Giga Energy: from Bitcoin mining to $350M revenue data center builder — with just $3.4M raised

Speaker 5: Talk to soon. Update, Eric. Good to see you. Have a good one. Yeah.

Speaker 3: You too.

Speaker 5: Congrats.

Speaker 1: Goodbye. Our next guest is Matt from Giga Energy. He made news recently because Giga Energy has scaled to $350,000,000 in revenue vertically integrating AI infrastructure while raising just $3,400,000 of equity funding. It's a amazing amazing story of what's possible in the AI era in hard tech. People say it's hard. Maybe it's not so hard. I don't know. Matt will tell us. How are doing?

Speaker 4: Howdy. How are doing?

Speaker 5: What's going on?

Speaker 1: Thanks so much for taking the time to jump on the I heard the story, heard a little bit of the news, but I'd love to go back and start with your origin story, how you got into this business, and then walk us through the shape of the business today. So where'd you get started?

Speaker 4: Yeah. So I started this business back at Texas A and M with a couple buddies from college almost seven years ago. Okay. Business kind of started around flared natural gas, bitcoin mining and quickly evolved into supply chain constraints.

Speaker 1: And and then And was your your your team, were there was the back were you guys studying engineering or or were you just interested in crypto? Like, what was the business? I don't know. What was the background?

Speaker 4: Yeah. Yes. I'm just super interested in Bitcoin, like, trying to find any way to get into it. Yeah. I was, like, out of desperation. Large oil and gas college. My co founder's generation oil and gas. Impetus was, hey, can you find us a well and can we mine Bitcoin on it? And then from there, just kind of went on YouTube and and little by little learned three phase power.

Speaker 1: No way. So we what was the first thing you built?

Speaker 4: So the first thing we ever built was a 50 kilowatt

Speaker 1: Okay.

Speaker 4: Modular data center in in Southeast Texas that mines yeah. Just like, I don't know, maybe 4 to bitcoin a month or something.

Speaker 1: Yeah. And and what does that actually look like? So there's a there there's a deposit of natural gas in the ground, and it's being flared and it and some of it would be wasted, but then you're converting it. Like, how much did you have to buy? What what what does what's the am I looking at something that's like a shipping container size or like building size or like washing machine size?

Speaker 4: Like twenty twenty foot ship container. Okay. Like tiny, like nothing nothing massive at at scale. And then, yeah, you have natural gas generator on-site. You're pulling from the wellhead. Okay. Stuff's going down, breaking all the time. And basically, in terms of scaling that business, the rate limiter was infrastructure. So very quickly, we said, how can we build our own Bitcoin boxes? Yeah. And then Bitcoin, 85% drop back. Next thing you know, we're selling more Bitcoin boxes and picks and shovels than we weren't mining the Bitcoin.

Speaker 1: Interesting. And then these Bitcoin boxes, are they GPUs? Are they a $6.04 4 Bitcoins? Like, where were we in the story of, like, Bitcoin mining at that point?

Speaker 4: Yeah. So your 2019, you're still at, 14 nanometer Okay. Chips with ASICs Okay. At time. So very rudimentary in the field, flies and dust everywhere.

Speaker 1: Yeah. Yeah. But they but they aren't repurposable because they're Bitcoin ASICs. So you wind up selling a lot of those as the market deteriorates. When do you start thinking about AI?

Speaker 4: Yeah. So and it's on about 1.2 gigawatts of those modular data centers over, a four year tenure.

Speaker 1: That's a lot.

Speaker 4: Then little by little. Yeah. Yeah.

Speaker 1: We we So so yeah. Wait. Wait. Really quickly. Like, you're selling 1.2 gigawatts of these modular data centers. Like, what is your supply chain like? How vertically integrated are you versus more of like a systems integrator that then is delivering that to a buyer?

Speaker 4: Man, it was incremental. It was everything from like building it at Home Depot to now, you know, restoring manufacturing back in the And then just like little by little over a seven year period, bringing it back.

Speaker 5: Amazing. And and what's the typical what was the typical buyer back in that era?

Speaker 4: Everyone. I mean, we captured I mean, we kinda had a nice niche, like lower market, middle market, public markets. You name it, we're selling to them. We're the guys with the Bitcoin boxes.

Speaker 1: Okay. So so But what what we so one of these companies that wants to they they they purely see it as like a economic opportunity. They're putting dollars in and getting Bitcoin out, and they think that the ROI will go up as they they have a Bitcoin price target. They come to you, buy it, and then they also have to go and get the land that has the access to the natural gas line, basically?

Speaker 4: Yep. Yeah. And and also on grid utility. Okay. And and so, yeah, kinda like, basically, we we got a really good mousetrap. We're selling it, making great margin. We realized everyone we're selling them to, why can't we do what they're doing if we got, you know, great margins?

Speaker 1: Sure.

