like they didn't do it on Monday, though. From the BOJ data. So that might have been more from Elgos or, you know, just the investor reaction where we saw the sharp moves in the yen on Monday. Of course, we only find out later. So we're watching still, you know, the idea that the BOJ needs to come out with a messaging maybe a bit more forcefully. Otherwise, you get the likes of the 10 year auction just a day ago out of Japan. Terrible auction. Weak demand. Super weak demand. Right. And I think not benefiting from the lack of clarity from the BOJ. How serious are they, you know, when it comes to that potential hike. I think markets are pricing in October fully. Maybe we see September. There seems to be a coin toss for now. But aside from that, of course, we've been watching as well what we're seeing in Japan market stocks specifically given the bounds we're seeing in SoftBank as we watch out for earnings. That is, you know, really the big big gainer today. And this is coming with the tailwind following on from SpaceX AMD. I think good vibes for the region overall. Really good vibes for the for the capital expand. Right. Yeah. And the
[01:11:56] Speaker 1: the supply chain for Asia. And I think that also goes into the good vibes regarding on the macro front with a softer dollar. It was already, by the way, softer oil. And then on top of that, I think in the last 30 minutes, this news headline perhaps is what's pushing oil down from the session highs right now. As you can see, we're reversing that to the downside, which is really helping Asia big tech already apart from the other reasons that we just laid out there. Just to give you an indication of so on a benchmark basis up up three up three up three. That's Japan, Korea and Taiwan. You want to put some names on that. You're coming up on your screens as well very shortly. So you had the likes of TSMC. Samsung is doing very well. Esky Hynix. Can we change the page please on top of that. And of course the optical space in China is also doing well. We'll get to that. But yes. Are we doing optical. It does seem like we're doing optical. Can we do optical first. Then we'll talk about this. So the optical names are actually coming off low. Thank you. There we go. Actually, no, I stand corrected. We're not actually back almost at session lows. We were trading at slightly higher levels than currently. But yes, certainly this royalties report is playing out. Markets are reacting first. And then I guess we'll try and figure out the difference later. Back to Asia big tech, which is the outperformer today. I won't spend too much time on that. We've showed you that already. But yes, the winners so far are again the winners today like Samsung. Yeah. And this is following on from its unveiling that AI
[01:13:18] Speaker 2: plan. So Samsung is in focus. It touted improved performance power efficiency and its most advanced memory hardware. And the system is aimed at overtaking its rivals SK Hynix Micron technology. Bloomberg's Asia tech editor Mayumi Negishi joins us now from Tokyo. So Mayumi the unveiling of this 3D roadmap. What does this tell us about Samsung shots at taking on Micron and SK Hynix. Right. So Samsung basically threw down the gauntlet to become the de facto
[01:13:56] Speaker 13: standard for the next. For future generations in high bandwidth memory for AI. So to be clear and all of this is still up in the air. We don't know which way the technology will go. But Samsung announced that they are there. Their next generation will go for sort of vertically vertically stacked HBM. They're calling it the HBM. And they're saying that in two generations time this will increase output and performance by multiples. So so it's you know they're the race is on to become the de facto standard for future generations. It's it's you know to be clear. We're we're still at HBM for and right now Samsung is saying this Z HBM will be like eight times ten times more powerful than HBM five which it has still to be announced and there's no timeline for this. All these memory companies. All these memory companies. Samson SK Hynix. Micron Keoxia. They're now scrambling to get to become to dictate which direction the technology is going to go. It's important to understand that if you if any of these companies choose the wrong processes or the wrong technologies that could throw them off in the race. Right. With the current HBM generation SK Hynix got got a two generation lead precisely because Samsung chose the wrong technological processes that wrong manufacturing processes and yielding that giving SK Hynix years of a lead in an edge with NVIDIA. So choosing the technology. So promoting their technology winning NVIDIA is backing for your technology is extremely crucial in the year or two ahead.
