JohnPaul: Welcome to the Digital Gold Podcast. Today we’re joined by Taylor Tower, CEO of Solomon Data Systems in a driving force in the next generation data center development. From scaling Merv Marketplace to $38 million in revenue in under a year to leading 250 megawatt in site sales and teaching mining seminars to launch dozens of facilities.

Taylor has lived at the intersection of Bitcoin mining infrastructure and power generation. We’ll dig into his journey, the future of AI ready data centers, and what it takes to execute fast in one of the most competitive industries on the planet. We also have another surprise guest, bill Schneider. He has been a sourcing and contract manager for SME with 25 years of experience across defense, aerospace, consumer goods and data centers.

He’s led billion dollar negotiations, managed complex PPAs and EPC defaults, and recently secured major metadata center contracts worldwide, based in Dallas. He advocates for reliable energy supply and supplier engagement outside of the traditional models.

[00:01:00] Welcome to the show guys, and it’s great to have you.

Taylor: Thanks, JohnPaul.

JohnPaul: So the first question we’re gonna get into Taylor is what made you think Bitcoin mining was the profession for you? And what would you tell yourself today or back then now what today? is it crazy to go into this space? Were you just that young, ambitious guy that thought he could take over the world?

And did you succeed in that vision?

Taylor: , , a combination of all the above, so to speak. I would say , that I definitely succeeded in the vision and , I’d do it again if I had the chance because, , operating in in an emergent industry as a young person, is one of the quickest ways that you can get to, , levels of seniority and managing a size and complexity of project that, , it might take you decades to reach in other industries.

And because it is an emergent industry that has so many young people in it, it definitely propelled my professional journey.

JohnPaul: And it’s one without the corporate ladder per se, where you’re able to jump in, set your own mark, and it’s based on the deals you [00:02:00] close, the power you find less of. Did you work for 2000 hours for this guy and did you fill this role? So talk to me about the first deal. How did it come about and what did you end up doing?

Taylor: , The first deal. So, I quit my day job operating, a wastewater treatment plant, robotics, and large ammonia coolers in a cream cheese factory, actually, , just on a whim because. I’d been day trading Bitcoin, largely over the course of the past year. And the first time I got exposed to Bitcoin mining and Bitcoin was in 2011 or 2012, and you couldn’t really find much on it about the, on the internet around then.

But, \ when I saw COVID hit, I just knew that there was gonna be inflation coming. And my experience with trading had caused me to dive a lot deeper into how the core protocol actually works and how it affects the world from a macroeconomic and sociopolitical, viewpoint. , And I was just , dude, I’ve gotta get , into this.

So I quit my day job I, and strike out. , But that , and this wasn’t one of those stories where, you work on your side hustle for three years [00:03:00] until it becomes large enough to support you, and then you leave your job in, , in a very safe way. I quit, I struck out, , going to my first conference, is actually , what sparked a lot of that for me, which was Mining Disrupt in Miami 2020.

And that just really allowed me to be around, -minded people, see that it was a real thing and do some networking in the space. But, , for me being the technical, analytical mind that it was, , I just bumped into an operator, had a long conversation with ’em, and shared some of, the technical computer and, , math skills that I have.

And , the first deal that I ever closed was actually, a consulting contract to help , an up and coming bitcoin miner, analyze the rate tariffs for a site that they were thinking about acquiring. , And that’s, yeah,, that’s what kicked it all off for me. Just a small few month long consulting agreement.

JohnPaul: Nice. So having no experience in the space, but having the passion, the drive, and the will to learn kicked it off to say, yeah, I can read a tariff sheet, , I can analyze this market based rate and tell you what your [00:04:00] power is gonna cost. , you mentioned you’re pretty young at that time.

What gave you the confidence to just step in both feet in and say, I’m the guy to do this.

Taylor: well, there’s a couple layers to that. I do come from a pretty entrepreneurial family, , designing and, , repairing manufacturing facilities when they’re, their onsite staff couldn’t do it. They call , my father and my grandfather to come solve problems. So there was a little bit of that.

But, , I’d say what most of it is when you’re gonna strike out on your own. You just have to do the work first, right? Before you can, , sell something to somebody , or get a job. And what by that is., I’d wanted to own and operate my own Bitcoin mines, so I went to all the local utilities where I’m from.

I started reading the rate sheets and putting spreadsheets together to figure out, how much is the electricity actually gonna cost. And having done that work for the six months and 12 months prior to that is what put me in a position to say, Hey, I just did this 20 times for myself. Yes, I can help you with this.

And , those 20 times for myself were unpaid, but they gave me the opportunity to, , be paid by somebody for that service.

JohnPaul: And to your [00:05:00] point, it wasn’t your first time looking at it, but you have the grit and the, , ability to go and look at those deals previously. What would you give someone in high school listening to this podcast? What advice do you have for them to jump into the bitcoin mining industry or really any industry knowing that you’ve been an entrepreneur for many years, some people are scared, , they have maybe some ongoing costs that they’re dealing with.

, What type of advice do you give that person to do what they love, and to build a name for themselves in this industry or any industry?

Taylor: I’d say one, read everything you can to connect with people in the space that you want to go into. \ Above and beyond those things, you’re never gonna be ready. It’s never gonna be the perfect time. Start now and learn as you go.

JohnPaul: And so when you got the call to deploy 16,000 asics in 30 days where you’re ready to start now, and talk to me more about that. , that is something that most people can never say they’ve done their life. Because that’s just , that’s what 45 megawatts of more basics.

Taylor: I was terrified when I signed my contract. , [00:06:00] Honestly, , , it was to, , receive disassemble clean test catalog. Sell 16,000 asic in 30 days. And, , we’d heard about the opportunity for , this lot of servers that was for sale. And , there happened to be a conference coming up, this was actually mining disrupt, 2022, I believe.

, At the conference , we’re being with all of our contacts,, shaking hands, doing the thing, and, , we had the opportunity and then we went out to the market to see if we thought that we could execute. And we did. So we negotiated the contract. Signed and went for it.

JohnPaul: And so explain to me that process, , , that’s a ton of servers testing them, maybe not necessarily racking and turning them on, but seems de racking , and preparing them for a sale.

Taylor: Yeah. So the, location of the lot was in Montana. And , we just figured if we’re gonna execute this thing, we need to be on present. We need to be local. Right. So myself and the partners that we executed the contract with, , as soon as we got back from the conference, we negotiated the contract over that weekend, signed it on a Sunday night, and Monday [00:07:00] morning we were in our cars on the way to Montana, rented an Airbnb, did the startup thing, , went and found a local warehouse, where we could actually receive and process the servers.

, So we rented the warehouse, flew up a team of 15 people, , stood up, , a testing facility, which was supposed to be, , a, a hundred kilowatt, , diesel powered generator. And it actually ended up being , a dozen little Generac units that we MacGyver a way to, to plug the Asics into.

, So the, servers would come in on the trucks, we’d take ’em off with a forklift, un palletize ’em, plug ’em into the generators, test ’em. If they tested good, they went in one pile. If they tested bad, they went in another pile. Everybody had a drill in their hand taking ’em apart, blowing ’em off , with an air compressor, , scanning in all the dashboards, resembling them, and then re palletizing ’em , to be ready to ship out to our customers.

JohnPaul: So were they already sold? Did you guys pre-sell them? do you have to pay for them in advance or were they just being sold in as is? Talk to me more about the financials there. ’cause it seems you went from not [00:08:00] zero, but zero to 15,000 is a huge jump when it comes to deploying a testing business and sale business of used equipment.

