In The Arena by TechArena - Giga Computing on Rack-Scale AI: From Training to Inference

Episode Date: January 21, 2026

Recorded at #OCPSummit25, Allyson Klein and Jeniece Wnorowski sit down with Giga Computing’s Chen Lee to unpack GIGAPOD and GPM, DLC/immersion cooling, regional assembly, and the pivot to inference.

Welcome to Tech Arena, featuring authentic discussions between tech's leading innovators and our host, Alison Klein. Now, let's step into the arena. Welcome in the arena. My name's Allison Klein. We're coming to you from OCP Summit in San Jose, California. And it's a Data Insights episode, which means I'm with Denise Rowski. Welcome, Janice. How are you doing? Thank you, Allison. Thanks for having me. It's day two of recording at OCP. Summit and it's been a wild ride. I am really excited for the guests that you brought with you today.

Tell me who we thought with us and what the topic is for today. Yes, I am always excited to get to sit down with Mr. Chen. Oh, thank you so much. Do it too caught. Oh, so today we're going to be speaking with Chen Lee from Gigabyte and talking about all things AI infrastructure. And Gigabyte comes up a lot with a lot of our partners and customers. It is a preferred solution preferred platform, and we're just excited to kind of unpack it a little bit and talk about the innovation. You guys are trailblazing.

Yes. So, Chen, it's been a while since you've been on. I think we talked to last year at OCP Summit, and Giggy Computing has evolved dramatically. We have. Can you walk us through that transformation and how the company appears in the market today? Mr. Ye have decided to create a networking division,

and that division grew to what is get computing today. I've known with the company in 19 years, and I have been seeing growth every year, but nothing, nothing, like what I've seen in the past three years, four years. The growth is phenomenal, of course, thanks to AI. And we have a big thanks to open AI to even opening that flick. And now everybody's in the game. And although it's exciting, but it's also very tough business now.

Yeah, yeah, yeah. Yeah. So you're right. All things are moving at a blistering rate. Right, right? It's so fast, sometimes it's scary. Very scary. So how do you mitigate this with your customers?

How do you help them thrive in this market? So what we do is that we try to provide our customer with the best possible service and then the delivery and let the customer know that we not only want to deliver the product to the customer. We want to make sure that after service and technical support and so on so forth are all there. If something goes wrong, we are there to mitigate the issues. We don't want to leave anybody standing. And to help the customer grow even further, now we have built up assembly areas around the world, right? Here in the U.S. are 100,000 square foot integrations that are coming in. We can do DLC, immersion, air cool.

And then we have a facility in Malaysia, which Southeast Asia are a growing market right now. So we're putting in a facility in Southeast Asia, and we're building up in Abu Dhabi as well. And then, of course, in Europe and also in Asia, right? So we have Taiwan and China as a strategic point. You know, I remember for the last interview that you talked about, the fact that you go play with the hypers from the beginning. And one of the things that I saw at OCP Dublin that I was really impressed with was your Gigapod and GPM solutions. Can you walk us through that in terms of how they help folks deploy at scale?

Yeah, so Gigapot is more of a modular and it's flexible. We're not locked down to a certain particular brand. So if a customer wants A brand or M brand or any of these type of a server, we can. can supply them with it. And with that, we have our GIPM, the GitHubPat management, and that will bring a lot of management, a lot of the serviceability to the rack, making the deployment a lot easier. What we want to do is make the customer feel comfortable when we roll the rack in. And they don't have to worry about doing this or that. We are there to help the customer and to get the system onto their premise, get it run it. That's fantastic. Well, we're hearing a lot about liquid cooling,

right? It's a hot topic, no pun intended. Yes. But what is Giga doing? around liquid-cooled solutions. We do have a lot of liquid-coil, right? Yeah, yeah. So like I was just saying, that we have a facility in the U.S. here. We're building the DLC, immersion, and so on and so forth.

And what we're doing for cooling, one thing, it involves green, right? We want to make sure that we sustain the earth for everybody that's come after us. Yeah. But most importantly is that in order for all these high-tower system to run optimally,

we have to make sure that it's pool to an optimum level where it can perform. Right. So we are focused heavily on DLC and also immersion. We are building server that are 95% covered. So literally your CPU, your GPU, your memory, our drive, everything is covered because that way we literally eliminate all the fans,

right, and that's a pretty heavy source power draw. So we go to liquid cooling, one, perform and provide a better cooling solution. Two, it will help eliminate some of this high-power usage. Now we're looking at this GPU at 600, 800-Watt, right? Just one, now you've got eight of those in there, plus the CPU, the memories, and so on and so forth. Holy cow, can you imagine how much one rag is running, and now if you have these hyperscalor, they're buying thousands of rad.

