Live from Sydney, I'm Paul Allen. This is Insights, where we dive deeper into the stories that matter with crucial context and sharp analysis.
Asian stocks slide. Treasuries remain under pressure, with the US ten-year yields near their highest since 2007 as elevated crude prices fuel inflation worries. Asia's chip exporters are reaping the benefits of booming AI demand. JP Morgan warns the so-called Dutch disease risks have not disappeared. The head of Asia Economics joins us on that.
OpenAI accuses Chinese rival Moonshot AI of a large-scale effort to extract data from GPT systems, intensifying tensions between the US and China over tech leadership.
Plus, India and the United States reaffirm strategic ties, with Modi and Trump focusing on trade and investment in a phone call as new tariffs loom.
Asian stocks are mostly rising at the start of a new month, with Micron earnings refueling parts of the AI trade. Haverhill Hong is here with a check on markets overall.
Yeah, you're really seeing AI leading the pack today, even though the gauge for the region is sitting a bit softer given those elevated energy prices along with yields. Don't forget, in the US as well, we saw a late session slide and some of those quarter-end quirks.
But it's interesting today, given Micron's beat and raise, we are seeing that AI trade refueled, particularly in Japan, South Korea memory names. It was a bit of an underperformance at the start, early goings, and some attributed this to some of those disappointments related to shareholder returns. But even that is pulling ahead now in terms of Japan specifically, maybe some support also from the Tankan data, which shows that the economy in Japan is getting stronger, but not so hot that it will warrant perhaps a move into aggressive rate hikes.
And that's part of why you're seeing dollar-yen also popping above the 158 handle for Japan stocks. Related names, specifically, we had the likes of Jefferies coming out with a note that they see potential recovery in valuations of some of these semiconductor chip-related equipment makers because of more visibility over market growth into 2027.
So the board, because when it comes to Micron, not only did it beat on earnings, it also came through with a very bullish forecast. And the stock, as you know, has been the top performer in US chips so far this year. So the question really for investors was how much better can it get.
As one analyst put it, this is against the backdrop of those supply-constrained concerns, and analysts a bit worried that this would ease in 2028. But what was interesting was how management soothed those fears. They talked about how demand would increase more so than that supply, that new capacity that is set to come online. We also saw them securing some of those new long-term contracts. So there's considerable revenue visibility, and I think that's helping to offset some of those concerns.
The warnings about profit margins being weighed down, given the increase in compensation that Micron is set to be offering as well. Paul.
All right. Thanks very much. Joining us now is Vikas, special Asian equities portfolio manager at M&G Investments. Vikas, thank you so much for joining us today. And just picking up on April's theme there, I mean, great numbers from Micron. You know what's coming next. Is it fair to say the AI trade's maturing somewhat there? And if so, where are you seeing opportunities right now?
Well, thanks, Paul. This is the right place to start. And it's a trade maturing, perhaps in some parts of it, in memory. What's interesting is the share price reaction that we have seen. The hurdle is quite high now for memory companies on these beats and these raises. I don't think it's a surprise that the numbers continue to be strong heading into 2027.
Soon we will be shifting towards 28, and that's when we will start seeing perhaps as a base case more downward pressure on non-HBM memory. So the non-AI server memory, the legacy DRAM as you would have it, partly because of the increasing capacity coming online in China from YMTC, but then also because of where we are and how elevated these prices are.
Also, what we've been doing more work on, and the market was quite focused on this a few months ago, I think attention will come back to this, is the equity-for-compute story and the circular financing, as some people call it in the industry, and how much capital has been committed relative to how much cash actually has been in the system. And then as that slows down or it unwinds, what the implications will be for the supply chain, and memory is a key part of that.
Do you see any evidence of that circular dealmaking unwinding?
We don't see any evidence yet. But what we're trying to estimate as a team is just how big are the numbers. And roughly our estimate is that it can be about 10 to 1. The commitments relative to the cash that has come into the system is, in some areas, it's a little bit more than that. That ratio, in some areas, it's a little bit less than that.
But there are some companies that will be less vulnerable to that. The semiconductor production equipment companies April was just highlighting in Japan, how the semiconductor companies are doing well. Our view is when you look at the surge in CapEx from the memory companies, the potential surge in CapEx from TSMC, the other sources of spending on equipment, those names win on a near-term view in every instance.