Taylor: Yeah, so we did have to pre-sell ’em and that was actually a gut wrenching part of the process because they required us to have pre-sold and funded, , two thirds of the contract size prior to receiving any of the Asics. , So we’d already , pre-sold some larger lots and we were waiting , for our big customer to close who was gonna do that.

And man, that week, I definitely got some gray hairs, but , when there’s a will, there’s a way,

JohnPaul: And, you guys are working , what, four hours a day drinking margaritas? , talk to me about this work schedule. Is it 18, 20 hours? That’s a day with sleeping on the floor, Elon style.

Taylor: it wasn’t quite sleeping on the floor before we got the warehouse. It was up at 6:00 AM off the phone at midnight every day for three weeks straight. , And then as soon as we got the warehouse, yeah, it was 14 hour days, for a month straight. Just from picking the Asics up out of the Gaylord and then putting them onto the table or the palette, by the end of that [00:09:00] month everybody there was ripped.

JohnPaul: Well, so it sounds you’re ripped, you’re getting the job done, but Taylor , did you hit your head , I can go work a nine to five, get my benefits. Why would I ever want to go do this? Why did you decide this is for me?

Taylor: I think one of the most important aspects of it for me is the fulfillment and the sense of self sovereignty and capability that you get from it, really owning your own destiny. , But then, , being a do it yourselfer too. But then above and beyond that,, there’s a, being an entrepreneur definitely isn’t for everyone, right?

There’s , the saying goes, your first seven businesses fail., But of course, once do get to the point of, of being successful, , , there’s a reward at the end of the tunnel. , it’s, it’s the only way that access to the American dream still really exists unless you’re in a high paying sales job and managed to invest your money well and save it over a long period of time.

, When you wanna buy your mom a beach house and retire the rest of your bloodline, you gotta be a business owner.

JohnPaul: So, what’s the driving vision for you? I’m hearing is buy the, on the beach [00:10:00] house and then retire my bloodline. , , , that kicks off after all this work. You don’t stop. Right. You’re not , oh, 16,000 a six,, tens of millions of dollars of sales. Let me be done. Here you go and teach people , how to build data centers, how to build Bitcoin mines, and , you successfully taught people how to build over 37 of those.

So, why the shift or what makes you go, okay, this was amazing. Sales was fun. I learned a lot here, but now I’m gonna go and teach people how to build these mining facilities. Tell me more.

Taylor: So I was actually doing all of those things at the same time. Because it would be , , I, I’d go for a couple of weeks seminar and then I’d return to the sales job. , I guess for me, ,, I’ve always been an educator. , my high school yearbook quote was, those who are crazy enough to think that they can change the world are the ones who do.

Right. , So for me , it’s all about leaving a lasting impact on the world , and the people around me,, and bringing everybody along for the journey. But, , you hit on a subject that is something I’ve gone back and forth on over the last five years, which is , okay, so I get a big exit, I close a massive deal, [00:11:00] and now I can retire if I want to.

Do I, no, no, I’d be bored. Right. Always need to have , another challenge to push against , just to keep life interesting.

JohnPaul: You sound me. It’s , you wanna just keep busy, but also you are making such an impact, which is you don’t have that opportunity on the sidelines and that’s how you’ve been able to scale this thing now we’ve talked about teaching people how to build sites, doing the ASIC cleaning.

, You’ve gone up the stack all the way to the sale of facilities. Now you’ve repped over two 50 megawatts of site sales. And talk to me about the fastest route from an LOI to energizing without locking up a lot of capital as someone who’s trying to do a lean. And how were you able to do that successfully?

Is there any deal specifically you can speak about that the audience would find interesting?

Taylor: What scale are we talking about?

JohnPaul: Scaling and selling facilities, basically anything on topic, even if they’re just

Taylor: Okay. Okay. , There’s layers to this the way I see it, right? Because , it’s a different conversation if we’re talking [00:12:00] about one megawatt, 10 megawatt or 300 megawatt, right? There’s very different things you need to do with,

JohnPaul: Yeah. , You can approach it. I would say just let’s break it into barriers, greenfield nut and just load studies and then just available power immediately.

Taylor: yeah, the, main thing is, finding pockets of capacity where the infrastructure, , to the maximum, extent that you can find. Is already there for another reason. Right. So, , if you want to do one megawatt of mining, maybe , you move into, , a warehouse facility that a previous, CNC shop was in, something that.

, If you wanna do 10 megawatts, you need to find a substation that the utility has recently built, , but hasn’t yet accrued its projected load growth., Or load has dropped off for some reason. Then of course, if you wanna do, , hundreds of megawatts and the really big scale stuff, well that used to be easier before, , traditional data centers came around and gobbled up all the capacity.

I’ll leave it at that.

JohnPaul: So Taylor, if someone has done the [00:13:00] one megawatt, the five megawatt, the 10 megawatt site, how do they go to the 150 megawatt site knowing you’ve done this a few times and you have the expertise, walk me through that process for those who might be interested in taking a part of the Bitcoin mining growth and AI growth story. , That’s to come.

Taylor: yeah, you wanna look for existing grid infrastructure that you can extend to your facility. , Essentially , the process is gonna be finding out where the main backbones are, finding land near that to attach to your load study. Submitting the load study with the local utilities serving that region.

And going through all that process. There’s gonna be engineering diagrams, there’s gonna be a few layers of the load study, and this is gonna be, a year long to multi-year process. , And throughout the course of that, you need to line up all the different variables, which is, , where are you getting your substation components from?

Because at that scale you’re gonna be building a substation. [00:14:00] Where are you gonna be procuring that much data center infrastructure? \ And depending on your business model, who’s gonna be your client or how are you raising the capital to purchase the servers that you’re gonna operate in the data center, you need to coordinate all of those things in parallel with each other, , and ratchet up each step as fast as you can in parallel with the other ones to unlock the next step because you need to do everything in lock stop.

But. A lot of times it is a chicken or the egg scenario. , , for example, the utility wants you to have the land to do the load study, but you want the load study to have the land, right? So it’s , okay,, how do I make that happen without buying 300 acres? Well, maybe you do an option on the land instead of purchasing it.

, Then once you have the option on the land, that’s long enough for you to find out if you can get the power to that location, then you go through the load study process and that’s replicated through everything in the business.

JohnPaul: It is really a stool, and to your point, you need all three parts of the stool, and [00:15:00] at any one point in time, a part might fall out and you are left with no deal. So it’s something where you’re running multiple deals at a time. These things don’t happen overnight because, they happened, it took years before.

Now it might even take decades, and we can talk more about that to get some of these deals done because of the inefficiencies in the system. So. , That’s a mining site. That’s a large scale development. What are you guys doing about AI and how do you view it? Shift your business shifting? If it is from Bitcoin mining to ai, is it a hundred percent shift?

Is it a half and half now? Is it 25% ai, the rest Bitcoin mining? Talk to me more about that. Taylor and then Bill, feel free to jump in as well, is what has changed, what’s been the eureka moment in the past two to three years in this space where traditionally, , the co-location guys just didn’t come to Texas.

They were in Virginia and that’s it. They didn’t want to build in Oklahoma.

Taylor: So, there’s a lot of layers and a lot of different answers I can give you there. So, , first off, the [00:16:00] business these days is 90% traditional data center and 10% Bitcoin. , And, the driving factor behind that is I’m very passionate about Bitcoin’s ability to. Stabilized civilization, , to bank the unbanked to give people access to money that is separated from state.

, But , when a data center doesn’t work, doctors don’t get medical records, the military can’t function, the police can’t function, right? So it feels , a much more tangible impact , on the local community in our country, , as well as having more stable cash flows, better access to investment, all those kinds of things.