So another thing that might be interesting to look into, and we're also stunning to cooperate with possibly nuclear. in order to provide that. But we're not doing it. We're not doing it. I was like, wow, that's a big statement. We're studying the possibility how to work with these type of company. You know, you do all of this for your customers.

And I think that one question that I had for you is these Gengupod solutions, can you give us some examples of how your customers have benefited from either driving more compute density or operational efficiency? What really stands out to you when it comes to customer success? I think the biggest success that we do is on technology front. We have talked to many customers, and then they have always said that Gigabyte had the technology know-how to deliver what they need. Not only we deliver a turnkey, a cookie-cutter rack, we can deliver customized rack, customized to the customer's particular needs, right?

Because sometimes when you buy a rack for it sort of someone, it is built exactly this way. There's no change, no nothing, no going about it. But we're trying to make sure that the customer can get the type of a performance, and the budget that it needs in order for them to be more efficient and in order to make money. We're all here to do one through, make more money. Yeah, always always. And we'll try to help the customer make more money. I love that.

I love that. Yeah. Speaking about making more money, you're also very big on making things more efficient, right, and sustainability. So tell us a little bit about your B-343 facility, which is manufactured made up of recycled materials. It's very energy efficient. Can you comment on? Some of that stuff is still.

in the works. But we are trying to recycle a lot of the stuff that we consider waste in the past. We can reuse a lot of components, a lot of the maybe sheet metal, cabling, all that stuff. We try to make sure that we are eliminating waste and we are able to take what we waste

and maybe reuse up. Because resource for us on this earth are limited. And we really need to make sure that we provide something for our future. So green is important to us and making efficient, also eliminate any waste. I'm going to assume it has some cost advantages too to bring some of that stuff into it secondly. Is that true? Yes and no. Yes and no. Because yes, we're not eliminating waste and we're bringing it back in. But now we also have a lot of manpower to sort it out and figure out that stuff out. So yes, it does help a little bit as far as cost concerned.

But mostly is that we're answering the call for Earth. We try to make sure that we provide something, like I said, for the future. Now, let's look ahead for a second. I know that 2025 has been an amazing year for you. Yes. When you think about what customers mean, it's never enough. So what are you doing with your rack scale roadmap to deliver new capabilities to your customers? So today we have ability to deliver racks.

But one thing is important is that training is where all the money is at right now. But inferencing is the future. Yeah. Because building racks to be able to fulfill inferencing business is becoming more and more prevalent to us. Yeah. We do see that with inferencing, we're opening up even more market, even more opportunities. Instead of being locked down to a certain model, now we can expand that and find even more opportunity and capability to support,

just IT field, right? Now we have medical, finance, oil and gas, even military government, education. It could all use what AI is bringing to us. We're on the cost of a revolution that is all saying, okay, 1800 will have the industrial revolution. It changed us human. Yeah. This AI revolution is going to blow away the industrial revolution. Our human life will change forever because of this. I agree. But we have to be careful. the road that would go down. And that being said, I think our focus on AI or focus on providing the solution, the modularity for a customer to be able to provide a better service to mankind,

I think that's most important. Yeah, just to switch gears a little bit to Chen. I'm sorry. I might have gone off course. I love it. We could talk to you all day and go all kinds of places. But how does your investment in the U.S. Assembly reflect kind of a broader regional support And how does that support the supply chain AI era?

Having a facility locally, it gives us the capability to provide our customer with better than cut. So if there's upticking sales and so on and so forth, our facilities there to build to provide the customer. Not only doing that, we're now looking at SKD so we can have assemble America servers. We are bringing the facility into U.S. So instead of shipping entire route from Asia and the U.S. shaking around on a boat or even fly on a plane. One, it cut down more carbon footprint. Sure.

Two, it cut down some of the shipping costs. And three, we actually can do better assembly, better testing locally because we don't have to worry about all that shaking movement as it comes into the port. So a local facility, it's paramount to any, not just us. I mean, I don't care. My competitor, it needs to be done. Otherwise, we won't be able to provide the service we need to the customer. market that requires it.

I love that. And I was a huge fan of gigac computing. I'm so glad that we spent some time with you today. Thank you. For those of the folks who are listening in, where will you send them to learn about giga solutions and engage with your team? So www.gigabyte.com will take you to the gigac computing website. And right there, there are email links and phone numbers and so forth.

You can reach out. And of course, you know, if you're on Lincoln, look me up, Chen Lee. I'm always there to serve. Awesome. Jen, thank you so much for being here today. Thank you. It's been a pleasure. My pleasure.

Thank you so much. And, Janice, that wraps another episode of Data Insights. Thank you so much for the partnership. Thank you so much for having us. And thanks for the time. Thank you for both of you. Thanks for joining Tech Arena. Subscribe and engage at our website, Techorina.

All content is copyright by Techorina.