This week’s podcast is about my visit to WAIC 2026 in Shanghai.

You can listen to this podcast here, which has the slides and graphics mentioned. Also available at iTunes and Google Podcasts.

Here is the link to the TechMoat Consulting.

Here is the link to our Tech Tours.

Take-Aways:

1. Robots and AI are converging rapidly.
2. There is a convergence of world models / building, video generation and 3D model generation
3. Tencent has a very rapid product cadence. “Choose Your Buddy” is a user-friendly approach.
4. We are seeing lots creativity and combinatorial innovation. A Cambrian explosion.

——–

Related articles:

• More High-Tech Flex by Huawei R&D (4 of 4) (Tech Strategy)
• 6 Big Events in AI Agentic Ecommerce (Tech Strategy)

From the Concept Library, concepts for this article are:

• n/a

From the Company Library, companies for this article are:

• n/a

—–transcription below

Episode 289 – WAIC.1

[00:00:00] Welcome, welcome everybody. My name is Jeff Towson, and this is the Tech Strategy Podcast from TechMoat Consulting. And the topic for today, my four takeaways from the World AI Conference in Shanghai, which was this past weekend. And it was spectacular. I mean, I go to a lot of conferences, and I, I do a lot of these visits.

This was on a level, like top level, spectacular, vibrant. Um, not just because the temperature was crazy, so every- It was like a heat wave in, in Shanghai, so everyone was sweating like crazy the first day. It was unbelievable. And then it was like a rainstorm. But, but apart from that, like, the energy of being at sort of the AI epicenter of China this year, given what’s going on, and then what it turns out is the floor above that was robotics.

So it was like the [00:01:00] epicenter of, of AI and robotics in one building, uh, you know, right now when all of this is sort of happening so fast. So yeah, it was exciting. It was really kind of great. I’ll sort of give you my– I’m going to do a lot of writing on this, summarize companies to look at, um, what companies that I follow, what they’re doing.

But I’ll give you sort of my four high-level takeaways, uh, for what I think, at least for me, that was the most important thing. So that’ll be the topic for today. Let’s see. Standard disclaimer. N- uh, nothing in this podcast or my writing website is investment advice. The numbers and information from me and any guests may be incorrect.

The views and opinions expressed may no longer be relevant or accurate. Overall, investing is risky. This is not investment, legal, or tax advice. Do your own research. And with that, let’s get into the topic. Okay. Yet again, let me sort of start with, I don’t know, a little update to my previous rants on sort of China versus the US versus Anthropic versus all that stuff, cause, man, stuff just keeps [00:02:00] happening.

And, um, the big news was Kimi. If you’re following the, you know, sort of the AI news. Kimi K3 gets dropped, um, open source, open weight, low cost mixture of expert model coming out of China, and people start claiming that it is performing at close to frontier level. So Anthropic, OpenAI, you know, leading edge frontier models.

Now, other people I talk to say, “Nah, it’s not quite that good.” It’s good, but it isn’t, you know… But it’s close. It’s, it’s, it’s much closer than anyone thought for a model, you know, where instead of $50, you know, per million tokens, you’re paying 50 cents. Right? That’s crazy. The cost, the performance. Anyways, and that’s sort of been dropped, and that’s kicked off political discussions.

It’s kicked off sort of real business discussions. How do you compete against something like that? If your plan was to be Anthropic and OpenAI and charge all this money for your high price subs, [00:03:00] and this thing does 90% of what you do, 95% of what you do That’s, that’s a major business move. And now what people are– If you go on Twitter right now, what people are talking about is like, okay, Fable is out there, the Anthropic Frontier one.

Fable is the one with the guardrails. Mythos is the one without the guardrails, but you have to be on a short list approved by some government quasi-entity to get on it. Okay, Fable’s the one out there with the guardrails, so it will not let you do certain things But what people are doing in the US is on Twitter, right?

So take that with a grain of salt. You know, they’re looking for security weaknesses in their sites, and when they go to Fable, it basically says, “I can’t perform that for you. Sorry.” Right? You hit a guardrail. Then they do it on Kimi, and it finds the weaknesses cause it, it says, “Fine, I’ll do it.” That leads– That’s a very weird situation.

That means if you’re trying to do white hat [00:04:00] stuff and protect your site, you can’t get access to the leading models, but if you’re doing black hat and you’re using another model outside, you can. So it’s like the US firms, and anyone dependent on this, has been left undefended because you can use one but not the one that’s approved locally.