And it’s , , I was still very passionate about Bitcoin, but , now I’m passionate about , the potential to retire everybody on the planet,

? , So what’s changed? , A lot of the skill sets have carried over, the ability to go and prospect these pieces of land, , get a large amount of [00:17:00] electricity served to ’em.

High density, high heat environments, whether that’s air cooling or liquid cooling. But another thing , that’s fun for me about the transition is that the technical integration of the systems to make sure that the server doesn’t ever turn off,, are much more advanced and robust. And then you get into a lot more questions of connectivity.

, And, yeah, , just a lot of fun technical rabbit holes that go down for me.

JohnPaul: Bill.

Bill: So I’ll lead off with . You guys know that Robert Redford passed away recently and one of my favorite Redford movies was sneakers. Do you remember that one from the early nineties? , , they were playing, a team of white hat hackers, , at that got in over their heads. So reference characters having a discussion with his old friend now turned enemy, Cosmo and Cosmo says quote.

There’s a war out there, old friend or World War, and it’s not about who’s got the most bullets, it’s about who controls [00:18:00] the information, what we see and hear, how we work, what we think. It’s all about the information. And so then he goes into discussions on what they’ve done with the information and the fact that they’ve turned over everything from banks to small countries, which of course was an allusion to what George Sowers did to the British Pound, , a year before the movie came out.

Alright, so with all of that being said, more computing that you have, , the more potential is, roughly aligned to how much power you can get to drive that. And what’s happened over the years is, for a number of reasons, which I don’t want to derail the conversation over, let’s just say that. number of political decisions resulted in a bubble in terms of name plate capacity versus dispatchable capacity on energy.[00:19:00]

So keep in mind that politicians , and most media types generally when it comes to complex subjects, they’re not very astute. So the politicians simply looked at nameplate capacity and said, oh my God, we’ve got all this spare capacity. So here in Texas Senate, I’m in Dallas, in er, so in Ercot, wow, Bitcoin liners, data center operators, we’ve got all this spare juice that you can come take up rather than the historical model, which is you contract with the utility.

And if you’re big enough, you have to coordinate with the utility to have either your capital build the power plant and the lines infrastructure. And then pay on the margins, or you have a long-term PPA and the expense of the CapEx and opex and margin is passed along to you. So in other words, in the classic model, they quote moms and pops never saw [00:20:00] that, right?

Because you were paying for what you need. But the politicians said, Hey, we’ve incentivized all this spare capacity. Come one, come all sign up. And everybody did. And then the grid managers went, oh crap. Because now you’ve got this very big gap between what’s on paper and what’s actually dispatchable you.

If the sun doesn’t shine, if the rain doesn’t fall, or if the wind doesn’t blow, and the worm turned. So the worm turned from, , we’ve got all this power to, oh my God, these data centers are taking all this p and in the meantime, now AI had been around, Taylor, correct me if I’m wrong, , going back to probably 2016, , but it really got legs under it with chat GPT around 2022.

That was when everything hit primetime and everybody realized, oh my God, the computing requirements for data centers running this stuff are [00:21:00] three times as high, and the power requirements are three times as high. Now, not getting into the whole discussion on fit for purpose on ai, ML or HPC, , and whether it’s AI for application or training and all of that, just very generally, you’re getting into far higher power density, far higher cooling density and the power use and water use that comes with it.

So now you’re in the big time politically and meanwhile ’cause of this gap between what’s really available., On the generation side what’s available on the line side, the t and d or transmission and distribution infrastructure. Now you’ve got this massive, massive queue of people who are signing up that may or may not have actual projects.

But if they did, , one example I saw, if all the load studies were approved, it would be three times the capacity of that infrastructure, which now brings you into [00:22:00] what am I gonna do about this? And the short answer is, you’re gonna go back to the model with Henry Ford. If you’re familiar with Henry Ford in the 1920s to build the Model T back then, he owned everything from the rubber plantations in India for the tires all the way through to steel mills in Michigan to build the model T Well.

If you follow Mr. Musk, a couple of years ago, he announced it at Boac Chica for SpaceX. As you’re aware, they’re using a rocket technology called methyl methane and oxygen or natural gas and oxygen, where he posited that at some point they would probably have to drill their own natural gas wells at the site to provide for security supply.

’cause , Elon, , he likes to have his hand in every part of the supply chain versus Boeing, which is the polar opposite. So that’s where you get , the intertwining of , where the data centers going in the [00:23:00] future. Where are the power and water requirements going in the future, and how do you guard against supply risks?

Equipment risk, which we haven’t even talked about yet. Security of supply risk , on getting equipment, and also the political or sovereign risk of, we mentioned SB six. SB six is a quid pro quo. Okay. Are you’re familiar with, SB six here in Texas, where as of now, , anything larger than , a certain figure 75 megawatts, you’re interrupted.

So what are you going to do? A data center can’t be interrupted, so you have to bring your own. And so all of these things have completely changed the landscape of where data centers were even, I’d say, five years ago.

Taylor: Heck, 12 months ago.

Bill: yeah. Well, it’s growing exponentially. It’s growing exponentially.

, I’ll leave that conversation with this. If I’m looking in my crystal [00:24:00] ball, what’s possible 15 years from. Now, , John, do the capacity of three gorgeous dam in China? The largest hydropower dam right now?

JohnPaul: It’s probably in the three five gigawatts range,

Bill: Mm-hmm. It’s 26. I’m predicting that you’ll see a data center, super campus, ultra super, whatever, mega campus in 15 years time. That will be 25, 26 gigawatts,

JohnPaul: Wow.

Bill: one cycle. Because as the technology matures, as you get, higher and higher rack densities, as you have more cases for training and development of very, very large AI modeling, and the AI modeling goes from mimicking to truly thinking, which I don’t think we’re there yet.

But ,, Taylor mentioned at some point AI does have the capacity now, won’t say all. I’ll say most, most especially [00:25:00] transactional roles today, , AI gone, first level tech support, AI gone, , transactional roles, that were typically pushing processes in organizations, ai, so all of that.

So as the AI computing demand grows, the data center demand grows because the capacity demand grows and the requirements for water and power grow. So that’s where we are.

JohnPaul: , this is a wave of

Bill: Yes, it’s a tsunami,

JohnPaul: And I think Elon even said it with Doge, he’s , I feel I’m picking up needles on the beach trying to fix the government when this wave’s coming and it’s going to disrupt everything. , what am I doing with my time

Taylor: Absolutely. Yeah. And everything in our modern lives depends on these data centers. the money of the future. Bitcoin depends on data centers. When we send a text message, pick up a phone call, go on Google Maps, Facebook, this podcast recording right now, , that all happens , in a data [00:26:00] center somewhere.

So everything about our modern lifestyle centers around these facilities that have become so integrated into our society and. , Everyone’s feelings may be about that. , I personally am an, am an avid user of technology. Therefore, I’ll always support a data center being in my neighborhood because as we look out into the future of the world, we all know the world is changing very rapidly due to technology.

And this AI thing is a, a big buzz right now. , America used to be, , a powerhouse of manufacturing and we shored that and maintained the world through the status of the dollar. While, , now bitcoin’s coming along and disrupting that, and now we have , this AI thing that’s happening.

And, , the question that I ask myself is, do I want that innovation to be happening in my country? And the answer for me is always yes, because, , do I want rent to stay low because robots are building houses? Do I want to be able to afford food or buy byproducts and [00:27:00] services? At an affordable rate, because it’s being produced locally, , by virtue of these data centers, the answer for me is always yes.

