GMI Cloud’s Alex Yeh on Scaling AI with Cloud Infrastructure

This is Textron tv. Hey, everyone. Welcome back here to techron tv.

I'm really thrilled to have this next guest on here. You know, I, I love introducing our audience to new companies. This is a fairly new company.

Well, it's new to us. Anyway, I want to introduce you to Alex. Yay.

Alex is the CEO and founder of a company called GMI Cloud. That's good. Well, he's gonna explain how, what GMI stands for, maybe.

But Alex, welcome to Tech Drunk tv. It's great to have you on. Thank you, Alan.

Well, um, happy to be here. Yep. So we're gonna get into GMI cloud in just a second, but let's talk a little bit about Alex.

Um, as I said, you're the CEO and founder of, of GMI Cloud. Tell, tell us your story. Yeah, it's definitely, uh, interesting, uh, story for me at least.

Uh, I stumbled upon entrepreneurship and went down the, uh, went down a rabbit hole, uh, I guess one mistake after another. Welcome to the club, man. Exactly, exactly.

So I was, uh, my background, well, I'm an engineering background. I graduated from Hopkins with a materials engineering degree. Mm-Hmm.

Um, I stumbled about cryptocurrency, um, really, uh, about, uh, in 20, 20 17. And I was just, I was just lucky, right? Um, my high school friend showed me something about Bitcoin and then Ethereum, so I bought both.

And then, um, because Ethereum was cheaper than Bitcoin at the time, I didn't know what that is. Yeah. Mm-Hmm.

Uh, and so I bought that, um, and it started rising, right? So I, I thought, wow, this is interesting. And I learned more about it.

That was basically my be my beginning into, into, uh, into the digital asset world. Um, but I also always had this, I would say, uh, traditional finance training, aside from my, uh, engineering path. It took, uh, I guess, I guess double major in business.

I understand under so, but about financial markets. Um, and it depends. I still don't understand what cryptocurrency is about, what it, what it does.

But what I understand is, like you, well, the definition of, you know, companies to create, I would say, uh, shareholder value or return dollars, right? But crypto at the time, I just couldn't understand like, why you creating something out of thin air and then without any value. And so the, the right move for me at the time was basically jumping, jumping into the fix and shovel space, because I think, uh, think about what built America.

It's the infrastructure that builds, right? It's steel, it's the real world, it's the oil, right? That, and also in the two thousands, right?

It's the, uh, IBM, the Cisco, um, the wind, the windows that act, it's sits at operating system or infrastructure for, uh, all of these amazing applications out there. And so I jumped into, I would say Bitcoin mining as a investor, excuse me. So, uh, that's what I did.

And, um, I, I just bought some machines and hosted at my friend's, uh, data center in, in the Mongolia, China, you've never been, okay? It's beautiful place. My, that's my only time there.

Uh, it's to, it's to, uh, as to, as to, uh, you know, see where my machines are running, uh, beautiful place. But I guess in 2020, she doesn't really like Bitcoin all that much, right? Um, so entire China shut down.

It got moved multiple place locations within China. Um, and, but in 20 20, 20 20, May 23rd, it got shut down in one day, zero notice, insane, if you think about it, right? They shut down this entire industry on one day.

And so, uh, I had two choice. Either one, I write down my investment or I send it somewhere else. And so I had plenty of options, but, uh, at the time I was like, all right, uh, US is the market that I understand.

I graduat from Hopkins. It's been, you know, more than eight years at a time, you know, in, in us. So, all right, so I put in one way ticket to, to New York.

I was like, all right, I'm gonna find a location or find data center. How hard could it be? Right?

Again, I, I kept saying that, right? That, that is, that is a mistake that I kept, kept on making. I think it's just like my naive optimism.

Uh, so how hard could it be? And so I was like, all right, I'm gonna call it digital realty. I'm gonna call vantage like data center.

No big deal. And difficult, right? Let me tell you that, uh, 80%, uh, at the time, miners were in China, and so everyone's moving out, right?

Imagine like zero, there's zero infrastructure right now. We're in a gold rush of data center as well. So it's the same thing that will happen in 2020.

So fast forward, it took me, I would say, six months from amda, that's no operating, uh, experience to try to find data center, and there's no results, right? Everyone on wants an arm. Like, and so what I did was like, all right, I'm gonna build it myself.

Another mistake. That was date. Uh, so I did this in 1999.

I want you to go, but go ahead. Yeah. So I was like, all right, uh, I'm gonna build myself, right?

And then I, I got some like, supply chain, uh, connections in China and Taiwan. I remember, I can figure it out. Uh, so in 2021, actually, it's a very difficult time.

