Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Under the Banyan Tree - How is China using agentic AI?
0 Plays2 seconds ago
Herald van der Linde talks to Charlene Liu, Head of China Internet Research, about the next phase in the global AI boom and how it's being implemented and monetised in China.
Click here for appropriate Disclosures, including analyst certifications, and Disclaimers that must be viewed with this podcast: https://www.research.hsbc.com/R/101/h6Pmvxq
Stay connected and access free to view reports and videos from HSBC Global Investment Research follow us on LinkedIn https://www.linkedin.com/feed/hashtag/hsbcresearch/ or click here: https://www.gbm.hsbc.com/insights/global-research
Recommended
Transcript
US vs. China in AI Capabilities
00:00:00
Speaker
From a capability standpoint, I think U.S. agents and models are still capable of doing a little more than what China has to offer. But then when it comes to cost, China is extremely competitive.
00:00:21
Speaker
Hello and welcome to Under the Banyan Tree. I'm your host, Harold van der Linde, and this is the podcast where we put Asian markets and economics in context. On the show today, we're tracking China's progress in the global AI race.
00:00:36
Speaker
How is it being used? How does the business of AI differ between China and the US? And why actually have Chinese equity markets been a little bit subdued on this front in recent months? Joining me to answer those questions is head of China internet research, Charlene Liu.
00:00:51
Speaker
Plenty to go through, so let's get the conversation
AI Landscape Advances
00:00:54
Speaker
started. From HBC Global Investment Research in Hong Kong, this is Under the Banyan Tree.
What is Agentic AI?
00:01:11
Speaker
Let's start with where we are in the fast-moving AI landscape. The buzzword in 2026 is a gen tick. Now we'll come to what it means in a moment, but it's important to note that it's the next big step onwards from generative AI. That's your everyday AI chatbots responding to human prompts, which was the talk of the town just a short while ago, and which we pretty much now take for granted.
00:01:36
Speaker
Let's bring in our special guest, Charlene Liu from Singapore. Charlene, great to have you with us. Can you start by telling us how Agentic AI is the next step forward? Absolutely. Yes. Essentially what agentic AI is, is this very significant improvement beyond what we're used to, which is the traditional chat box. And these systems can initiate and complete tasks without human intervention. They can do things very independently. And that's why, you know, it is considered a major breakthrough.
00:02:08
Speaker
And then you have the models. That's the stuff that we're used to, right? Correct. So essentially what it is, is agentic AI is powered by the model that sits behind it.
US vs. China Approaches to Agentic AI
00:02:18
Speaker
But they can operate ah independent of human intervention, as you said. ok Correct. Yes.
00:02:25
Speaker
Okay. And it seems that the way the U.S. is approaching differs from the way China is approaching this. Is this fair to say? And can you give me a couple of examples of this? Sure. um If we take agentic AI as an example, so the way that U.S. has approached it is that they developed more general purpose agents.
00:02:47
Speaker
And what they focus on is to increase productivity at work. Right. And um in the case of Chinese companies, what they do instead is that they make them into sort of like a super agent and they integrate them into their equal ecosystem like e-commerce.
00:03:06
Speaker
social or service ecosystems, etc. So one example would be Alibaba. ah What they could do is that they can order tea for you, order a cap for you, etc. So this kind of the first layer, if you will. um Second of all, I think, you know, what we have to take into consideration is perhaps its cost as well as capability. um you know very I'm generalizing a little bit here, but in short, from a capability standpoint, I think U.S. agents and models are still more powerful, capable of doing a little bit more than what China has to offer currently. But then when ah when it comes to cost, then China is extremely competitive. Oftentimes, they're offering the same service at a fraction of the cost.
00:03:51
Speaker
Is that because the cost of energy is lower in China? And also they've taken a approach whereby a lot of this is on open systems. So there's a a general platform where companies can develop their own sort of AI agents, if you want to put it like that.
00:04:08
Speaker
I think energy cost, you know, based on some of the experts that we have spoken to would definitely be increasingly a very crucial driver going forward in keeping ai costs more intact for China and but be a continual kind competitive edge.
Chinese AI Companies and Monetization Challenges
00:04:24
Speaker
But also at the same time, it's important to also note that from a CapEx standpoint, China's CapEx spending to date is probably one-tenth of what the U.S. has put into currently. So I think that is probably another very decisive factor as well. Okay. So the key differences are energy costs, the sort of investment that they put in place, open sourcing, and therefore the sort of different approaches being taken in in China, more developing general models, agents that that can be applied in corporates. Within China, are there then also differences between companies? Do some take one approach versus another, or are they all taking the sort of same approach, but look at different sort of applications
00:05:09
Speaker
I think generally speaking, if we were to take, say, Alibaba and Tencent side by side, there are definitely differences in terms of how they run AI, but that's also owning to their respective business models.