Speaker 4: So I started going down the pipeline, figuring out how do we get powered land, how do we get cheap power. Yep. In the same vein, started introducing new product lines, so transformers and switchgear in '24, we introduced. Yeah. Started selling that in in in at Nauseam and Nen. Yeah. Mid twenty four, we built a go to market function around C and I, renewables, electrical, and just started selling to non Bitcoin miners, realizing that our electrical equipment was was fungible.

Speaker 1: Interesting. And so what was the scale of the company in '24 when you start doing that? I mean, you have like hundreds of employees?

Speaker 5: Wait. And when did you when did you raise that Yeah. Few million bucks?

Speaker 1: Just through the fundraising.

Speaker 4: Yeah. No. So that that the funny thing, that 3,400,000 didn't really, you know, go to any of our growth. They just like kinda kept us alive for a couple of year period figuring out, you know, our product market fit. Mhmm. So we raised like a million bucks in 2021 Mhmm. And we raised the remaining amount in March '22.

Speaker 1: Wow. And then just and then just sales come in.

Speaker 4: Yeah. It's profitable business.

Speaker 1: Yeah. Was there was it were there any like

Speaker 5: Did you get thrown out of any VC office for being profitable at any point?

Speaker 3: Yeah. Wasn't in them.

Speaker 1: Sounds like it didn't even stop by. But in in in terms of the cash conversion cycle, did you ever have to use debt or payment terms? Because I imagine if somebody comes to you and says, like, I want a Bitcoin box, you still have to go make it. There's like supply chain. Like, the like, your work your your your your working capital could explode if you're not careful.

Speaker 4: It was all payment terms. So we we were able to convince customers to use thirty, forty, 50% down and then match our payment milestones relative to cash out the door. And and just kind of continue rolling forward. Yeah, our customers finance the whole business.

Speaker 1: Okay. So then take me through, like, the growth of the actual workforce, like, much is in in house? How how big is the the team at at various points? Like, how did you scale the workforce?

Speaker 4: Yeah. So we had about 90 employees end of last year approximately, maybe a 100. We're up to about 200 now Yeah. This year by the end of the quarter. Really kind of continuing to continuing to scale it throughout '24. We got into renewables, C and I, electrical division. Yep. Selling to churches, prisons, hospitals, transformers, switchgear systems, and then ended up getting a massive barfque with an AI data center, and that's like just kind of led us back to the watering hole in terms of, you know, how how big the opportunity was. Next thing you know, we're selling transformer switchgear to everyone in the AI space.

Speaker 1: That's amazing. Where do you want this to go? Are you going to raise money at some point, or do you want us to continue to grow it profitably?

Speaker 4: Yeah. So Giga has now kind of again, we went through the cycle of like, we're selling the picks and shovels. Yeah. Wait. The guys who were selling the picks and shovels to are making a ton of money. Why can't we do that? Mhmm. Same thing's happening again on the AI side. We got the powered land. Have massive pipeline, multi gigawatts in terms of our capacity.

Speaker 1: Yeah.

Speaker 4: We have we have healthy margins, and so Giga today really exists around rapid data center deployment.

Speaker 1: Okay.

Speaker 4: We build all the main all we manufacture all the long lead time equipment items build the data center, and we build the data center with our own picks and

Speaker 1: shovels. Okay.

Speaker 5: And What kind of what what kind of challenges are you facing on on the ground? Specifically with like local Yeah. Where's your community opposition?

Speaker 4: Yeah. There's a lot of that. Yeah. We got town hall meetings all the time talking to people and kind of walking them through, know, hey, this is gonna use less water than than your golf course does with an area. There's lot of noise concerns. So a lot of this is kind of sitting down with the residents walking through what does this practically look like in your backyard.

Speaker 5: On the noise front, what's what do you believe what do you believe is the solution there? I think there's every single day there's videos on x circulating where you could Yeah. Posting like, you know, this is what it sounds like if you're living in this, you know, suburb and it's like a low from the generators. You've you've been continuously making new hardware, you know, innovations and products. Do you think there's a hardware solution to noise or is it just put them farther away from people?

Speaker 4: Yeah. I mean, the unfortunate thing is like almost all the setups are in compliant with local laws. Right? They're probably going be below that 65 decibel threshold, but that kind of low is still very irritating. The extent that, you know, you can kind of continue to to innovate around the products and get lower decibels, put up sound walls, dirt mounds further away, but also, I mean, in my opinion, it's a bit of a easy thing to latch onto when just, you know, opposing the general concept of an AI data centering record.

Speaker 9: Oh, sure. So it's sort of like

Speaker 1: a straw man. Like, that's not what people actually care about. They care about the bigger picture, bigger question.

Speaker 5: Why don't they dig putting them underground? Is that a is that a solution? I've I've seen a bunch of people posting like very creative ideas around these things. It sounds like a lot of work to dig a big enough hole in the ground, but maybe it's

Speaker 6: Yeah.