JohnPaul: And I think there’s a shift to your point. we all use the tools, but we don’t see the data centers ’cause they’re not bias usually. And now it’s becoming a much more mainstream conversation. I wanna switch it to the consumer. And talk about. So in 1930, the US lagged behind Europe in providing electricity to rural areas, and only 11% of the US farms had electricity.

So by 1937 in the New Deal era, right, they, 1936, they created the Rural Electrification Act, which then gave loans to build these power systems and basically subsidize building out the rural grid where investors from investor owned utilities would not see financial return. And that helped lower costs for consumers across the board.

But now in 2025, it seems [00:28:00] the consumer, will be getting squeezed on the electric bill and already is getting squeezed today. Now AI is a good scapegoat to point out and be , you’re the problem, it’s you. What is the real issue? increasing the rate on the consumer and how do we ensure that all this bloat in the system and all these load queues, , that’s projects that don’t, might not actually get done, don’t lead to a higher cost of to everyone because of increased generation or an overbuilding of generation?

Let’s have you guys jump into that.

Taylor: So there’s , a handful of factors there. One is inflation is a real thing. Energy prices have been pretty stable for a long time, while inflation has definitely occurred over that period. Secondly, , the grid hasn’t really been substantially upgraded since those times that you mentioned. And we’re now bearing the cost of not only needing more energy for our home devices, for our ac, for our electric cars, [00:29:00] for automation and manufacturing, and for yes data centers.

, Now in the context of the grid, having aged and having this new demand, not just from data centers but from our modern lifestyle, , come onto an aging grid. There’s been a combination of lack of investment. Inflation that’s occurred, that’s raising prices. , And , it’s easy to just point at a target and say it’s all because of this, and they should pay for everything.

Which , yes, to a large extent, you , should be a good neighbor and make sure , that what you’re doing in the community you’re in is a creative to everybody. But, , one of the fundamental issues is that capital chases returns, right? And investing in utilities while typically being a secure and stable long-term return, it doesn’t get you , large accumulation that most investors are looking for.

So all of the money for the last 20 years has been pouring into tech stocks and data centers and AI and software , and all these [00:30:00] items leaving investment in our grid to be. by the wayside. And now, , we’re collectively paying the price for that, so to speak. , Bill, do you wanna jump in and add to this?

Bill: Yeah. , I’ve read an article in LinkedIn, recently. , One of my friends is Doug Sheridan, who has the Think Tank Energy Point Research here in, Houston. And basically what Doug and I came to was that energy policy over the last 20 years has had the effect of playing Pied Piper to capital to one specific area.

And so that’s just sucked the air out of the room for everything else. Whereas historically , you had a utility , that had a long-term growth plan for its residential and small industrial and commercial and large industrial commercial would contract directly with the utility as I previously explained.

. have growth plans for the moms and pops that took into [00:31:00] account a new subdivision needs new lines, infrastructure to the substation and the lines, infrastructure in the substation back to, , the main, , transmission trunks to the power stations. All that’s covered as part of , the utilities growth plan.

All of that got wrecked, in the early nineties. First on , the thing of, hey, we’re gonna open up the grid to quote competition, except the mechanism wasn’t put in to take care of all the infrastructure requirements along with, , new generation capacity. And then that was turbocharged with the ITC and PTC towards, renewable power in the early two thousands.

, As I mentioned, now , you’ve got all these large power users wanting to come in, but there’s not a mechanism there to properly. Account for what they are going to do if they’re connecting to the grid, to all of that transmission infrastructure. So it’s very ad hoc and they get caught in the middle.

Right? Because if you’re an investor in a new [00:32:00] data center that’s 20 megawatts and no one’s demanding that you pay for additional lines infrastructure, you’re not gonna do that. Your business case is not going to account for it. And the lines infrastructure that has got money towards it has been towards intermittent generation that’s south in Timbuktu, because that’s where the wind and sun happened to be.

But that’s not necessarily where the industrial and commercial loads are, including, data center loads and Bitcoin mining loads. . So there’s a mismatch.

JohnPaul: And to your point, the mis mantle was created. I-T-C-P-T-C, these are tax credits that created generation to be built, not where load was. We’ve never approached the

problem.

Bill: , The incentive you’re chasing is the credit. You’re not chasing the real world issue that needs to be addressed by new capacity. The credit is a phantom capacity, and because the credit is a phantom capacity and it’s not [00:33:00] tied to a real world need, that’s where you get misapplied and misallocated capital.

Taylor: Right.

JohnPaul: We’ve basically allocated capital, to your point, these generators and the consumers not benefiting from this negative price energy in this certain area. Everyone’s getting their tax credits for building it. They get back their money plus their, , 8% return a year. But the consumers really left holding the bag with a more volatile grid and political , utilities with non-retail access states not allowing this cheap energy to flow to their consumers.

Bill: Right.

Taylor: exactly. , We’re all for being good stewards of the environment, not polluting. , All of those kinds of items. So the spirit behind that movement, was well founded, but it, created the effect that, , a lot of people don’t know this about the grid. , Every time you turn on a light bulb, there needs to be an equal and opposite reaction.

Somewhere in a power plant that turns up just a little bit to power that light bulb. [00:34:00] So you amplify that out across. Factories, charging cars, data centers, et cetera. And then we have a concentration of this energy that, one, you can’t turn on and off when you need to. Two, you can’t turn it up and down.

And three, by the way, the weather’s not perfectly predictable. We all watch the weather, right? , So we don’t know when this energy is gonna work. While at the same time we’re trying to play at , this perfect balancing act. And that combined with the lack of investment and the ability to get energy from power plants to areas where people use the energy, and now the energy source that we’re getting, \ isn’t constant, is unpredictable and can’t be turned up and down., It’s caused a problem where now we essentially just don’t have what we need in the grid to be able to power our society.

JohnPaul: And now we’re talking about this consumer, the consumers hearing that AI is gonna make their bill go up. Let me ask you this question. Utilities are incentivized to [00:35:00] grow their asset base. Why are they incentivized to grow their assets base? ’cause they can make a fixed return regulated by the government X percent per year on their asset base.

You have these massive data center customers coming in with almost an unlimited. Amount of capital to build new data centers and to Bill’s earlier point how far do they go in upgrading and paying for their system? Is it just the lines? Is it their substation? Is it the generation? That’s a conversation we’re having, but from a utilities perspective, they see money printers everywhere.

They’re , wait a second, I can have a data center park. I can only let Apple and Amazon use this power. I’m gonna build a 200 megawatt substation. I’m not gonna let anyone else use it, but all my rate payers are gonna pay for my substation, pay for this transmission line to my industrial data center park, and I’m gonna go find my client, but I’m not gonna let anyone else in.

It seems we’ve moved away from this open retail [00:36:00] access. The grid is for the people. Everyone can access the grid if you’re paying for it. To , oh no, I’m gonna grow my asset base, but I’m only gonna let certain players use this come into this 200 megawatt site in this example. How do you guys see , these misaligned incentives affecting consumers and then also affecting, the AI players in this space.

Even smaller operators , , where me and you are in that 50 to a hundred megawatt scale where we can’t go compete in the gigs right now.

Taylor: So from, call it 2005 to 2018 ish. Those scenarios where you just mentioned where the utility is gonna make an investment into infrastructure in order to bring a user to that location did happen. And the reason that happened was because communities want high paying jobs in their region, right?

, , a lot of people want to go work in a data center, make a, a top 20% salary, have a pretty cushy office position, , have the tax incentives be there. ’cause , , these data centers, they are [00:37:00] given tax breaks, but it’s normally not a hundred percent.

, And even, , if you’re paying 25% of the taxes in one location, over five years, that builds you five new high schools, right? That’s the reason that utilities made that investment to bring those kinds of businesses to the area because it was good for the community. But what’s actually happened.

