00:00So, I think the major theme this year is the advancements of agentic AI.
00:04We heard, obviously, Jensen Huang talk a lot about that.
00:08Cristiano Amon of Qualcomm talked a lot about agentic AI.
00:12How much is that changing the way you are employing edge computing
00:17and also Internet of Things devices in the industrial space?
00:22Yeah, thank you, Stephen.
00:23So, I think we have a very long partnership with NVIDIA since 5 or 6 years ago.
00:28And we covered with NVIDIA from across AGI platform, from NVIDIA, like IGX and Jackson to MGS,
00:36also on RTS Pro, and, of course, on their software on the universe.
00:40So, recently, we're working with NVIDIA with their, they call FOX, factory operation blueprint,
00:49and also with their Nemo car, and together with Advanti-wide software.
00:53So, we deliver a series of the industrial agent.
00:56So, AI agent, that's what we call agentic AI.
01:00That will help our customer, especially in our factory, we can build an AI factory brand.
01:06So, AI factory brand is actually the concept that was agentic AI, that was a super agent,
01:12that can orchestrate and coordinate different agents within the factory on the workflow
01:17to make it end-to-end improve the efficiency.
01:19Now, edge computing, from my understanding of it, and now they're calling it AI edge,
01:24is essentially to have the solutions assessed and implemented on the floor, locally.
01:32It doesn't go up to the cloud.
01:33You don't have to wait for the determinations from the cloud.
01:40And it's more immediate and it's more private, right?
01:42Right, you are right.
01:44Actually, a lot of applications at the edge talk about healthcare, or manufacturing, transportation,
01:50or infrastructure and energy.
01:52Everything needs to be processed real-time at the edge for the low latency.
01:56For those that implement, it will call the edge AI.
02:00In previous, I talked about AI factory brand.
02:02That's agentic AI.
02:03Right.
02:03But a lot of stuff we talk about is physical AI.
02:06If you talk about robotics, and if you talk about hyper-visionary and transportation info,
02:12those decisions need to be made at the region, I mean, on prime.
02:16So, that's why we call it edge AI.
02:18I think physical AI is the one also in addition to agentic AI will be booming in the upcoming month,
02:23not year.
02:24Yeah, so how many of your clients are actually now using that kind of agentic AI through edge computing?
02:32I know you mentioned NVIDIA, NemoClaw, agentic AI.
02:36So, you're doing that through them.
02:38So, how many, what percentage would you say by the end of this year of your clients are using these
02:42kind of solutions?
02:43Yeah, I think so far at the edge, most of the customer, no matter from industry, manufacturing, healthcare, or infrastructure
02:50-wise,
02:51maybe around right now 20 to 25%, they are using AI technology because they know the labor shortage.
02:57A lot of things need to be autonomous or production need to be efficiency.
03:01But we see, because the technology of the NemoClaw and physical AI is around the corner,
03:07we expect by the end of the year, I think the increase may be from 20 to 30% because
03:12the AI technology is there.
03:14And the other way to accelerate is that we're working with NVIDIA closely on three computer frameworks.
03:21That means three computer frameworks that are simulation, training, and inference.
03:25Adventai is one of the few partners that are able to provide the total solution on infrastructure-wise.
03:31By doing this, that seems to be real, real to seem.
03:34So, why does that mean that you can do your AI model in the development environment?
03:39Make sure simulation, training is fine, then you can deploy it to the edge.
03:43That's called inference.
03:45In the edge, that will be everything on prime.
03:47And that will be one advantage working together.
03:50NVIDIA equals them together to bring the AI down to the edge for physical AI.
03:54I mean, you're using NVIDIA in a lot of areas.
03:57I mean, you're in about 20 different areas of robotics, including humanoids as well.
04:02I saw Aeolus Robotics is another partnership that you've announced recently.
04:07You have the Jetson AGX Orin chip from NVIDIA doing that.
04:11So, what's your outlook?
04:12And when is humanoid robots going to become less hype and more a reality?
04:19I would say human robot, I would say in a different way.
04:23Because most in the factory world, it's not necessarily look at human.
04:27Right.
04:27They're not that great with dexterity.