So it’s a very weird situation. People are checking it right now. It’s interesting to watch. Um, anyways, I want to talk a little bit about that cause, um, the guy behind Kimi is actually pretty interesting. This guy named, uh, Yang Zhilin, and if you go on X, you’ll see people are finding talks by him. He’s 34-year-old guy, a rockstar computer scientist trained at Mar- He did Carnegie M- Well, one, he came out of Tsinghua, which is, you know, sort of top, top of China.

It, it’s more like the MIT of China. Peking University is more like the Harvard. Then he does his PhD in machine learning at Carnegie Mellon, which, you know, I think it’s still the number one computer science program in the country. It was for a long time. It’s in the [00:05:00] top three, right? Under all the big guns.

He’s the guy. Then he goes off and, uh, he works at Meta AI. He works at Google Brain for a while, and basically, you know, 2023 in March, he launched his Moonshot, Moonshot AI. The product is Kimi. Pretty much everyone just calls it Kimi now. Um, interesting factoids there. His name, you know, people have like an English name they go by if you have a Chinese name.

His name apparently was Kimi. Like, apparently that was his nickname cause people, you know, they can’t really say Zhilin. Uh, I don’t know. That’s what I’ve read a couple places, so apparently it’s named after him maybe. And Moonshot AI actually in Chinese is closer to Dark Side of the Moon because that’s like his famous album, the Pink Floyd album, and he released or he founded the company in March, uh, 2023 with a couple classmates from Tsinghua.

He founded it on the 50th anniversary of the [00:06:00] Pink Floyd album , right? I-interesting guy. Anyways, you can watch his talks online. Very technical, but yeah, he’s really interesting. So you got this guy out there, um, clearly high power. He reminds me a bit of, uh, you know, there’s this whole new generation of these young entrepreneurs coming out in China.

Liang Wenfeng, that’s the founder of DeepSeek, right? And he launched DeepSeek in, you know, July 2023. But, you know, he’d actually been doing AI for a decade before that, doing quant stuff and making a ton of money in a hedge fund doing all this stuff. And DeepSeek, you know, is funded or was for a long time funded by the profits from the financial service trading business.

Now they’ve raised some money, things like that. So you got these two interesting guys. There’s others, but those two are interesting. When you listen to them speak They sound a lot like Elon Musk in 2015 when he launched OpenAI.[00:07:00]

They talk a lot about intelligence should be ubiquitous and inclusive of everybody. Everybody in the world should have it. We- we’re going to democratize intelligence and make it open source. Open weight, open source, and let’s make it really, really cheap so that everyone can afford it. Now, that’s good business strategy, but it also lines up nicely with an ideology, which is a lot of how OpenAI was supposed to be.

So there’s an irony here that, um, DeepSeek and Kimi are kind of what OpenAI was supposed to be. You know, and Elon Musk’s– I think it was him who said this. His comment was he founded OpenAI as like a rainforest protection, you know, nonprofit, and it decided to turn itself into a timber company, which is a pretty good criticism.

They sound, when they talk about what they’re doing and their mission in life, they’re building frontier-level, [00:08:00] low-cost open models that they’re giving away to the world, everybody. I really like– I mean, now we’ll see in practice how this plays out. But if you listen to them speak, that’s how they sound, which is a real contrast to listening to, say, Dario and Sam Altman over at– They don’t sound like that at all.

They sound like two guys with a technology they’re turning, trying to turn into a regulatory, uh, duopoly by a lot of fear-mongering, and so that’s my opinion. Anyways, that’s interesting. Okay, so that’s where that is. Now, last time I said, you know, look, the next shoe to drop would be if the US government, probably the White House, starts to ban, uh, Chinese open source models.

Now, the real enemy, in my opinion, is open source at all, and Dario from Anthropic has been talking about that forever. Oh, you know, open, open weight, open source are bad. They’re risk. You know, it’s– He’s– Now, I think that’s self-interest, but I don’t know. But it’s easy to throw that [00:09:00] stone if most of them are coming from China cause then you can play the geopolitical card.

And that’s what you hear, and now the White House has started to say, “We are starting to review Chinese open weight models,” and what are the concerns, the national security. So you can see that shoe starting to drop Now, if they do that, if they ban these things in the US, well, that’s a very interesting situation because we’ve seen that situation before.

That’s what happened in EVs. The entire rest of the world is buying Chinese EVs right now, BYD, Xpeng. They’re fantastic cars. They’re super low price. You can get a BYD for $9,000, the small one. Um, everyone in the world can get those except for Americans cause they’re not allowed in the US. Is that good for Americans?