I think by the time I was like the most knowledgeable 26-year-old about US power market, I flipped every stone of every state possible that have power on west coast, Washington state, get California expensive power. You guys, you guys should know about it. Texas, Arkansas, Oklahoma, um, uh, to Wisconsin, to Mississippi, to, uh, east Coast, Pennsylvania, north West Virginia, uh, you name it.

Uh, all over, all over the place. I even, I even people even sent me deals about Ukraine. That's pre-war.

And Georgia, the country, not the state. No, no, I get it. Mm-Hmm.

And, and like somewhere in northern Europe, I went to Iceland in middle of December. Don't, don't go there. During that time, no, not in December.

Yeah. So in the end, you know, I knew everything about the power market. I signed a deal in three days.

Imagine I bought a piece of land in three days. You actually bought the land to build the data center on good for, for you. That's bold.

As, as a 26-year-old Asian, Asian guy going into middle of Arkansas and talk to a talk to God, 34-year-old landlord, it's like, I wanna buy your land. I wanna buy your power. I wanna build data center, which is, which is a surreal experience for me.

Um, but I did it in three days. I think it would be a surreal experience for everyone. But go ahead.

So, but, you know, uh, that was, gee, that was accurate. Reviewing hundreds, I would say hundreds, hundreds of term sheets, right? By that time, I already knew everything about the, about this market.

So I was able to make precision very quickly. Day one, I see their data, uh, the data room. I know this is gonna work, but I called the landlord, was like, I'm flying the next day, the third day, basically handshake.

And they also done, because he was a very, very nice Texan person. So he's like, uh, he's big numbers guy. It worked out.

Um, so that was like the beginning, right? And after that supply chain, and it was P Covid, remember P Covid, yeah. Everything shut down.

Uh, I wasn't able to navigate, uh, supply chain, all that. All difficult, difficult, uh, in Taiwan, in China to ship containers, ship transformers, ship a lot of equipments out, um, and dead servers out at the same time. I had my container shipped.

I was tracking on satellite, and, uh, it was sitting outside of Houston port for 20 days. It was sitting out there. You see thousands, hundreds of ships, ships, big container ships, kinda looking in Singapore.

If you ever been to Singapore, it's like that every day, right? Yeah. Mm-Hmm.

Crazy. So yeah, fast forward. So we built site one and beautiful, uh, worked out really well.

Uh, and then, uh, raised venture funding, um, and raised two, actually two rounds after a PC eating seed, and then built out, the first one is the diff most difficult one built besides right? And then about a year ago, our investor, uh, and also founder friends was like, Hey, Alex, actually, I know you're an HPC business, high performance computing business. And like, we, we need a lot of GPUs and, uh, our portfolio needs GPUs.

Were like, we need GPUs. And they're like, okay, well that's interesting. Uh, I can learn more about it.

And then I met, uh, I would say a mentor, uh, now advisor and, uh, investor, which is the, the, the co-founder of Digital Ocean. Um, mm-Hmm. And I was like, oh, yeah.

So I'm thinking about pivoting into GPU, uh, like basically cloud, cloud business. You tell me more about it. And he was like, so excited.

He was like, Alex, you're building the GPU for, to, like, that's a deal, deal for GP, basically. Mm-Hmm. Um, and so if you guys know about Digital Ocean, it's basically, I would say, uh, a neo clouds, um, taking down, I would say giants taking down Amazon, AWS and so forth.

So I really got that inspiration. Um, so the pivot was also very difficult because from a hardware data center into a cloud stack, and I did that, right? I would say in six months.

So October of last year was the first time when I, when I pivoted and I hired an entire software team, I hired a team. So the CTO of Mineral ai, which is part of Google X board of beautiful Google X project, uh, the CEO came out, brought VP engineering, brought director of, of, of a product as well as, uh, uh, LM came out, um, that's a software team. And then hired a very experienced, uh, financial team, finance team from, uh, JP Morgan.

And Morgan Stanley. Spent 12 years there, and he was a, he was a group CFO of, uh, of, uh, one of the largest NASDAQ listed data center company came out. And, uh, our COO, who was one of my first employee, uh, from Super Micro.

And he, and he was like, Alex, I, I was, I was a chapter director or super micro building, H 100, which is the, the, the GPU that everyone's talking about. It's like, mm-Hmm. I think everything about this machine.

And so he, he went to Super Micro and took a lot of his old, I would say, colleagues, uh, to, to join us. So that was really, I would say the, the hardware and the software team was finance, finance team that really built that under six months. And on top of that, we built three data center.

I mean, clusters, GU clusters, um, in Taiwan and also in Thailand within six months, right? Fantastic. And then find close to a dozen customers that wants, uh, our, uh, software as sports hardware.

So, you know, in six months time, we built a software team, hardware team, the finance team, built three data centers, built a product, and got customers. That was a crazy, crazy time. Crazy, crazy.

Yeah. Good for you, Rick. Congratulations.