The pressure is probably on other parts of the supply chain, perhaps on the GPU accelerators, perhaps on the memory companies as well. Also, what we should think about is today we're seeing inflation everywhere. You were highlighting that earlier, or the team was highlighting that earlier in the China show, where we are not seeing inflation. We're seeing the opposite is in the unit cost of compute. The cost of computers is collapsing. That has implications for the frontier model companies in the US heading into their IPOs. That has implications up and down the supply chain all across Asia. And we should all be paying more attention to what's happening in China all across the supply chain as well.
Yeah, I want to get to the China question in a moment. But, you know, we talked about circular dealmaking. That's one risk which is often associated with this story. I mean, another is concentration risk. I mean, so much of the market now is either directly or indirectly linked to AI. Do you feel that that risk is potentially diminishing as well, as the concentration risk?
Well, it's still pretty concentrated for us. As 2026 has progressed, the setup for prospective returns in 27 and 28 and beyond has improved because the rally has been so narrow. We were just highlighting the strength in Micron share price year to date and how in the US that's the best performing chip stock. You look at the market cap accretion in a small handful of names out here in Asia this year, and that leaves a lot of alpha on the table for long-term active managers.
So for us, it's a good setup. If you look at many of our key strategies, the top holdings are not exposed to the AI story. It's a broad and deep universe that we have out here. Whether you look at India or China or Japan or South Korea, Taiwan, even or even Southeast Asia, there are a lot of opportunities out here. And it's not just all about AI.
Is there a bigger risk we need to consider as well? I mean, we heard from the Bridgewater CEO saying, look, AI might dislocate almost a fifth of the entire US workforce. BMW is looking to AI as perhaps eliminating 20% of all management jobs as well. So can you see other pressures starting to emerge elsewhere in the economy if we start getting a lot of unemployed people all of a sudden?
Yeah. Well, the thing about predictions is that they'll be wrong. But we need to think about these issues. We need to think probabilistically. And yes, we are seeing some of that impact. It's not just in the US. It's not just in China. It's across Asia, it's around the world.
But if you look at the IT services sector in India, for example, net hiring has slowed quite a bit over the last two to three years. The mix has shifted a little bit more. The high-end hiring of engineers, but net hiring has slowed down materially. We're seeing that in other sectors as well.
I was in Bangalore two months ago and I toured a lot of factories. I would have been on the same factory floors pre-COVID, and the biggest distinctions or the biggest differences for me were how few people there were on each of these floors. What was interesting was how high-end the machines had become. A lot of high-end Japanese machines, some German machines as well. So these companies are investing in R&D, they're investing in advanced manufacturing, but fewer and fewer people. At the same time, AI is coming for a lot of jobs. And that is true. We need to think about that.
Yeah. India's typically been a popular hedge against the AI trade and against AI concentration risk. It sounds like you think that that's about to change. Are the days of that trade numbered?
Well, that's hard to predict. But what I will say about India, and this is a good topic of conversation, if you look at the benchmark in India, it's very concentrated in a small handful of companies, many of whom are not being helped by AI, some of whom are being hurt by AI in the short to intermediate term and certainly in the long term.
But when you look at the AI supply chain, if you think about the $750 billion of CapEx from the top five companies in the US, you add on to that another quarter trillion dollars from other data center hyperscalers and other sovereigns around the world, you're looking at about $1 trillion of CapEx. A lot of that is making its way to India, whether it's in the form of optical fiber cable expenditures, transformer, transformer oil, the power sector, cooling. There are a lot of those companies in India. We own some of them. They have done well this year. The outlook still looks pretty good.
So to call it India, the AI and India story is a little bit misleading with the impression that some people have. It might be the anti-AI benchmark, but there are plenty of opportunities related to it. You can draw a direct line from the capital spending from the hyperscalers in the US to many companies in India, and that universe is growing.
Well, you did mention China a little bit earlier. And I know your largest allocation in Asia is to China. Is that a function of just the sheer size of China's market, or are you more bullish on the China story?
We are bullish on Chinese equities. It's not just the valuation. It's the quality of companies that we're seeing. On the valuation point, there are some companies in our portfolio that have a higher dividend yield than their PE ratio. We're not going to find that in India. You're not going to find that in Japan these days, at least not beyond a few one-off situations.