No, not really. Is it good for the automakers? Yeah, it probably is keeping them alive. Let, you know, let’s be honest. Okay, is this going to be the same scenario where the whole rest of the [00:10:00] world’s going to get low cost, super affordable, frontier level open weight models and, and Americans aren’t? Is that the scenario?

And if that’s going to be the scenario, is the same thing going to happen for robots? You know, the joke that’s been going around the last month is Europeans don’t have air conditioners. Um, okay, how did that evolve? Like, is it architectural? Is it green ideology? I don’t know. But some point– at some point, they weren’t adopted, and now you get this weird scenario where the rest of the world goes to France, and it’s like you don’t– like only 20% of homes have air conditioners, really?

It’s just bizarre to everybody else, whatever the reason was. Is it going to be like that where you go to the US, it’s like, “Dude, your cars suck, and your models are crazy expensive. What– You know, this is weird, and your robots suck too.” Like, is that the path? I don’t know. Anyway, so that second shoe started to drop, and then one last bit, and then I’ll get off [00:11:00] this.

At the World AI Conference, President Xi came the first morning, and this is the first time he’s been there. And he gave a pretty– I wasn’t there for that. I, I– You got to– It’s actually pretty hard to get tickets. I had tickets for the exhibition, and then I had– I was at the Tencent Forum, but I didn’t have those tickets.

Um, his speech is fascinating because… I mean, I’ll read you the parts of it. This is my opinion, so this is not an exact quote. But this is positioning China as the world leader for inclusive open source AI ecosystems. I’ve– And that is a strategic contrast that’s as stark as can be to what the US is doing, where not only are these high-priced proprietary models, but the US government is also [00:12:00] increasingly cutting certain people and groups off from those at will.

That’s not inclusive at all. They’re going the other route, which sounds a lot like what Kimmy and DeepSeek are talking about. The other points he made, and I’m paraphrasing, that the concept of national security should not be overstretched within the area of technology policy. My impression of that is like, look, don’t use this as an excuse to ban everything for whatever reason, which seems to be kind of what we’re hearing in the US.

Um, we don’t want technological monopolies or digital divides between haves and have-nots. That’s my language. And there was an interesting phrase he used. We, we want to prevent new historical injustices That’s interesting. It’s almost like a colonization thing. Anyways, the whole talk, one, that he was there was important.

Two, you know, when you hear that, that’s a real contrast with the US. I actually agree with this vision. I do. I think that’s the right North Star for how [00:13:00] this should develop. I don’t like this idea that some people get the great stuff and some people don’t get good intelligence, and entire populations- and politicians decide, and entire populations are just cut off.

You know, we need to control this because we don’t want… That, that would be like if electricity was built, and instead of helping the whole world get electricity, it’s like, “No, no, no, we don’t want those countries to have electricity cause then they could challenge us, so we need to prevent certain entire populations from not having electricity.”

I mean, that, that’s pretty evil, in my opinion. So I like this as a North Star overall. Uh, anyways, that’s sort of the update. Things are moving. I’ll get off this topic. But, uh, I, I mean to keep getting off this topic, but, like, literally every couple days, something important happens. It’s really fascinating.

Okay, let me get into my four takeaways. This won’t take very long. Number one, we are seeing robots and AI sort of combining in real time. It was fascinating- This is an AI [00:14:00] conference, right? The big robot conference is in August in, uh, Beijing. But man, 30% of this conference was robots. And every, you know… Now, they didn’t have all of them.

They had s- It was more humanoid and custom- consumer-facing than, let’s say, industrial. But every robot company you’re talking to, they’re talking about their embodied AI system and what they’re using. Man, it’s fascinating. And then the other models are all about… Well, one room was… There were three rooms.

One room was robots, one room was really AI, and one w- room was kind of chips and telco and more of the infrastructure side. Okay. Now, if these Chinese open source models are such a big deal right now, Kimi, DeepSeek, what happens when the open source VLA and world models drop? Cause all of these robots, the embodied intelligence, well, they’re all based on foundation models, not LLMs.[00:15:00]

You know, there’s usually, let’s say, four foundation models you need for an advanced robot to have embodied intelligence. You know, you need the, the VLM, the sort of vision language model. That’s really the robot plans what it’s going to do It, it’s reasoning, it’s planning. I’m going to move my– I’m going to go over there and I’m going to lift my arm, so and so.