I didn't have a lot of sleep. So that was basically the entire pivot and how we build this. And you can GMI, right?

I think you asked that question. mm-hmm. GMI stands for General Machine Intelligence.

So, okay, so we're here to build the HR feature. We're built, we're building the foundation, um, for the HR future. And what, what, what that means is basically we wanna build, I would say a Wix or a Squarespace for ai, right?

Right. Now it's so hard to pull AI, need a team of 20, 30 people, or even 50 people to build foundational model. Or you wanna fine tune some product, create a vertical ai, and we wanna make that super easy and seamless, right?

Allowing any individual worker, any founder to build an AI application instantly, right? Allowing anyone to deploy AI instantly is our, is my vision for, for the company. I love it.

I love it. You know, Alex, I see so many parallels. So 30 years ago, 35 years ago, there's no such thing as hosting, there's no cloud per se, right?

We didn't call it cloud. And I did the same thing, started in a friend's data center, went and got our own data center, sold my company to a company that was at Big Morgan Stanley, and, and Chase money behind them. We wound up being actually a, the predecessor to Digital Ocean was called in to, and they were one of our competitors.

com, right? We had data centers in Tysons, Virginia, and in France and in, uh, Atlanta and bunch of other places. 36 acqui, 30 acquisitions in 36 months.

Crazy time. Wow. So, I, I, I, you know, listening to your story gets my juices flowing, but we're in a whole new world there, right?

Because this, the GPU and the AI is where it's out. We're not gonna have time to get into everything. I'd love to get into with here, but we're gonna see you in, in CubeCon.

That's, and we're gonna continue this conversation. 'cause I want to hear about the software, right? I think we all understand the hardware and then the Nvidia and all of that GPU.

But what people don't get is that it's, it's always about the software, right? It's always the software. It's, it's, and we, we wanna jump into that, but a couple things right now just to get out of the way, and we'll pick this up.

You're going to be Q con, GMI cloud, is that Q con? You're at booth Z, like zebra, two Z two, and people can come by That's right. And get more information there.

But basically what you're offering is almost sort of GPU as a service, right? AI is generative AI as a service. Is that fair?

So it's, if, if I give you like a quick, quick, like one minute, uh, the state of the market is, uh, everywhere's just selling the boxes, and I hate that. People are like, oh, we got gpu. And I'm like, so what?

Right? You're here to build a solution. You're not here to buy GPUs.

That is the means to the end. And so what we're trying to build is build a vertical solution. If you wanna build a rag, we got a rag playbook for you.

If you're, if you're building a text to image model or text to video model, we got a playbook for you, right? You, you need from basically optimizing on the, on the model selection to the data procurements, to the data processing et l data, uh, all the way to kind of visualization of how your model's performing to the orchestration of the software for G Fuse. And so wouldn't it be nice to have all of that under one platform, right?

We wanna build a useful tool for anyone to build ai, right? We're not here to just bounce reviews because like, what are you gonna do with that? Right?

And a lot of people, you know what this, this is exactly where I was 30 years ago, right? The web hosting providers were selling what we used to call them power paint and pipe, right? Power ping and pipe.

And that's what they sold you. We were selling application service provider back different apps. Then you had Lotus Notes and PeopleSoft and Oracle.

Yes. But we were selling you the whole hosted application, not just Yes, power ping and pipe into your one you or four U server. That's rights.

It's what's old is new my friend. It never changes. Yeah, it's the same.

Yes, same, same kind of suite. Doesn't, doesn't, doesn't repeat that. But it certainly definitely ranks.

Yeah, it absolutely. Very cool. Hey, what's the website?

It's GMI cloud ai. ai. Love it.

Alex, we're way over that 15 minutes. I probably took 20, 25, but we're going to continue this acute con next week. Absolutely.

Actually, by the time, yeah. Next week people will see this, this week. And, and we'll, we'll be there next week.

Again, for people who wanna visit, you know, booth seven, two Z two, that's Z as in zebra. Two. Also, just real quick, we're doing a breakfast Aon Tuesday morning, Alex, I want to, it, there's no, it's not a sales thing.

Just come have a good breakfast and coffee. If you're there Tuesday, Anton and the rest of your team is all invited, inviting anyone out here who's gonna be there, come on in. You can register, get to it off our website.

Um, Q Con's a week away. Alex, what a great story. I always tell people, you know how I could tell successful entrepreneurs their passion?

If you can't feel their passion about what they're doing and talking to them, they're not gonna be successful. Your passion comes to loud and clear, man. I love it.

Thank you. Thank you, Alan. All righty.

Thanks for having me. Look forward to meet, meet you at coupon. Absolutely.

Alex Yee, CEO of founder of GMI cloud, the future of AI infrastructure and cloud computing right here. We'll speak to him live next week from Q Con. We're gonna take a break here on Text Drug tv.

We'll be right back.