[01:16:11] Speaker 1: That's a good point. I mean is is first mover still an advantage given current dynamics in this market where also you know things are moving very quickly. And you know on on one side of the supply chain you have you know the top players asking their supply chain to create new products. So how do we need to be looking at you know whether or not you are in fact in an advantageous position to come up with the newest technology. Now this is interesting because NVIDIA is boasting that they actually hold the important levers that they can change the architecture to sort of change the reliance of AI systems on different different parts of memory to different kinds of memory that they could say OK well there's a shortage of HBM. The HBM processes are too expensive.
[01:16:40] Speaker 13: So we're going to change the architecture so we're going to change the architecture so that AI systems are going to change the architecture so that AI systems are more reliant on NAND or SRAM or SRAM or SRAM or SRAM or SRAM or SRAM or SRAM or SRAM. So we're going to change the architecture so that AI systems are more reliant on NAND or SRAM or SRAM. So there is a lot of these factors are moving you know are in flux all the time. And as you said everything is moving so quickly that it's it's it's it's it's like all these companies are jockeying for position right now and lobbying NVIDIA and AMD and saying they're technology is going to change the architecture so that AI systems are more reliant on NAND or SRAM. So that there is a lot a lot of these factors are moving you know are in flux all the time and as you said everything is moving so quickly that it's it's it's it's like all these companies are jockeying for position right now and lobbying NVIDIA and AMD and saying their technology is better their technology is more energy efficient. Their technology can pack more memory and data into AI systems as inference demands more and more memory and for AI to perform.
[01:17:46] Speaker 1: Well you mean the Gishi there thank you so much our Asia tech editor on the what we understand so far about perhaps the future for a as it pertains to memory. We have lots more head here on the program. It's a big day in India today so we'll get into that in a moment. This is Bloomberg.
[01:18:25] Speaker 2: Right also looking at Japanese pharmaceutical manufacturer Shionogi how it's considering manufacturing some of its capacity you know in the U.S. over the next few years as it looks to try and diversify its production base. CEO and President Isao Tashirogi also told Bloomberg's Lisa do exclusively that Shionogi is pursuing a series of M&A deals.
[01:18:48] Speaker 14: We are looking for a lot of opportunities actually we're actively pursuing minimum three deals today. And of course the you know we're not able to guarantee you we're going to go although three but the this is something that we continuously will do. And of course the you know Japanese market is very important and U.S. market is very important too. And we are going to keep growing the European market as well. So those are the three major geographies we're looking into.
[01:19:30] Speaker 15: Chinese competition obviously going back to this it's you know it's a hot topic globally. Some pharmaceutical companies have mentioned there being more cautious about disclosing you know certain clinical trial results early because of Chinese farm abilities to kind of kind of replicate the results or make very similar drugs in a short amount of time. Is this something that is on your radar and that Shionogi is similarly cautious about.
[01:19:53] Speaker 14: I think a small molecule area even historically you know was exposed to the you know the things you described. So when we're submitting the intellectual property we have been very cautious and we have been you know aiming to us the thoroughly as possible to create a good IP. So yes we recognize but I don't think it's only unique for China. The pharmaceutical companies are so competitive and everybody is holding a lot of radar screens for the innovation. So it's a competitive competition we admit that. So having said that how we maintain our strength the only thing is we keep running into the top notch science in terms of the how we're going to differentiate ourselves from the other companies.
[01:20:55] Speaker 15: Okay so the strategy around hasn't really changed because it's always been quite competitive when it comes to small molecules. Right. Well I think I believe you became CEO and president of Shionogi in 2008. Right. So you've been running the company now for almost two decades. Yeah. How are you thinking about kind of the next generation of leadership at the company.
[01:21:11] Speaker 14: Well this is one of the you know most the you know important things for our board. Again in my company's board members almost outside members so they have the course to be concerned for the future leadership team of Shionogi. So we are you know creating the team both the you know outside board members and the we picked up the four or five the younger generation who they believe can be a possibility of my successor. So since last year we're very active in the you know training all those folks but at the same time we acquired the you know U.S. company. The relationship with VIV is getting stronger and stronger so that takes time to heritage what we have been doing to new leadership team. Right. So it takes probably three to four years to almost create the good candidate team to succeed in the current Shionogi you know the business model. We just nominated Dr. John Keller the first known Japanese board member. John and I are almost teaming up to take care of various things including the relationship with the VIV, GSK, U.S. companies. So we try to convey all those experiences and networks to a younger generation in the next three to four years.