You know, then you get to sort of the VLA model, the vision, um, language action model. That’s when it actually, okay, now I move my hand. Um, you have that sort of the take action phase. You have sensory models. That’s a lot of feedback, right? The hand grabs something, it feels it. The eyes see it. You know, you get sensory, you get the feedback loop.

And then you have the world model, which is, you know, the apartment has been mapped out and, you know, you can sort of see how things are happening in your vi… Well, you need at least four types of foundation models for embodied intelligence. It’s more [00:16:00] complicated than that, but that’s generally true. Okay, who do you think’s going to make those?

Are we going to see the exact same… I mean, everyone in the US, well, the politicians are freaking out about Kimi right now because it’s low cost open source. What happens when the VLA models come out of China? Cause they’re next. They’re coming really soon. They’ll be a DeepSeek or a Kimi in embodied intelligence models very soon Well, then it’s the same problem.

The difference here is you need the hardware and you need the models. Well, how do you compete with both of those things out of China? I can kind of see how Anthropic can compete with DeepSeek or Qwen. You know, it’s software versus software. But in this case, it’s going to be software plus the hardware, so you have to compete with the whole sort of open VLA model ecosystem, and you have to compete with Chinese manufacturing, right?

How do you beat that? So this sort of story we keep [00:17:00] hearing from, uh, the political realm, I think we’re going to see another version of it very quick, which is going to be em- embodied intelligence and robotics, and it’s going to be a combination of this open weight, open source thing. And look, you really can’t compete with Chinese manufacturing.

Nobody can. That was another thing like that really struck home. When you walk around the, um, the conference, there were so many frigging robots. I couldn’t believe it. I– There were so many companies. They were everywhere. And stuff that was really shocking 12 months ago, like a robot dog that can do a flip, Unitree.

Well, one, the robot dog can now do a double flip. I’ll put the video up online. Like it, it jumps from a standing position and does two flips in the air and lands. It couldn’t do that six months ago. And by the way, robot dogs, everyone’s got those. That was special 12 months ago. It’s everywhere. Everyone’s got a robot dog.[00:18:00]

And then you go and you get coffee, like, you know, these, these retail… They’re calling them retail cubes, where you put a robot arm in there and it can make coffee, it can be a barista, it can be a bartender, it can make ice cream. That was interesting a year ago. Everybody has those now. In fact, most of the booths, not most, let’s say half of the booths at the exhibition hall, you could get free coffee cause they had a humanoid robot up front or an arm robot making you coffee.

It was nothing. It went from special to ubiquitous. Like, eh, it’s just another robot making coffee. That’s nothing. The dogs, nothing. Humanoids, everywhere. Like so many companies. It was crazy. Like the stuff that impresses me Um, there are a couple, couple things that got my attention. Okay, within all of that, the two that I thought were interesting, um, definitely the humanoids.

Right, now the humanoids, kind of two versions. You have the, the bipeds, right? They walk [00:19:00] around, you know, and they do kung fu now, and there’s interesting training for them where you can basically put on exoskeleton gloves and helmets, and it will basically copy you in real time. So you can, you, you can do tai chi with their little exoskeleton on, and the robots behind you do the same thing.

So that’s actually useful for training. But there’s the bipeds, and then there’s the ones where it’s sort of like a centaur. The bottom half of the robot is wheels, the top half is a humanoid. That’s kind of the most common one that I saw. Uh, cause you can use those for… Well, the bipeds you use for stuff like cleaning the house, uh, household chores, being a companion.

They can be small little guys that can kick the soccer ball, which they have. They play sports not bad. They can be medium-sized ones that do MMA. You can see robot MMA online, just go on YouTube now. Or they can be, you know, if they have the, the wheels underneath, then they’re probably doing loading or [00:20:00] unloading in a warehouse, industrial handling.

Uh, maybe they’re doing logistics, they’re doing sorting. And within that, you know, you got two types of hands. You got the basic gripper claws, which are more primitive, but it’s enough for, let’s say, sorting or picking up a crate. But the real impressive ones are the dexterous hands. Um, those are five fingers and- In November, I went to one of these conferences, and you could sort of show the camera your hand, and the robot hand would sort of mimic it slowly.

It’s instantaneous now. If I put my hand in front of the camera and I start making hand signals, the hands react and copy what I’m doing immediately. Like, it’s instant. And the dexterous hands, they can play the piano. They can do fine sort of manufacturing assembly, like putting in small components within circuit boards.