In recent times, now that those resources have become constrained is that the data center operator pays for everything. They pay for the power lines to get from , the existing transmission corridor to the data center. They pay for the substation, build out, they pay fuel riders. And it really does pay , for a lot of things.

And I think that right now what’s happening is that we have , a perfect storm where we have unpredictable energy supply, aging grid infrastructure, and massive demand for electrification from a variety of sources, not just data centers, combined with an [00:38:00] inflationary environment. , And I think that is really what’s driving energy rates up.

It’s not just data centers coming in because the data center operator , and the, in the past decade, pays for everything , when they come into a location. And , that’s my fear.

JohnPaul: And Bill, before you jump in, Taylor. You’re right, the data center is paying for it, but guess what? They don’t own the asset. The utility says This is my asset. Now I’m gonna charge 8% on it. Even though I didn’t pay for it, the ratepayer didn’t pay for it, but I’m still gonna charge my rate base. 8%.

Bill: Historically, I’ll give the example. I, , years ago I worked for an Illumina smelter down at the bottom of New Zealand. you need to know about that site is it’s powered by a hydroelectric, , facility called the Manor Power Scheme. Manor is unique. It’s not a regular hydro dam. It’s a lake here and a lake here, and a race between the middle, which generation.

800 megawatts between the two lakes. Originally, then owner of the smelter that was coming in [00:39:00] Alco was going to build and operate this power scheme, but then a bunch of decisions were taken. Then the New Zealand government finished it, and as I described earlier, they tacked on a capital cost and the margin cost, which was inked into a contract prior to the first stage of the smelter opening in 1971.

Now to your point, the, what’s the famous quote from Empire Strikes back? , I’m altering the deal, pray I don’t alter it further. , Well, that’s what the New Zealand government did to the smelter in three years time.

JohnPaul: Yep.

Bill: so there’s, an issue around trust where if you’re contracting with an, a government owned entity and the government owned entity says, well, .

That’s a nice PPA, but now I’m gonna open it up and you’re gonna pay what I require. Where else? Then you get mistrust in there. , And on that topic, when those [00:40:00] guys were under negotiation for power in 2008, you had people leaking to the main papers in New Zealand claims of costs and, , opinions that this smelter is quote, paying less than everyone else.

Well, logically yes, because if you’re buying 600 megawatts of power, you’re paying less than little homeowner who’s using, , a thousand kilowatts a month. Okay. So, but that’s the point. It was political positioning. So where are we now in all of that? Look at meta, look at what they’ve done in, I think it’s Richmond Parish, Louisiana with Entergy.

They have contracted with Entergy, where Entergy are going to build five gigawatts of CCGT. So combined cycle gas turbine power generation for that ultra mega hyper, whatever it’s called these days,

Taylor: Hi, carry on.

Bill: Parish. Okay? Now, [00:41:00] that generation is not going to be owned by meta it, it is going to be contracted so that the capital is paid.

I don’t know the particulars of the deal, but I would presume that the cap cost for all the infrastructure going from the electron generation at the CCGT all the way across the lines, through the substations to the site, is going to be tacked onto that PPA and yet. You’ve got media going out there, my favorite Bloomberg, , putting op-ed articles out there about how all these data centers are, quote, taking power away from people.

And the big challenge is, while Taylor is right, salaries are high. Once a data center is built, there are very few jobs for the square footage. This is not an office park where you’ve got thousands of employees in an office building., A 50 megawatt data center could have less than a hundred [00:42:00] people in that facility. Now, it may have a bunch of outside contractors doing everything from, working on the servers to cleaning the floors. But those aren’t FTE, those aren’t employed , by the operator. Okay. So what people see is. Oh my God, this big power users coming into my area, they are taking all the power that’s available and through, , whether there’s any truth in it, into it or not.

The media color, the media area is positioned such that the uneducated reader is led to believe that the data center is getting quote one over ’em

JohnPaul: Hmm.

Bill: and then add to all of that, the actual logistical issues of how long load studies take to accomplish, whether the power is interruptible or not.

And, , I’m talking about interruptable from a [00:43:00] regulatory standpoint, and then the uncertainty of dispatchable versus non dispatchable energy. Okay, so you’ve got all of those uncertainties driving risk. And this is why I am saying that the future. Is behind the meter, bring your own power. And then another thing I haven’t even talked about yet is the difference in load between a classic, what I would call an analog load.

So if you think a large mining operation, the equipment is on

Taylor: physical mineral mining.

Bill: yeah. , so for example, there are mines that have rock crushing circuits. , The ball mills on those mines can be anywhere from 10 to 35 megawatts. The largest ball mill is in Western Australia, 35 megawatts.

Well, when that thing is off, it’s off. So you immediately have this drop of 35 megawatts. Oh crap. Something has to be turned down versus. The way, and Taylor’s the [00:44:00] expert here. The way, , AI works when it’s learning or it’s, or it’s executing on its models and the load is very digital, it’s very chunky.

And those chunks can be within cycles that are less than the second at the time that if they’re not coordinated right, can set up demand waves very much physical waves, that now you need something in between to shock, absorb that load. And that’s where I think, best systems , electric, , battery energy storage systems rather, can perform that function because the load is so different.

And whether you’re on the grid or behind the meter, you’re gonna need that. Otherwise you’re gonna be creating problems upstream for everybody. it’s a very different world. It’s a very fraught political world because of, we’re coming off of subsidies for certain types of power, [00:45:00] but we haven’t addressed the fact that , we’ve neglected the lines infrastructure, the t and d infrastructure from the generation point out, and we still haven’t yet, , solved the issue of allowing the capital to really go back to dispatch for generation base, load generation.

JohnPaul: Wow.

Bill: a mouthful.

JohnPaul: That’s, no, that shows you the complexity of the topic from the local level where you have thousands of landowners in Facebook groups saying, stop the data center to the mayors speaking out in these towns for and against these loads to

Bill: , Do you realize the ultimate irony there? The Facebook groups are creating the demand.

JohnPaul: yeah.

Bill: know, for very quick example, I worked at a steel mill in 2008. It was going to expand to land that it owned for what we would call a demolition landfill. Do the term a demolition landfill , is [00:46:00] industrial type waste. when you break up a road and you’ve gotta do something with the concrete. Okay. It’s that stuff. , They were gonna expand , this plot that was in this beautiful rolling valley, and the locals were all miffed over it ’cause they loved the views, but they didn’t understand that years ago, this has been sold for this purpose. And they’re spraying in spray cans, signs about fighting the steel mill.

And I remember thinking, did anyone bother to ask where the steel came from for the cans, for the paint that they were using to make their signs?

JohnPaul: We forget. We forget because we’re not next to it. We’re not close to it. We don’t see it. And , that’s what makes it abstract and much harder to understand.

Bill: Yes.

Taylor: Going back to the consumer. , These issues of rising costs of employment per use of resources in our local regions are, , these , civilizational, [00:47:00] societal level problems , that we’re grappling with, not just as a country, but as a planet right now. , But I think one of the things that really adds fire to the flame, so to speak, with the discussion of, rising energy costs for rate payers, , is that, , it’s not just electricity.

Everything’s more expensive, rent’s going up, food’s going up, gas is going up. , So when you, see it all go up at the same time, it’s when you talk about. The price of food going up, for example. , It’s hard to find, , a direction to point yourself in, to be emotional.

I’m about to be , okay, this is the problem that we need to solve. But, , when it comes to electricity, , the grid is , , before I got into Bitcoin mining, I don’t think I ever noticed a power line in my life. And now every time I drive past a substation, I’m , Ooh, look at that.