00:05:27
Speaker
For Tencent, for example, ah the kind of kicker from AI was in improving the advertising business improving how they make money you know in their games business. For example, they can analyze user data more thoroughly. They understand when's a better time to sort of market to players.
00:05:49
Speaker
ah For example, you know they do it around pay date, etc. I think Alibaba, on the other hand, monetized quite heavily via the cloud business, which was also printing close to 40% growth in the last couple of quarters. and doing really well on that particular front.
00:06:07
Speaker
Okay. And this is something I'd like to talk about now, Charlene. um If I look at the sheer price performance of these Chinese companies, in particular some of the larger ones, they have not done so well over the last six months. ah It seems, therefore, that maybe there's something about how they make money out of this. Is that fair to say? And how do they make money, these Chinese companies, out of ah out of these new AI models?
00:06:31
Speaker
When we move to sort of the stock market, there are definitely multiple forces at work. The hyperscales in particular, you know, AI is just one facet of their business. So sometimes it could be other things at play that could hold back maybe Shepra's performance. I think ah that's number one. Number two, you know, we're seeing more of these frontier models coming to the market and and listed.
00:06:56
Speaker
And, you know, these are, you know, relatively pure. They're definitely pure plays. And they are very scarce. There are very few of them. And therefore, you know, they have done really well from a share price performance standpoint. And sometimes it would have maybe taken away some of the demand from a share price perspective from these hyperscalers. And then one last point is that the hyperscalers, Tencent and Alibaba alike, they're starting to invest more to acquire users, individual users for AI. But the market is a bit more skeptical in terms of you know how they can eventually monetize those acquisition efforts.
00:07:38
Speaker
um Obviously, subscription is a model that I think U.S. has done. you know quite extensively and OpenAI had some success with it. um but But when it comes to individual subscription, hasn't really been as successful for China historically.
00:07:54
Speaker
Okay. Why would the subscription model in China be not as successful as it is in the US?
Subscription Models in China
00:08:00
Speaker
What what do you think is the reason for that? I think Chinese consumers' propensity to spend is generally a bit lower. Okay.
00:08:07
Speaker
I think, yeah, this could be related by simply the dynamics in the consumer market in China. In China, you have an economy where there's a lot of supplies. There's oversupply in many industries. And corporates give out kind of sometimes if you buy their product, they give you a free bubble tea and these sort of things. So they're not used to paying for services as much.
00:08:27
Speaker
While in the U.S., you have a much more concentrated economy. Concentrated, I mean, among few industries and corporates. You don't have that oversupply issue and you don't have that that you if you open up a bank account, I'm going to give you free bubble tea or something like that. So maybe that's something that differences in consumer behavior as well.
00:08:47
Speaker
Definitely. i think China sees a lot more competition in general. Yeah, absolutely. Okay, so broadly speaking, what you're telling me is that China is actually quite close to where the US is. um ah But when it comes to making money out of this, it seems to be that on the corporate side, applying these models on the corporate side in China... That seems to go very well, but it's really the subscription models that seem to work in the U.S. somehow don't seem to work as well in in China.
00:09:16
Speaker
And that seems to be a key, one of the worries that people have with these stocks, in addition to what you mentioned, that there are new plays coming into the markets or the stock market needs to digest it as more companies they can invest in when they invest in the AI theme. Is that fair to say? do you really To be really fair, I don't think these Chinese companies have even started the subscription model. But then I think the market takes the view based on historical data that this is going to be a much harder route versus overseas consumers.
00:09:47
Speaker
So I think it's it's an expression of that concern as opposed to actual data that you know China is lagging behind. I think Tencent had told us that, you know, the pro-consumers, what they call the professional consumers, at least in the initial stage, seems really receptive to paying for subscription. It's just that I think the market takes a more conservative view on whether, you know, there is a very big pool of people that they can potentially really tap into from a subscription standpoint.
Future of AI in China
00:10:16
Speaker
Fantastic. I think there's been a fantastic oversight over what's happening in China, how they differ with the US, and what kind of the way forward is for some of these Chinese companies as well, and particularly looking at getting individual subscribers hooked hooked onto their models. Thanks a lot, Charlene, for dialing in from from Singapore. And no doubt we'll be having you back on the podcast to see how this story, yeah this important story in Asia, is unfolding over the next couple of quarters center and years.
00:10:46
Speaker
Pleasure, it's mine. Thanks, Harold. Thank you. Bye-bye. Well, that's all we've got time for on this week's podcast. Another great discussion under the Banyan Tree. Listen, like, and subscribe on Apple, Spotify, YouTube, or wherever you get your podcasts.
00:11:02
Speaker
Under the Banyan Tree is an HSBC Global Investment Research production produced by none other than Graham McKay. Thanks for listening, and we'll be back again next week.