That can happen now. [00:21:00] That was pretty impressive. So the dexterous hands got my attention. The humanoids that are centaur-like got my attention. Uh, the retail cubes, which I mentioned before, we went to Galbot a couple months ago. You know, the retail cubes, you basically put them out on a sidewalk. It’s got a robot inside it with a screen, with a cute digital avatar, and you can, you know, you can get coffee whenever you want.

It works twenty-four/seven. It never stops. You can get drinks or whatever. Well, pharmacies have those now, and 7-Elevens have those now. Lawsons have those now. So you can get these retail cubes anywhere. And so we saw a decent number of pharmacies, retail chains. Those are pretty impressive. Those are going to be everywhere.

So anyways, there, there’s a bunch of types, but those are kind of my five types I was paying attention to. The robot dogs, which are mostly for security and inspection, uh, the humanoids, which come with a couple varieties. The hands are a big deal, the retail cubes. Um, pretty interesting. So [00:22:00] yeah, that was kind of my first takeaway.

Like, we’re going to see a tremendous– And I’ve been digging into the models, and you can put these into two buckets. Bucket one is when you have someone just making robots, and they’re not doing the software themselves. So they might be using software created by Tencent, which is creating robot software.

Robot X is their division. It’s called Tyros. I’ll talk about that. I’m going to talk more about Tencent in another, in a podcast. They’re just providing the software for anyone who makes hardware You know, it’s like Android or something. You can… Anyone with a smartphone can download it and put in. That’s what they’re doing.

And then you have other models like Agibot, uh, Galbot, which they’re doing the full stack themselves, the software, the hardware, all integrated, uh, which you probably need for high performance at this point. Uh, that’s kind of how it plays out. So I’m digging into the model side much more than the robots.

Okay, so that’s takeaway number one. This intersection of robots and [00:23:00] embodied intelligence, it’s happening right now. Like, we’re going to have a DeepSeek moment very soon in this space, I think. All right, takeaway number two is there’s an interesting intersection that I didn’t really appreciate before, and now I’m– it’s, uh, sort of on my radar, which is there’s an interesting combination happening between world-building, 3D modeling, and video generation.

Let me explain first, like, okay, 3D modeling, right? This is pretty common. You can build a little digital avatar yourself. And this is just generative AI. Um, you know, put in a prompt, put in a picture, and it will turn it into a 3D version of that very quickly, which, you know, a lot of people, these… You put them into video games.

These become 3D digital assets, which you can then upload into video games and things like that, which is very helpful. You can also 3D print these things. I’m going to start playing [00:24:00] around with 3D modeling a little bit more. I think it’s pretty fun. Okay, fine. So you can kind of see that, and that ties nicely in with gaming, and that usually tees up the idea of world-building.

Okay, now world building in the video game version. Now 3D modeling we could do not just ob– we could do s- we could do weapons, we could do characters, we could do, uh, cities, right? 3D modeling you can do a lot. But at a certain point, you start to get into the idea of world building, you know, world models, where, you know, you have to sort of bake in the physics of these things.

Like if you, you know, if this building falls, what’s going to happen? And that’s, you know, that’s where you, you can see the sort of intersection there in gaming, right? You do the assets and then you sort of get into world building. But then there’s also that sort of intersection I just mentioned between world models and robots.

Because it turns out training robots– Well, one, they have to [00:25:00] perceive the world around them, so they have to be able to understand it and sort of build an internal model against that. And I’m kind of mixing terms here, but, but at the same time, a lot of the training that happens for robots is done within these worlds.

So a company like NetEase, when it was getting into generative AI a couple years ago, you know, they said, “We’re, we’re working with, with robot companies, and they’re using our video games and the models we build as a gaming company to train their robots.” Cause getting training data as a robot in the real world takes a tremendous amount of time, and you’re not getting all the edge– you know, the long tail of weird things happening.

So no, you, you know, the robot sort of can get trained in this interesting combination of being in the world and seeing things and trying things, but also syncing that experience with sort of digital twin world models and, you know, you can train there as well and then go back and forth. So anyway, the world building and then world models are, they’re a bit different, but you’re, you’re kind of in the same category of [00:26:00] product.

Okay, so there’s this interesting intersection of 3D modeling and world building. And then the other bit which I didn’t really appreciate was just video generation that, you know, if you’re going to do video generation, let’s say you take a photo, standard stuff like Kling or, or whatever, Happy Horse. You know, if you’re going to do video generation, okay, you take a photo, you tell it to make a video.