[01:22:50] Speaker 15: There is a lot of rhetoric now globally I mean around the world about geopolitical tension, trade barriers. Just last month U.S. President Trump also had mentioned that he intends to put a big tariff on generics going into the U.S. I understand Shionogi doesn't have a big focus on generics but you guys are also are very you know focused on your message about access to health care and medicine. I'm just curious how what you think about this type of kind of tense rhetoric and how it might impact the future of the pharmaceutical industry globally.
[01:23:22] Speaker 14: Well good question. The pharmaceutical manufacturing as you know is heavily regulated by the each country's health authority. FDA, EMA, PMDA, MHLW. So it's not easy to move the manufacturing site from country A to country B or place A to place B. Having all said that the probably five to ten years landscape our company's way of thinking is the we really want to almost diversify the manufacturing capabilities Japan, U.S. and part of Europe.
[01:24:06] Speaker 1: Okay. I have a feeling this is a headline we'll be talking a lot more in the next couple of hours or so. Reuters reporting Samsung and SK Heinex are testing Amex chip tools at its well at factories in China. We're looking at this specific stock we're up 11 percent on the back of that 12 percent. This is of course a listing on the Chinese mainland. We'll keep an eye on this and how this obviously affects the supply chain conversation ever evolving supply chain conversation. But yes we are trying to figure out more details on this market is I it's fair to say there is some some reaction across markets today on the back of this. But the stock was already up going into the last couple of minutes.
[01:25:00] Speaker 2: Yeah. I think I cross currents is a fair way to put it perfectly given how we've also been seeing it's interesting foreigners are still selling South Korean stocks. Right. But we have been seeing that rebound steadily on the Korean one. Let's take a look at the E.M. FX complex as well. Given how these lower oil prices brand below 80. That's something that's helpful. Perhaps another tailwind. Maybe they break out to the upside as well. We're watching some of these very energy import reliant economies including India. We have an RBI decision due today not expected to do anything. expected to keep the repo rate at five point two five percent for an analysis of bringing Pranju Bhandari chief India economist and strategist at HSBC. So Pranju set to keep things on hold the RBI today but maybe some tightening in the later parts of the year. What's the thesis. What do you think needs to be done here.
[01:25:55] Speaker 16: Yeah. Absolutely. You know just to take a step back. I think India is in a sweet spot at this point of time. Growth has been holding up pretty well. External finances have really improved with the FX package that the central bank had announced. And inflation hasn't really taken off. So all of this means that the RBI can sit and do nothing in this policy meeting. policy meeting and you're expecting no change in policy rates. But there are two more policy meetings coming up later in the calendar year in October and December. And our senses that inflation would have risen enough by then to call for a rate hike both in the October and the December meeting. Overall recording for two rate hikes in this RBI rate hiking cycle. Nothing today but in October and December. Right. And how does that then feature into your growth forecast.
[01:26:45] Speaker 1: Perhaps even for next year because I would imagine any hikes this year would show up at the very early at the earliest next year. I think there are
[01:26:55] Speaker 16: just too many things going on. I think what we have right now is that growth has been remarkably resilient in India. You know on the back of a lot of the policy stimulus in the system last year like GST tax cuts which are still holding up consumption pretty well. And then in the heart of the energy crisis we found producers actually produce a lot more rather than cutting down on production because there was a lot of front loading of production. There was a rise in exports to the U.S. because tariffs levied on India had fallen dramatically from 50 percent to 10 percent. So all of that has held growth pretty well. And I think that's going to be the story of most of this calendar year. But come early 2027 I think India needs to grapple with the El Nino which is expected to strengthen around October December. And the impact of that I think would be felt in early 2027. But you know we're not there right now.