So China, the valuation is compelling, but a low multiple and high dividend yield does not necessarily signal cheapness of an equity. We do think Chinese equities are cheap. In aggregate, we think the portfolio companies that we have are also cheap because of the quality of the companies, the growth that we see, the margin expansion opportunities that we see. And again, just like across our broader strategies across Asia and even in Japan or dedicated to China and India, the top holdings are not related to AI. We do have exposure, but these are broad and deep markets.
Just to summarize, if you look at the Taiwanese benchmark, we would tag more than 75% of the benchmark as related to AI. In Japan, it's probably 20 to 23%, India less than 10. But that doesn't mean that there aren't AI-related opportunities available, but also that if you take one minus each of those numbers, it just signals how broad the opportunity set is beyond AI in the renewable energy space, in consumer retail, in financial services, in health care. All those are very exciting to us as well.
Because you mentioned there Japan as well. I know you like the pace of corporate governance reform in Japan, but is also part of the appeal, just the enduring appeal of the carry trade in the very weak yen at the moment?
And that is part of it. But no bet that we have made, no single active bet one way to underweight, is contingent on a particular level of the yen, just like it's not contingent on any particular level of interest rates in Japan or in the US or elsewhere. We are long-term bottom-up stock pickers, and we see a lot of opportunities to generate absolute returns and relative returns in that market, largely because of the corporate governance transformation underway, but not only because of that.
If you look at where bottlenecks are around the world, not just in the AI supply chain, but if you think more broadly when it comes to renewable energy, when it comes to water treatment, we're finding a lot of those companies in North Asia and in Japan in particular.
All right. Vikas Shad, Asian equities portfolio manager at M&G Investments. Thank you so much for your time today.
Well, OpenAI is publicly accusing Chinese AI startup Moonshot of a coordinated effort to pull proprietary data from its GPT systems. Now, this escalates a growing fight over how advanced AI models are built and protected. Bloomberg's Min Min joins us now with more on what OpenAI says it uncovered and why this matters for the AI race. So, Min Min, can you talk us through the scale of OpenAI's allegations against Moonshot?
Yes. Well, I'm going to start to sound like a broken record because we keep hearing these similar accusations from Silicon Valley companies, right? If this is going to sound familiar, it's because Anthropic has made a similar accusation as well. Now, OpenAI does it the first time, but is now accusing Moonshot AI of large-scale distillation. And here you can see Anthropic also released a study showing that Alibaba is one of the biggest distillers as well.
So OpenAI has found that in July, there were as many as 16,000 attempts to uncover hidden reasoning, or to uncover the hidden information on how OpenAI models approach reasoning through problems. And what OpenAI has done so far to stop distillation is that they have concealed their reasoning. When users enter a query, they get an answer instead of an explanation.
But according to OpenAI, how Moonshot has gone around this is that they have copied that encrypted information and then asked a model in a different conversation to decrypt and transcribe that data. So, of course, Moonshot AI has so far not provided any comments in relation to these accusations. But the Chinese government has, of course, in the past already refuted some of these accusations and said they would take countermeasures if the US were to release any punitive measure.
And this is really the crux of the matter here, because OpenAI has already communicated what they found to the government partners with other Silicon Valley companies as well. And if the US does take some sort of policy that would be punitive for these Chinese AI models, that will, of course, have tremendous implications on their ability to scale globally.
Also, the DeepSeek-Huawei software partnership. Can you tell us a bit more about that? Does it suggest China's becoming less dependent on American chip ecosystems, especially Nvidia's, for its AI development?
That's right. Already we know that Nvidia's advanced chips are not able to enter China right now, but Nvidia's software is still very dominant and very integrated into frontier AI labs, whether it's in the US or in China, to develop AI systems. And DeepSeek is now collaborating with Huawei and creating their own software system to help developers basically develop the AI chipset or program those AI chips.
This software system is going to be free for download, is going to be open source, and is going to be compatible with these Huawei Ascend chips. Now, one of the reasons why Nvidia's CUDA ecosystem has been so dominant is because there is a high migration cost, right? It's kind of like if you've been using Apple iOS, it's hard to switch over to Android, for example.