But when it makes the video, you have to import or figure out a tremendous amount of physics. Okay, you have to build a city in the video. You have to have the physics where if something falls off a table, here’s what happens. Well, generating all that physics and such within the video generation tool is, is quite consuming.

But if you’ve tied it to a world model, you can sort of combine the physics between those two. So there’s these interesting linkages between 3D modeling, video generation, and sort of world model, world buildings that I hadn’t really thought about. And I’m, I’m digging into that cause I think [00:27:00] that’s kind of sort of an amazing space.

I’ll give you a couple examples of companies. Uh, there was a… Tencent has released a new agent called Miora, and it’s basically– They’re, they’re releasing multiple agents right now. It’s crazy. There’s WorkBuddy, there’s CodeBuddy, there’s Qclaw, but this is their design agent, so this is for people who design things.

You know, you, you basically get an unlimited canvas where you can start to engage with the canvas in various ways. Here’s a photo, here’s some text, here’s a video, here’s a digital tool– you know, a digital 3D asset I have. And you can sort of engage with the agent, and it will put these things together so that you can build, you know, whatever you want.

You can build ceramic art pieces that are printed out. Well, not ceramic, but pieces that are printed out. You can build worlds, you can build images, you can… anything. You can build logos, whatever you want as a designer, right? It’s kind of the Adobe killer. And because it’s an agent, there’s a couple things it gives [00:28:00] you.

One, it gives you sort of tremendous memory Where, you know, if you try and do stuff online now with, with video generation, it’s kind of a pain because, you know, you have to give it a prompt. It takes a try. You maybe edit one of the results. You can try again, but you’re always starting over. Well, this has sort of long memory, and it knows what you’ve built before, and it can look at all…

So it’s sort of this unlimited canvas where you can keep going and keep going and keep going. Well, that, that gets you into 3D modeling. It gets you into video generation. It gets you into world-building. So I thought that was sort of an interesting example of that. I’m going to play with this one. It’s on my shortlist.

Um, there was another company called Hyper 3D, which basically just– pretty cool company, which just generates 3D digital assets for gaming companies. So anyone who wants to build a video game, you can get your, your characters built. You know, that’s what they specialize in. There’s another company called Wondershare, which is basically just a pure AI TV [00:29:00] studio.

If you want to start your own television studio or movie production business, you go to Wondershare, and it helps you write the script, it helps you generate the characters, and then it starts putting together… And this is all video. And you can use models that you want. You can use Kling, you can use Happy Source, you can use Seedance.

It has its own model too. So that one’s just purely video generation. The other one’s sort of more gaming. And then you have a company called Shengshu, which sort of is in between world-building and video and all of that. So anyways, that whole space, I’ll put the names in the show notes, but the-that opened my eyes.

I wasn’t appreciating how interesting that space is becoming. And it’s just a half step from building stuff as a 3D image to printing it out or having it done by a manufacturer. You’re one half step away from product development in that Really interesting. Okay, um, number three. I’m going to finish up here pretty quick.

Number [00:30:00] three, Tencent. I’m focused a lot on Tencent this year. Um, the products they’re rolling out, it– Like, the speed is, is crazy. It’s crazy. I think this is going to be a story this year. I think this has not been understood. Like, I talked about this before. They redid their sort of generative AI approach, uh, over about nine months ago they started, and they kind of redid the whole process.

They hired a new team. They created a new data approach. They created a new process, and then they integrated it in with their product teams, cause they have, like, a million products, and they sort of co-design either new generative Apr– AI products, or they integrate AI into their existing products. And the first sort of attempt at using this new process was the HY3 model preview, HunYuan 3, which, you know, that was their open source, open weight sort of version, and it’s, it’s good, and it did real well.

That was the first one. Now they’re just rolling out [00:31:00] products like crazy. So they had a big thing, a big sign. I’ll put this, put it in the show notes. It was just this big sign at the entrance to their exhibit that said, “Choose your buddy.” And they had basically nine or 10 different buddies you could choose.

Some of them are AI assistants, some of them are f-full agents to do whatever. One of them was Miora, the design agent, but one of them was WorkBuddy, one was CodeBuddy. Um, the list is just growing really fast. Uh, and then robotics is there too. So I’m going to… I’ll do a separate article and podcast on Tencent cause I also attended their forum, and they’re much more into robotics than I was aware.

I knew they were doing it. I didn’t know they were doing this aggressive. They’re doing HY world models as well. So yeah, they were on my shortlist of, okay, I got to… I’m pretty good at Tencent, but I’m going to get a lot more into that company. All right, number four, and this is high level. This is just last thing You know the old phrase I’ve talked– I [00:32:00] talked about this years ago, this idea of a Cambrian sort of explosion.