04:28They're not that great like human, right?
04:30But generally, what we call is the human robot with wheel.
04:33Yeah.
04:34Then the feet, right?
04:35And with that...
04:36Cobots.
04:36Cobots.
04:37They say cobot.
04:37And some people call...
04:38Collaborative robots.
04:39Yeah.
04:39Some people call it NMR, manipulator mobile robot.
04:42I think that is crucial for the factory flow.
04:45And now, because this is now on the ecosystem partnership.
04:49With the brand, with NVIDIA together, with that partnership, that adventail,
04:53we also need a lot of ecosystem with ARM, with the motion.
04:56So, I think the tech country is moving very fast.
04:59Where are you seeing supply crunches right now?
05:01It's been a theme as well.
05:02I mean, Jensen Huang has talked about it.
05:04Memory has had some, you know, I guess, tight supply in the high bandwidth memory.
05:10That's not necessarily what the kind of memory you use on a regular basis.
05:14But where are you finding those constraints most acute?
05:17Right.
05:18Indeed, in age, we are not using HBM.
05:21That's more on a hyperscaler.
05:22But because HBM, I think, indeed, memory is one of the supply shortages that we are facing.
05:27Start from the end of last year.
05:30Now, I would say it's still the challenge for us.
05:32But now we face another big challenge.
05:34It's about NEM flash.
05:36Because everything...
05:37NEM flash.
05:37NEM flash.
05:38Yeah.
05:38Yeah, for SSD.
05:39So, I think NEM flash will be the one, even together, of course, with memory,
05:44maybe in the upcoming two or three years, it still remains a challenge for everyone.
05:48Not only on the hyperscaler, also everything on the edge.
05:51Where is the priority going to be this year as far as hiring?
05:55Or hopefully not retrenchment, but hiring when your stock price is up 90% already year-to-date.
06:02You just had in the first quarter top line and bottom line at records.
06:07So, how does that going to translate to growth?
06:11Right.
06:13Back to you on the most important thing.
06:15I think people is important.
06:16So, we keep hiring everything in Taiwan globally because we still need...
06:20People is our core foundation for organization.
06:24And regarding the growth is that even though we suffer a lot on the supply chain for memory
06:28and SSD, we actually will benefit a lot from AI.
06:32Because AI infrastructure on the data center, on all the smart grid and automation and those
06:39power monitoring, that's where IoT device is coming from.
06:42That's where the benefit is from energy on those IoT devices.
06:46The other part, we also benefit from AI is on the semi-contour.
06:50So, we suffer because of AI for the supply chain, memory SSD.
06:53The other hand, we have benefit a lot of opportunity through the AI train.
06:58What sectors do you want to expand into?
07:00Because you're three.
07:01You just mentioned two of them.
07:02Semiconductors and medical, but transport as well, and smart cities, right?
07:07Yes.
07:07So, where do you really want to focus, say, in five years?
07:09If this trend, because what we're hearing is the red-hot AI boom is going to continue
07:13from other people we've talked to for the next three to five years.
07:16So, what sectors do you want to be in?
07:18Okay.
07:19Because AI will be driving the incremental opportunity for everyone, including industry.
07:24So, we are working on this smart manufacturing.
07:27So, physical AI robot is definitely one of them we will focus on.
07:31But in addition to that, I think everything regarding autonomous machine, intelligent machine,
07:36that's where the benefit will grow.
07:38So, you're going to still hire, though?
07:40You're not going to retrench?
07:42No, we definitely will hiring.
07:43We are actually hiring really aggressively in Taiwan, because you know in Taiwan, so many
07:48good companies here.
07:49Everyone is craving for talent.
07:51You're not worried about a bubble in AI?
07:55I think, as I said that, right now, it's only like 20%.
07:58To 25%, maximum 25% of our customers, adult from the age, normal age to the AI age.
08:03So, we still have the long way opportunity to grow.
08:07So, I don't think there's a really like bubble at the age side, because it's very fragmented.
08:12Healthcare, transportation, smart wheel, retail, smart manufacturing, and machine.
08:18So, very fragmented.
08:19We are not only focused on one market, we focus on very many industrial-embedding markets.
Comments