Right? But, , so this thing that’s out of sight, out of mind that, nobody really thinks about how it works, how it was built, why it does what it does, and all of the energy policy and history that we’ve talked about throughout this [00:48:00] conversation. So it’s easy to just look at data centers and be , that’s the problem.

I don’t it. Right.

JohnPaul: So how do you guys stay focused on, and what is the goal of, the business now? Taylor and Bill, with this minefield in front of us of political, , death stars using lines I’m altering the deal. You just invested $10 billion and the deal has now changed to your community of Facebook groups with rioting, with their paint cans that are made by this aluminum shelters.

Where do you find your guys’ business and how do you add value to the space?

Taylor: , There’s two questions there. , So , I’ll address the first one,, in order the way that I see it, which is that educating the local communities that we want to put these things in is very important, right? , Don’t let the community hear about , the giga site that you wanna build, , when you’re one week away from signing the deal.

And the city board’s gonna vote on it, , , get involved in the local community, donate to some [00:49:00] nonprofits, have some hot dog cookouts at the local ball game, , and talk about what’s going on. , And, , let people know. What it’s doing for the community. for example, even if you give a data center, a 75% tax break, the scale of capital that’s being brought to the city is gonna pay for new roads, new schools, , new fire station, and, during the construction part, employ a lot of people.

And even once it’s over, maybe not employ , so many people, but, give the it and client youth of the region the opportunity to go and work in it near home. Right? , So it’s really important one, to educate the communities that we go into. , And two, to be good stewards of the, resources that are being allocated to us.

Build to the extent that we can, our own power generation, , build to the extent that we can, data centers that consume less water., That’s the, the first part is , how do we step through this minefield, right? Basically be a good neighbor, , is what it comes down to. And, , the second part, , that we’re really passionate about ties into that, [00:50:00] which is designing and building data centers that are more efficient, more quickly than what has been done over the last 20 years.

, And that’s, done through rethinking the design of how do you get the power to the rack? How do you get the cooling to the rack, right? Because, , just to give an example, a lot of, , data center campuses have , these centralized cooling plants and these centralized energy stations, right?

, And what that ends up causing is now you need to pump, , a thousand gallons per minute of water, 600 feet away across a bunch of twists and turns through pipe. You’re using more energy than you need to cool the data center because of its design. So when you bring everything closer to the rack, you increase efficiencies in the data center.

So that’s one thing that we’re really passionate about is, building centers that are as efficient as possible. And then above and beyond that, , innovating, in the contracting and construction side of things where [00:51:00] we, and , this is a long-winded subject , that I can bang on about for a long time.

But, , in essence, , creating scenarios where the people who own the data center and the people who are building it, they’re both incentivized to build that as quickly as possible on budget and on quality, right? Because , the way the industry’s operated for the past 30 years, , if there’s anything I learned in Bitcoin, and it’s actually Bitcoin that taught me this, is is that incentives rule the world, right?

So if the contractor who’s building the data center is. Making his money on a cost plus basis. So they’re gonna make 10% above cost. So say for example, the data center costs a million dollars, they make a hundred thousand. Well, what incentive does that create? Now the contractor wants the data center , to be as expensive as possible, to increase the size of the 10% that they’re gonna get.

So innovating in the, contract structure of , how we build things, has been an important part of our business.

JohnPaul: And , are you guys now doing what cost minus structures or what? What’s the innovation?

Taylor: Yeah. [00:52:00] Yeah. So, there’s , a portion that’s allocated just for the services , that are being provided for our counterparties, right? But then there’s a portion of the margin that’s rewarded for delivering the data center on time. And there’s a portion of the margin that’s delivered for the data center being delivered on budget. , But we gotta be careful with the on time and the on budget part because. When you try to go fast and when you try to do things cheaply, what happens to your reliability? Right? , So it’s making sure that a portion of that, , compensation that’s paid for services is carried out over time as it’s proven that the data center operates the way that it’s supposed to.

JohnPaul: , That’s, a really good, good way to look at it and, , innovative. So great job with, with

approaching it that way.

Bill: , To add onto Taylor’s point, I’m a big fan of Dale Carnegie. If you’re gonna convince somebody to do something, you’re going to explain what’s in it for them. , , How are you going to benefit from this? And the first biggest thing is when you put in an operation that, even if it [00:53:00] only has five people in the building, it’s still earning a lot of money.

And the operation is cited in the local county or in the local town. And that town is going to realize cash flows from that operation. It’s also going to employ to degree that it’s been cited properly, it’s going to employ local labor for the building upkeep that I described earlier. You hear about facilities that are being built in the middle of nowhere.

Okay, well great. Now try to go hire for people to work at those places and try to hire for. An electrician , to go, perform a job at one of those places. Good luck. , And so you wanna bring the community along. Here’s what’s in it for you. Here’s what we’re going to do for you. And yes, this facility is going to use power.

Here’s what we’ve done to mitigate any impacts on you. It’s going to use water. Here’s what we’ve done to mitigate any impacts on you. And as Taylor mentioned, getting [00:54:00] out in front of those discussions so that some room temperature person in media that has nothing to do with your business and knows nothing about it, isn’t just pulling stuff off the top of their head.

’cause that’s what they heard in some Facebook group. , When they ask somebody about it or a slack that’s filled with other journalists that together create a black hole of non knowledge. So you have to get out in front of people. You have to show ’em what’s how they’re gonna benefit.

They’re not necessarily going to benefit in the ways that they think they’re gonna benefit. It’s not gonna employ a thousand people. , It’s not gonna be building widgets that they see rolling out the door that they’re gonna be able to buy. So that’s where you educate ’em, because otherwise, , we all have our areas of expertise.

I knew absolutely nothing about intermittent power generation. And so someone requested that I review a proposed PPA for a wind farm back in 2008, four months later. I knew everything that [00:55:00] I needed to know and it weren’t pretty. But the point, , is I went in with an open mind. I hired experts and I jumped in with both feet.

Now I have to treat people in the local communities the same way. They’re not stupid. You treat ’em they’re stupid. They’re gonna see right through that in half a second. , And all you’re gonna do is tick ’em off. So you need to be at the council meetings. You need to be at , the local events, the fairs and that, and get to know people in the town and always be explaining, here’s , what’s in it for you.

JohnPaul: I can’t agree more. , with the meetings we’ve had that are successful are the ones that you get ahead of it. The ones that are 20 people asking questions and everyone, , poking at it, saying, what about this? What about that? You’re already falling down the hill. You have no chance. Even if the size of the data center is a megawatt, it’s the perception which rules the reality of the situation.

I wanna switch topics to modular generation. ’cause [00:56:00] you mentioned that this, the grid congestion, grid cues, frequency control, demand response by ai, how that is gonna cause an issue. Do you see modular nuclear reactors, before 2030 being live and helping support data centers? Or is it more just turbines?

We have tons of natural gas turbines in the queue to be made cogen and batteries.

Bill: Here’s why I don’t see SMRs from a regulatory perspective. If a nuclear power plant has a trip, you don’t go, just flip a switch back on. There is a massive regulatory requirement on, , doing an investigation on what caused that trip and ensuring that the plant is safe. That could take months.

So if your behind the meter site is tied to an SMR and that SMR has a problem, well, now, , you can’t build n plus one redundancy. In other words, if you need three SMRs of a capacity to run the [00:57:00] place and you do plus one, which means four, you can’t run it that.