Is, is it explosion? Anyways, they talk about the Cambrian era, like the, you know, back in the evolutionary times, you know, all the h- all the life forms or a lot of the life forms we know all sort of emerged in a very short period of time. You know, not many life forms for a long time, plankton, simple cells, and then in a very short window in sort of, you know, universe timescales, all the animals emerged very complex.

And what happened? Well, it’s a good, it’s a good sort of a analogy or example of Ca- you know, combinatorial innovation that, you know, a lot of innovation is putting two things together. Well, at a certain point when all the core elements are in place, you can start to put them together, uh, very quickly. But you need all the pieces.

You need the data systems. You need the software. You need the semiconductors. You need the AI tools. You [00:33:00] need the robots, right? But once all the pieces are there, then everyone can just go to town and like, “Hey, let’s make a robot shark. Hey, let’s make a robot that does, uh, MMA boxing. Hey, let’s do gen- you know, generative AI videos in a studio and then print them out and sell them in stores.”

You know, you need all the pieces to finally be in place, which I guess is what happened in the Cambrian era that, you know, the cells were there, the mitochondria are there, the DNA was there, you know, all the p- and then all the life forms just came. So that’s what was running through my mind as I was walking around, cause there was all this stuff that was just strange.

Like there was just all this creativity and stuff I didn’t even think about. Um, there were smart guitars. I’d never even thought about smart g-guitars. Basically, it’s a guitar. Um, it doesn’t have strings, but has sort of soft pads where you would put on the neck, and then it ties to a bunch of AI, and you can…

You know, there’s a big screen you can touch. And you can basically play guitar without having all [00:34:00] the difficulty of learning the chords or pressing those strings down, which is very difficult. And yeah, you, you could basically become a guitarist very quickly now without all the years of training if you want to do.

And I, it was, it was fantastic. You know, is that a good thing? I don’t know. But it’s like if I want to, if I want to make videos, which I like to do, I don’t need to go to film school. I can use Seedance. I make videos all the time now. If you want to make comic books, I can’t draw. I could make a comic book right now.

I could make paintings and put the… I could make Van Gogh-level paintings right now. I, and I have no– I, I never trained to do that stuff. So, you see all this sort of crazy stuff happening. The smart guitars were cool. Uh, exoskeletons I thought were neat. You know, these exoskeletons you can put on your arms and on your legs to increase your strength.

Those were pretty interesting. Um, human faces and hands, these were on Twitter a lot, that, you know, some of these robots… A lot of the robots are [00:35:00] just, you know, they have this mechanical shell, right? And they have, for the face, they might have a cute little digital image. But then there are certain people that are making robots that they’re trying to make them look just like humans.

You can literally buy this rubbery-like skin, they look like gloves for hands, and put them on the robot hands, and they look kind of like human hands. And, you know, instead of having a humanoid body, they build the body out and put it in a suit, and it looks like a human body when it walks. Not 100%, but it’s getting close.

And then, of course, they have the faces. They have these– They’re trying to be attractive women that are, you know, chatting with you or singing on stage as pop stars and… I mean, there is just all this creative and having this sort of human similar robots. That’s pretty weird. I saw vending machines where they’re just for sampling and trying things.

So, you play a game on the [00:36:00] screen, and if you win, you get a free sample of whatever. So, it’s a way to get people to try your product. And if you have a perfume, it sort of sprays out a little perfume, and you get points, and that was kind of strange. The little robots playing soccer. Turns out, uh, the robots can kick the soccer ball pretty well, the little ones.

That was interesting. Maybe the one I thought about the most was the smart glasses cause I really like the Quinn, which is the Alibaba smart glasses. I think they’re great. I mean, the technology’s cool. You can sort sit there in a conference and, you know, everything you see on the screen, if it’s in Chinese– I read Chinese pretty quick, but my, my listening is not that good.

It’s mediocre at best. You know, it just plays it either in my ear as audio, or it, it scrolls across the screen. Or if I want, it just ties to my phone, and I just look down at my phone. You know, I was doing that stuff. They’re great. Now, I go back and forth on the smart glasses because [00:37:00] there’s a couple of use cases I like.

Um, I like that case for translation. I think that’s very, very helpful. I like the smart one, uh, the sports glasses Where if you’re a bicyclist, you know, you have those wide wraparound glasses that bicyclists always wear. You use those glasses to sort of look for potential accidents, look for problems on the road that could, you know, throw you off your bike.