Speaker 4: I mean, on top of trying to build a data center nine months, putting one in the ground is is not on on the highest priority, I would say.

Speaker 1: Yeah. Just too too too fast. Well, what about energy for local, like local energy prices? There's been the rate payer production pledge signed by a lot of hyperscalers. Like, how how real is that? How are you working through those those those talking points when they come up at a town hall meeting?

Speaker 4: Yeah. Right. Every utility or electrical tariff relative to a municipality can be different. Even within our card, right, you have different center point or encore. Right? Two different systems, two different ways of calculating that method. There is the ability to ensure that consumers don't receive increased pricing relative to that. And there's a lot of education in terms around what soft load or front load looks like. You get backup diesel gensets on on-site, but also at the end of day, gotta be pulling, you know, four four five nines of uptime. So the LMP pricing or local marginal pricing of these sites probably will go up a marginal amount unless the hyperscalers are willing to

Speaker 1: step in. Yeah. How far do you wanna go up the stack? You're building actual data centers now. Are you going to be the one racking the GPUs, or are you going to partner with Neo Clouds? Are you going to be, you know, networking everything together and offering an API for a particular LLM? Like, you could go very, very deep in the integration if you wanted to, but what's on the near term goal set?

Speaker 4: Yeah. I think near term. Right? Gays are very much a ground up company.

Speaker 1: Yeah.

Speaker 4: Right? Manufacturing the stuff from the ground up with our own CAD file sets. Yeah. Moving the dirt, putting up the building, and energizing the facility. We stop at the rack level currently, and and that's really what we're excellent. Very much the top down approach of heavy financial systems or, you know, Yeah. Whatnot. We we view our our bottom up approach very novel.

Speaker 1: So I I mean, do you do you is there is there a different financial like like underwriting procedure or thought process around actually acquiring GPUs and racking them because it sounds like you would have a tenant that brings in their own equipment and Yeah. Runs whatever they

Speaker 4: Yeah. Exactly. Right? So for us, value proposition is, hey. You need to burn these chips. We can get you fifty, hundred megawatts of IT online in nine months.

Speaker 1: Yeah. And we've seen that from Sadeghian Alexander at Microsoft. He says, we actually have the chips. We just don't have enough powered shells. That's why you're in the powered shell business. That makes a lot of sense. Fascinating. What else goes into delivering a data center in nine months? Feels like is there a secret to not getting hung up in permitting? Is it the fact that you have a more fluid workforce that can travel across the country? We've heard stories of like plumbers going on private jets. I don't know what else you can tell me about like bringing your 200 person employee base to bear in a particular region that might not be right next to your headquarters.

Speaker 4: Yeah. So our mantra is building in the factory, not in the field.

Speaker 1: Okay.

Speaker 4: And we did this kind of implicitly in the Bitcoin mining world. We were able to turn on Bitcoin sites in almost sixty days Yeah. From bare pieces of dirt to energized racks. Yeah. In the AI space, nine months is done because we do majority of the commissioning and the integration at the factory in a controlled environment.

Speaker 1: Mhmm.

Speaker 4: We've got a 95% reduction in the the need for on-site labor at these data centers because of our modular design and the prefabrication going in. And that's really much the focus upfront.

Speaker 1: Tents versus buildings. We heard that Meta recently shifted to tents to speed up the time to deploy their data centers. Do you have a opinion on the structure that should hold the the GPUs, the racks?

Speaker 4: The whole industry should be focused on time to token. Right? What what gets us online as fast as possible? Whether that's a tent, whether that's a pre pre engineered metal building on-site. If all, you know, 15 different companies could be aligned on time to token, that's really what we need right now.

Speaker 5: What city has been the most welcoming to data centers in America that you've come across yet? Like any any anybody standing up the whole town just just saying You

Speaker 4: know, I mean, it's hard to make everyone happy, I would say. The I couldn't name a particular place that I would say is a 100% on board.

Speaker 1: Are you worried about any of the potential regulatory or political shifts in the next twelve months?

Speaker 4: I mean, I have to imagine at some point, this becomes a federal issue. Yeah. Like, where, you know, sign sign an order, you know, saying, you know like, there's no reason, like, we should have random municipalities blocking, like, AI inference coming online. It's just, like, not in any way shape or form in terms

Speaker 1: of what

Speaker 4: we're trying to march forward. So I think that's gotta come at some point. How soon? I I don't know.

Speaker 5: Yeah. How are you thinking about space? Feels like Giga has been very adaptable to date around different trends. Could we see you guys building satellites in the future? Data centers on board?

Speaker 4: You know, we're we're we're focused on the dirt. We're focused on we're we're we're not the most high-tech. We're not the largest. We're very much focused on on speed. And at the end the day, I think that that is is is the difference.