[01:27:48] Speaker 2: We're in a place right now where growth is extremely strong. OK. So growth strong. We've also been seeing how the rupee has been relatively supported. There's been that bit of intervention as well. What are you expecting in terms of how sustainable that all is. Are you expecting more intervention to come as well. Well the RBI announced a very big package
[01:28:15] Speaker 16: on 5th of June trying to attract a lot of money from NRI is the non-resident Indians. You know putting it back to India. And I think the scheme did very well. India has been able to get in 40 billion dollars which is quite a lot. And today the RBI is war chest of reserves is strong enough for it to intervene quite a lot. And it has been intervening in markets for the last 10 to 15 days in a fairly aggressive way. And my sense is it has firepower to continue that. So from that perspective I think our outlook on the iron hour is fairly positive at this point of time. Of course a lot depends on oil prices. A lot depends on what happens to the U.S. dollar. But the RBI has a lot more reserves to use right now than it had see three months ago. I was going to ask you that. What do you think they will be using that money.
[01:29:05] Speaker 1: I mean I mean they could they could do two things. They could retire. I mean they could accumulate those reserves or they could just manage some of their near term liabilities. Which one do you think they'll. And it's a good problem to have obviously these these these these inflows financial. Well look
[01:29:21] Speaker 16: you know we'll have to see how things turn out. But my sense is that at this point of time the central bank would want to see an iron hour which is stable or perhaps outperforming the region you know which is what has been happening since the fifth of June. And I think it will use reserves if needed to you know to create that environment for the iron hour. So for now I think the mode is going to be a little more interventionist.
[01:29:46] Speaker 2: Perhaps later on you know it can take a step back. You talked about how growth looks quite robust so far. But what are the risks to that aside from maybe if we look at El Nino further out into the impact in 2027 because there are some of these structural ball necks for the Indian economy as well including concerns about job creation. Well yes you know I think job creation is an issue globally right now.
[01:30:14] Speaker 16: But in terms of the main risks that India is facing there are a couple. I think number one is oil prices as when it goes closer to 90 dollars a barrel. I think it starts hurting the economy. 80 dollars is still fine. But as soon as it's closer to 90 I think that hurts. So that's one big risk. The second is how strong will the El Nino become towards the year end. I think that will matter a lot for agricultural production which makes up about 20 percent of India's GDP. And then of course there is what happens globally to the U.S. dollar or to U.S. rates. I think a lot of people will also be watching the interest rate differential between India and the U.S. So if you know U.S. yields U.S. rates were to rise then you know how would that look from the effects front is also sort of an important risk that has to be monitored. But I also agree with you that it's not just about all of these short term things. There's also a lot of reforms that the country is doing. I think the remarkable reform that I have seen in the last one year has been the speed with which many trade deals have been signed you know with UK with EU and a couple of other big countries. And my sense is once these get activated we could see a lot more FDI inflows into the country and exports. So I would say policymakers really need to focus on getting these trade deals implemented as quickly and as efficiently as possible. I think that would be good for growth and flows
[01:31:35] Speaker 2: into the country. All right. We'll watch whether that unfolds. Thank you so much Pranju Pandaria's chief India economist and strategist at HSBC. Now let's also take a closer look at some of these lines across the terminal a short while ago against the backdrop of what seems like progress among some of these chip tool makers in China. So AMEC seems to have been testing some of these tools for China plants at Samsung and SK high name. So that is according to a report by Reuters and coming at a time I suppose where China seems to have been making a fair bit of headway on deep ultraviolet lithography. Yeah. So you know maybe that speaks to the point that our analyst was talking about earlier the the A.I. race and the headway that China is making. Yeah I mean this is a suddenly we're trying to figure out most
[01:32:31] Speaker 1: people are trying to figure out of course the company. It's a Shanghai based of course makes micro fabrication equipment. Hence of course AMEC the sea being of course China. Lots to watch out for today. What else are we tracking. I think I was looking at something else on my telecomputer. But anyway I'll read what's on it. That's what our job is. OK. The Hong Kong dollar. We're very close. We're as close as we have been recently to the weekend of the of the trading ban to watch it closely whether we get there. Of course we cannot cross that of course with the peg. Brent is down just to mention that too. What else to watch today. Cathay earnings. We have a CGB auction 30 year and updates in the weather. Lots more ahead on Bloomberg television. We will see you all tomorrow on the China show.