Uh, they do communication with you and your, your teammates, so you can talk to the other riders easily through the headpieces. So I thought, like, the sports glasses were quite cool for a lot of things. I think the smart helmets for motorcycles are amazing. Uh, translation. Um, there’s these use cases I kind of buy, and then there’s this big question of, look, I don’t think people like wearing stuff on their face.

I think– They just don’t. If I can get everything I just said to work on my phone and not on the glasses, that’s [00:38:00] what I would probably do. I- if I could put a little camera on my collar and it did everything I just mentioned, and I just had to s- quickly look down at my phone in my hand with an earpiece, I think I’d probably do that.

So I’m always on the fence with the smart glasses, but there were a couple use cases I thought were great. So anyways, I’m probably going to get the Quinn ones. I thought they were pretty awesome. And it’s Alibaba, so, you know, the model is amazing. So that software-hardware integration’s going to be… I think it’ll probably be the best in the world.

I mean, Facebook, yeah, they got the Meta glasses, but their, their software, people aren’t using their models that much. You know, who’s got the hardware and the software? It might be Alibaba. Uh, maybe Tencent could do it. Anyways, we’ll see. So anyways, there was just all this crazy stuff. Um, that’s kind of point number four.

There’s just sort of explosion of creativity happening. It, it’s almost like I heard it described by someone that we are at the end of the document era, and now we are at, we are at the [00:39:00] beginning of the build era, where people like me, you know, we would do a lot of talking and a lot of thinking and a lot of writing, documents, uh, PowerPoints, whatever.

And then based on that, we would launch a team to do something. But there’s a lot of document writing involved in planning to do things, looking at things we have done, fixing them. We, we spent a lot of time on Excel and Word and, you know, spent a whole life doing this. And now it’s almost like, are we just…

Is that an anomaly, and we’re just going to have the build era? Any idea you get in your head, you can just do it immediately. You know, I would like to build a 3D digital asset that’s me as a samurai warrior. I want to put it into a video game, and then I want to build the video game around it, and then I want to print out the merchandise and make all those things and sell them in stores and put it up on Taobao, and I can do all that with no planning.

I just talk to the agent, and pretty much all of those things can happen right now. [00:40:00] So it’s like we’re just jumping to the build era. Any idea, just do it And you see people sitting at their desk with all these tools on their screen that can basically do that now. Now, organizations are more complicated than that, but at least at the product level or at the content level or the video level, yeah, I think we’re, I think we’re just at the do it, sort of build it right now.

No more documents, no more typing, just do it. Uh, apps, you can build apps that way. You can build websites that way now. I heard that described, so that, that was kind of got my attention. Okay, that’s it for first pass takeaways. Um, robots and AI, it’s all happening right now. China’s definitely the epicenter.

It’s amazing, both individually and combined, amazing. Uh, point number two, world models, world building, video generation, 3D modeling, those have a lot of interesting connections that I think are quite powerful. [00:41:00] Uh, three, Tencent’s product cadence is kind of crazy. Um, the whole choose your buddy thing is really interesting.

That’s a good… Their focus in a lot of this is to put things into their existing ecosystem and make them super easy to use for anybody, which is kind of, that’s kind of an important piece right now. And then the last thing, just sort of Cambrian explosion combinatorial innovation. Anyways, that is it for the content.

I hope that’s helpful. It’s kind of fantastic. I have a whole stack of like reports I, I got, well, brochures and pamphlets by so many companies. Um, yeah, it’s crazy. I’m, I’m learning a lot. I feel like I’ve been studying full stop for the last, uh, week. Anyways, that is it for me, and, uh, I’ll, I’ll have a lot more on this and get into some more actual strategy lessons within this other than just fun stuff.

Okay, that’s it. Talk to you next week. Bye-bye.

——–

I am a consultant & keynote speaker on how to increase digital growth and strengthen digital AI moats.

I am the founder of TechMoat Consulting, a consulting firm specialized in increasing digital growth and strengthening digital AI moats. Get in contact here.

I write (a lot) about digital growth and digital AI strategy (3 best selling books, +2.9M followers on LinkedIn). There is a free book and email newsletter below.

My Moats and Marathons book series is a framework for building and measuring competitive advantages in digital businesses.

Note: This content (articles, podcasts, website info) is not investment advice. The information and opinions from me and any guests may be incorrect. The numbers and information may be wrong. The views expressed may no longer be relevant or accurate. Investing is risky. Do your own research.