’cause now for months on end, you don’t have any redundancy, which means you’re right back to what’s my non-nuclear redundancy. So let’s put that one aside. , I think once the discrepancies on non dispatchable power are flush through, and honestly that’s gonna take 20 years ’cause that 20 years is the life of the facilities that are now just being brought online. But as the subsidies on the ITC and PTC fall off, the number of new facilities that being built is gonna fall off a cliff. Just what happened yesterday. , The EV tax credits. Just fell. So you had a mad dash for people to buy before the tax credit ended, and now you’re gonna have the hangover period where nobody’s buying.

And as the market writes itself, there’ll be a lot less. ’cause now people won’t be buying them for tax credits. [00:58:00] Same with power generation. So as all that flushes out, you’re gonna have 20 years where really you’re gonna be looking at, , a number of different technologies. You’re gonna be looking at, , gas turbines, number of different versions of gas turbines.

You’re gonna be looking at reciprocating or rice, , reciprocating, internal combustion engine. Right? And you’re gonna be looking at what I call novel technologies that use natural gas and similar fossil fuels in a way that has a far less, emissions footprint. And I’m not talking, just talking about COT, I’m talking about NOx and item, other items that affect, , emissions in what are called non-attainment zones.

EPA, non-attainment zones. ’cause today, if I wanna build the behind a meter facility in the Dallas Metro, because that whole area is under EPA non-attainment, so the chances of me putting significant generation on a site, no, but there are [00:59:00] novel technologies coming out. There are linear generators coming out, there are fuel cells, on a number of technologies coming out.

And eventually the preponderance of, , the market penetration with that equipment will bring prices down to where they’ll be, quote in the money to, , provide either supplementary or , primary generation for those sites. It’s gonna take time. everything else, , but there are options.

Taylor: The other thing that I see is that, , a lot of things in life come down to cost, how we were talking about earlier and even conventional nuclear these days. , The cost to provide the energy before any margin is made on the billions of dollars that gets invested into them is somewhere around six and a half to 8 cents.

Whereas, coal and natural gas are in , the four or 5 cent range these days. , So it’s, it can be hard to, [01:00:00] build a business case around those things in that scenario. , SMRs and geothermal right now are in the, I’ll call it nine to 15 cent range, , for producing energy. So for that to become really feasible and get rolled out, , we’ve gotta figure out how to hit some economies of scale and production to get that into a place , that’s competitive with other energy generation methods.

, And then of course there’s , the fear of nuclear fallout, which, if you look at , your one nuclear meltdown history, the only reactors that have ever melted down were first generation reactors, first time we ever tried it. That’s the only type that’s ever melted down anywhere in the.

Right. , And now we’re on generation three or Generation five nuclear or something that. Any reactor from generation two and on has not ever had a single meltdown. And actually , the deaths per year via nuclear compared to the deaths per year of people operating in, natural gas or [01:01:00] facilities it’s zero for nuclear.

And , well, , when you stand next to an exploding piece of steel all day, , that can be hazardous, right? A combination of the cost and the fear of nuclear meltdowns is preventing us , from advancing in that direction as a society. But, , if you think about it in terms of, powering a data center directly, Bill said, , one of those things trips off, you’re gonna be offline for weeks or months.

So nuclear directly powering a data center, at least a mid-size campus, may or may not be realistic depending on what other generation technologies you have on site and all those kinds of questions. But, for society at large and, , the world, that whole, , the abundance of electricity is directly correlated, in my opinion, causative of , the wellbeing of the people who live in that region.

And it’s my opinion that , the best thing that we can do as a society right now is build as much consistent generation as we can to stay ahead , in this [01:02:00] race for super intelligence that we’re in.

JohnPaul: And I, agree with you. I think that , we’re gonna win that race of super intelligence with pure force and capital. , Versus where the west, is looking at it of, as a way of trying, to, let’s say, be more efficient in the boat , and go faster with what they have, the tools they have, with some of the deep, steep optimizations that we’ve seen.

But to, your point earlier, which is the nuclear, it sounds big and scary, similar to data centers. Sound big and scary. They’re gonna make my power bill go up, but they actually are less risky than the stuff we’re familiar with. It’s almost we’re in the same conversation between nuclear and natural gas as we are with data centers and their actual impact to the community.

I wanna end it with, , you guys are experts in, the field and I want to end it with. Almost a stock pick, but, and let you guys talk about there’s companies in the space that one that , you’re watching that are public, that you think are undervalued and of course not financial advice, but just from your expert opinion, what you’re seeing.

And then one that’s, , overvalued that you think is overhyped. , We, , we’re [01:03:00] talking about Schneider Electric earlier today and , that, talk me to me through, there’s a ton of capital moving in the space. , even the bitcoin mining companies that were left to dry are now, some of them are doubling in size or in market cap within a week.

, And so there’s a lot of capital flowing into this space. Where does someone take action from all the insight we’ve given you to make a, maybe a financial decision and , get ahead?

Taylor: Not financial advice,

JohnPaul: Yeah. Not financial advice, but , , what do you ?

Taylor: , For me personally, for to do ai, so to speak, you need a few things. You need energy, you need chips, but you also need data, right? , And , that’s , where a lot of people kinda. Misread the three parts of the bar stool to do ai, so to speak. , And in that regard, Google and Xai both have access to massive amounts of data.

how often do we use Google on a daily basis? It’s probably 20 or 30 times for me. And that’s all just getting logged. And , that’s meaningful [01:04:00] human interaction that can be used to train synthetic systems to operate us, right? And then Google also happens to be a leader and developing their own chips , that are more efficient and purpose suited while at the same time being a leader and multi-campus training.

, So how essentially all models are trained right now is it’s not efficient from a capital or an operational sense to. Be having a single model train at a facility that’s separated from one another even by a mile. Right. And Google’s currently the leader and being able to make it work, , with campuses that are separated by tens of hundreds of miles.

So they have the data, they have the chips, they have the power, and they’re learning how to make it so that to make Gemini be the next best model that it can be. They’re learning how to be able to effectuate [01:05:00] that in multiple data center campuses instead of one data center campus. ’cause essentially , you reach a limitation on scale, right?

to put even one gigawatt in one location. It’s crazy. some people are doing five, well how do we get to 25 or 50 or a hundred gigawatts training these models? Well to do that and on one piece of land is basically impossible. And Google’s the leader in solving that problem right now.

And then, , with, , X ai, it’s this a similar story. They have the data because they have Twitter and starlink. They’ve proven they can move fast, solve problems quickly, and scale quickly. And, , combining all of those things together , makes me bullish on those two companies.

JohnPaul: Taylor, that’s great. I didn’t get the Google connection. I know they’ve been relatively flat. They obviously are pushing the edge with Gemini, but they’re maybe not the open AI is getting the spotlight there, per se.

Taylor: A lot of people don’t know, Google invented the LLM.

JohnPaul: Yeah, the transformer,

Taylor: Yeah. Yeah. Were you gonna jump in, bill?

Bill: Yeah, I was just gonna say, . The AI boom today [01:06:00] has a lot of, now I’m a big fan of history and I’m a bigger fan of the old saying that there’s nothing new under the sun. Do you remember your good book? where that comes from? So go back to the late 1990s, you had stupid money being thrown at anybody who said, I’m on the web internet, I’m gonna do internet commerce.

And then, a fellow Tulane alum looked at Amazon in March of 2000 and said, hold on. This is a business that has inventory and they’re not making any money. What gifts? And six months later, the bubble just went poof. Now that didn’t kill the industries and the businesses who figured out, I can’t be about, wow, I’ve got this tool.

So fast forward to today. Everybody’s got ai. I want to throw something at the television. Whenever I see an ad on a [01:07:00] YouTube video of this and this fortified with ai, it’s , , fortified with calcium or something, or fluoride, or vitamin A and it’s not that at all. It’s how am I going to take this tool and do something with it that’s going to make the product or the service that much more efficient?