So now that DeepSeek has come up with this new system, it's going to change the playing field a little bit and make it easier for many companies to use Huawei's Ascend chips with this software stack from DeepSeek.
We're just getting some news as well from the Financial Times reporting that Tencent reached an agreement with Oracle on data center leasing. What more do we know about this?
Yes, this is the latest news. And Tencent reportedly making the largest leasing deal yet in recent years with Oracle. We're looking at 100,000 or accessing 100,000 at once AI chips from Nvidia through these leased data centers around Southeast Asia. So this is one of the loopholes, if you will, that allow Chinese players to access advanced Nvidia chips that cannot be imported into China.
And this is a deal totaling about $7 billion, of which there is a 30% upfront payment. And according to the Financial Times, this was what contributed to the very first negative quarterly cash flow that we have seen in over a decade for Tencent in the last quarter.
I believe China correspondent to mid Menlo. They're still to come. We're going to hear from Netflix co-CEO Ted Sarandos, the Bloomberg Screen Time home, the Warner Brothers bidding battle, and the company's growing bid on live events. This is Bloomberg.
Welcome back. You're watching Inside. Netflix co-CEO Ted Sarandos says the company's pursuit of Warner Brothers Discovery was a unique opportunity and not the start of a broader acquisition strategy. Speaking at Bloomberg Screen Time, Sarandos also discussed the company's search for new growth drivers, including live events.
We are growing engagement. So we're growing on 200 billion hours of watching. We grew 2%, you know, in our last announcement. About 2% is not what people are hoping for out of you right there. Used to double-digit growth in at least and certainly in revenue and to some extent even in viewership they had expected. This is my point about the growth in general. Yes, overall, we're not growing as fast as I want us to, and we're working on making that move faster.
We are, though, also doing things that create a lot of headwind to that number. Meaning when we do live programming on Netflix, which is a relatively new thing, we spend about 5% of our content budget on live events. They generate about 1% of our watching right now. They are, but they do a very different job. They generate a lot of signups. They're really effective for signup, retention, advertising, all those things that they do.
But it creates engagement headwind in how you invest against it. And remember when I say we grew 2%, that's an easy number to sneeze at. But it's through all the growth of live. It's through incredible headwinds from things like the World Cup and world sports and all those things that are going on too. So it is a we are growing the business. We want to keep growing it faster.
This past quarter, we did double-digit revenue growth in every region of the world. So the business is great and growing fine. I do, I mean, if you ask me when if we were growing at 20%, I would be telling you, I wish we were growing faster. Right?
So when you said you wish you were going faster and we're working on it, what are the things that you are working on to kind of re-accelerate that growth?
Well, it's these are some of the things are in the expansion of what we do. So like it was an expansion of what we did that didn't necessarily bring more growth engagement, but it brings very valuable engagement. So it's not a mystery that all engagement is not equal, because it's you probably are not surprised to know that an hour of Judge Judy in the middle of the day does not generate as much revenue as an hour of NFL football.
And so those are I saw some CPM chart recently that showed what you make per view versus what YouTube makes per view. And yours is higher. Yeah, we generate more. We monetize better. Our programming monetizes better. Now we do. We're looking constantly about how do we do things to bring more value to the members, more things to watch, more ways to watch all those things.
So that is some of these things that are and I don't want to over-characterize things like podcast and those things, because they're very small investment and very small additions to what we're doing. The best the vast majority of what we spend on is professionally made movies and television series and games.
But do you ever buy one of the big studios, Take-Two? Microsoft seems to be unhappy with that.
Well, you know, we're not we have not been traditionally big buyers. The Warner Brothers notwithstanding, that we've not really been we've been beam builders from the script, from scratch. So I do think we've looked at we have built some, worked with some of the smaller studios and programmers and developers, and there's like this of those things come the the big problem with with when you do these deals.
And the reason why I go back, why the Warner Brothers was attractive, even though I said we were not going to do that, is it's very rare that so that this kind of asset was so clean. It was just you're just buying just the things we wanted to buy. And so we were able to very easily look at this and say, how will this add value to Netflix and not destroy value in other ways?
How do you take this growth engine and have it supercharge our growth engine? And for us, the ability to do that made that deal super unique. We're not looking for the next opportunity to backfill that deal, because we are our plans. And how we're going to grow is primarily organic and will look for opportunities that complement the business as we go.
