00:00Now you have your Matrix 3, that full-stack humanoid third-generation.
00:05Tell us a little bit about where you are in the production timeline.
00:09Are those pilots, have they been already deployed?
00:11Where are they going to be used?
00:17Okay, this is a very good question.
00:19We just launched the Matrix 3 Dismegg, which was near about two weeks ago.
00:23And this is built to develop the most advanced physical AI intelligence.
00:30And also, as you can see from the picture, that coming up was 27 degrees of Dolph, degree of freedom,
00:38Matrix 10.
00:39And also, we have our custom-made AI model, which we call as the wave.
00:44We can use in the brain or the cerebrum to instead.
00:48So, actually, you can see the current China market is one of the most important robotic markets in this decade,
00:55for sure.
00:56It is decided by three things.
00:58First, is the physical policy backing.
01:01And also, China has world's most deepest manufacturing supply chain.
01:05So, this market is willing to deploy the robots at a scale.
01:10So, we truly believe that 2026 will be the inflection year.
01:14And the Matrix 3 was built to fit naturally and safely in the human space.
01:20And as well, working alongside with people and help us to take those dull, repetitive and dangerous work.
01:27So, this is what we are all built about.
01:30So, Matrix 3 is ready to go.
01:34Ready to go.
01:35So, what's the timeline right now?
01:37How many units are we talking about?
01:39Who would be your biggest clients, especially since you talk about users beyond just industrial cases in factories?
01:51We are talking about 500 to win-sour unit deployments starting from January this year.
01:56And the production line is already launched in Shanghai, Zhangjiang.
02:02We call it the Manufacturing Fabrication Hall.
02:04And we start the pilot production for the early entry customer already.
02:10And the first shipment is coming out end of June and middle July.
02:16So, the cost, as I understand it, would be around $80,000, start from around $85,000.
02:28So, where would the margins come from?
02:31Are we talking about the hardware or the software customized service for you as a business?
02:40Actually, we have two models.
02:43The Matrix 3 is coming from the $85,000.
02:45And the pro version will come with about $99,000.
02:50So, the margin will mainly come from the annual recurring fee.
02:55So, we don't want to make too much profit on the hardware part.
02:58But we really believe that the tokens, that the computing powers that allow us to reasonably at the end of
03:06the physical scenario, that's super important.
03:09As what we did for the digital aliens, that the most cost expense is still on the token side.
03:15So, we think in the long run, the people, if they want to upgrade from OTA or deploy more skills
03:21and using more physical tokens to do the manipulation and help them to make the physical world automatically,
03:27we think that will be the very strong growth from the token cards.
03:33You mentioned how important the hands were, right?
03:36The dexterity there.
03:38When it comes to the manufacturing process and being able to scale it up to those hundreds of thousands of
03:43units that you want to put out eventually,
03:46where are the biggest bottlenecks and challenges?
03:52We think the world's biggest bottleneck is not the hardware itself.
03:57As you can see, China and us have a full stack vertical integration supply chain.
04:03And it is very easy for us to get low cost and the fast iteration of the key components.
04:08But the software today, you know, the data is scarce because the physical manipulation is not like the large language
04:17models that we have already saved about 20 years of the Internet of Data to train the language model.
04:23But for the physical manipulation model, it really relies on the people, on the synthetic data.
04:29And also, currently, we're using egocentric and try to immigrate the human behavior to the humanoids and make sure the
04:37matrix hand can perform the, you know, daily life and do some automation or work through their portraying the neural
04:46network.
04:54You talk about how there are so many companies in China and an ecosystem being built around robotics and humanoids
05:05right now.
05:05Does that also mean more competition for you as you have dozens of startups trying to do what you're doing
05:11and how do you keep that competitive edge?
05:17Actually, you know, not only China, but for the global market, it is super crowded.
05:22But we see the consolidation in the next 24 to 36 months.
05:26You know, only the winner will come out from who can have a very competitive cost and deploy their robotics
05:33or the humanoids in the real working scenario,
05:36no matter the commercial to public services or the medical assistant, industrial or even logistics.
05:41So it really depends on if we, if any company can assemble a very good hardware and the software and
05:48also with the physical AI intelligence into the, into the real working scenario.
05:54So, so that will be the Q&A app.
Comments