Telling me that a laptop has AI means three eighths of nothing. It’s marketing crap. Alright. Telling me , that the laptop is built to leverage connection with AI to do all these tasks from. Faster and more capable spreadsheets to, , being able to put videos out at much higher bandwidth and much more professional effects. Sell me on that. I don’t care how the sausage is made. You [01:08:00] know, the joke about, , go back to up in Sinclair, it used to be that, Italian sausage was authentic. ’cause every now and then you’d have two Italians fighting over the bats who want to get knocked in, right? , But I don’t care about that.

I want , the nice food sizzling on my plate because it’s good to eat. I don’t care how it’s made. Okay, don’t tell me that it’s got AI in it. Tell me what having AI is going to do for me. So , the companies that are able to train the models for an end purpose, that has a real world application that’s going to save people time and. It’s going to make them smarter now, that’s the trick. It can’t be a crutch. It’s gotta be something that makes you smarter , and makes you able to do more with less companies that can solve that. One. Show the public what their products are gonna do using this [01:09:00] tool. They’re the ones that are gonna make money.

Taylor: And that’s, where we are right now in this whole wave is, , we’ve gotten to the point where these models have been trained to the extent that , they’re pretty smart. They can solve hard math problems, they can write long papers, almost a human, but it’s , okay, , right now, at this point in October of 2025, it’s , okay, how do we translate that into a real world business case?

Whether that’s a consumer or a business to business product. That solves a problem at a value proposition that is better than the way that we’ve done it before. And that’s what everybody’s talking about right now is, , are we actually gonna get to that and how do we get there?

JohnPaul: I, agree with you and on that, I think I want to go back to a point that Bill mentioned, which is , bill . Is the round tripping schemes that are happening with OpenAI committing $300 billion to Oracle, and then Oracle investing $40 billion to buy Nvidia cutting edge GPUs. Then Nvidia pledging a hundred billion dollars to back into OpenAI.

, , are we at the peak of the bubble? Are we in [01:10:00] that situation , where there’s a lot of financial euphoria, there’s a lot of round tripping schemes going on where the money, , NVIDIA’s receivables , aren’t being actually paid for, but they’re being booked and these sales, is that gonna affect the, not , the short term growth, or is this

just such a big

train?

Bill: me. Yeah. Don’t show me the circle jerk of those three. I’ll leave that visual to you, but, show me what’s gonna , come out of all of that collaboration. What is that going to do for the products and the services that each of those companies are able to offer to the customer base outside of , the little

JohnPaul: , It almost feels we’ve

lost, we’ve lost that , forest from the trees here where we’re not focused

on what’s

Bill: back to, I go back to the late nineties and the.com boom. Remember, for a brief while there, everybody’s running around saying, market shares everything. Well, no, it’s not. Profit is what your [01:11:00] shareholders want. And so for a while there, people were able to convince venture capitalists to throw money at growing market share until the market share did not translate into money, into profit.

Then the whole thing blew up. I think if it’s not managed well, that little thing you just described has the potential , to make a big mess all over the place,

Taylor: , personally,

Bill: something will come of it elsewhere.

Taylor: personally, this wave of growth doesn’t crest, so to speak, until at least late 26, if not 27, 28. Just given . The capital that’s been deployed and is in motion will continue to be in motion for at least that timeframe because it takes that long to build out and deploy these things. , So not to mention I just, , I see more bold when I see Bayer right now personally.

, Regarding , the Nvidia investment and OpenAI and Oracle and whatnot, , I do [01:12:00] see that when these models get larger and when they’re trained for longer, they become more intelligent. , Because I’ve , been an avid user of chat GBT since 2.0 became popular and just, , use it frequently.

, So if it is becoming more intelligent than theoretically can solve more problems. And at the same time, it’s not just on the, which one is most intelligent side, it’s also on , the other side of the versions of the models that are cheaper to run, and smaller. Also get distilled down from those larger ones.

So that means that we’re then able to solve harder problems , with less computing. And all of that is derived from the size and the power of the chips that train the largest models. So , it doesn’t seem to me that we’ve hit a point of diminishing returns yet., So I do think , , that investment that was made between those three companies is accretive, in the sense of advancing the technology.

But, , as we were speaking about earlier, how does that translate into a business case that solves the problem better, that somebody wants to pay for? And that’s, that’s the part that, that the [01:13:00] industry is still figuring.

JohnPaul: I think the amount of capital spend in AI versus the revenue in the AI software space is pretty dramatic and the software revenue has to catch up. But is this investment for, , these three year investments for GPU life cycles, are they 10 year investments, 20 year, a hundred year investments that’s yet to be seen?

And obviously there’s different parts of the stack from, we’ve talked about generation to the Rackspace, to the GPUs, and, , that’s gonna greatly impact and affect the bottom line. Well , thank you guys so much for coming on and having , this deep conversation about not only energy, AI and tailor your background as well.

What last words do you wanna leave the audience with before we close up?

Bill: Here’s what I would leave people with. Okay. Whatever your endeavor is you’re going to need. More power. And that’s not just a Tim Allen joke, but you really are going to need more power. You’re going to need reliable power. And at the same time as [01:14:00] a developer, it behooves you to be aware of the political risks of what you’re doing.

what Taylor said, don’t wait for the news cycle to write your story. I would leave everyone with, if you’re going to invest in, in a particular area, get somebody out there as the face of your company with the city council , and with the local populace, if it’s a smaller town, so that they don’t just learn about you when , you’re put on a docket for a town council meeting and then they’re all ticked off.

’cause someone lift them up into a frenzy, whether it’s,, hearsay or truth or some combination thereof. So develop those relationships with people and then. To the degree that you’re able to, I would say that for , the next 10 years, , any gener , demand, power, demand of size, look to what you’re buying, the meter options are and be the master of your own [01:15:00] ship, the captain of your own ship.

Because if you don’t, then you could get out there with that , brand new shiny data center and discover that somebody that has no care of your operation at all can just flip a switch if they need to and shut you down. And at the very least, that’s gonna cost you a lot of money in standby power generation, which depending on where you’re located, you may not be able to do because running , those diesel gen sets may hit you into not get you into non attainment.

So I would say look at the risks, look at your business case and don’t. Apply stupid money for the sake of suit. Stupid money. Understand what it is, what service that , you’re going to develop and how to get there and really pay attention to all the logistics that are involved.

JohnPaul: The details matter.

Taylor: thanks, bill.

Bill: They do, they absolutely do.

Taylor: Thanks, bill. I think my, closing note would be that, , the world is changing so fast right now [01:16:00] that the most valuable thing, we can do as people is to learn quickly, stay nimble, and stay on our feet.

, And also that, , the world’s changing really fast and. , This wave does really carry out and so much gets automated. , There’s gonna be some change that we go through as a society, and , that’s not quite that far away. So bet on yourself. Bet on yourself. Now. Work hard, grow an online presence.

Join a startup. Solve a hard problem, do something and, , tuck away resources for yourself , for the next decade.

JohnPaul: And don’t forget to buy some Bitcoin if you haven’t already. Right. That’s

one of the best resources to be holding. Well, great advice guys, and great conversation. I really appreciate the level of detail we were able to get into some of the nuances, , that we’re with mining and with the energy and with ai.

, It’s great to hear, , from both of you and to see the business grow. So best of luck in, the next gigawatt of development.

Bill: Mm-hmm.

Taylor: Thanks, JohnPaul. Good luck to you as well. Thanks for having us.

Bill: Alright.

JohnPaul: you guys and thank you for [01:17:00] listening to the Digital Gold Podcast and remember to mine on and never stop using ai.