Fireside — The Power Bottleneck to Digital Infrastructure | Boston Blockchain Week 2026

Welcome back to Boston Blockchain Week, Jake.

Thanks.

Good to see you have you back in Quinsey. Um why don't you tell everyone first who you are and what Bridger Solutions is.

Uh so Bridger what we do is we look to uh monetize trapped and stranded energy. Uh it started internationally now with a a domestic uh flare as well. We do that primarily through Bitcoin mining. uh because Bitcoin mining is uh probably the best technology that's ever been uh invented to uh give a flexible load and any grid operator any utility will tell you that uh that is absolutely paramount for them. Uh the only other technology that addresses this issue probably at all is utility scale uh battery systems uh battery energy storage systems or best.

Uh but Bitcoin mining is kind of the workhorse because it can take on large slugs of electricity and kind of act as a digital peaker plant. So if I'm utilizing 50 megawws, but all of a sudden the grid needs that 50 megawatts, I can cut off and curtail. there's no customer that is, [snorts] you know, calling a data center and and being upset because they don't have access to their data, but now the grid has that 50 megawatts back. Uh so we look uh at that really electron layer of um everybody focuses on Bitcoin because that's the lightning rod term, but that's just who pays for the electricity. It's really just an energy business.

uh but obviously that sector has come in uh direct conflict or direct association with uh the boom of traditional data centers really AI HPC data centers so when people are thinking about the grid they're thinking about electrons and managing all that uh the two go hand in hand

so a line that you had said before to me which I think is really um sort of will set the tone for this conversation is that AI data centers turn electricity into intelligence and Bitcoin data centers run electricity into money you know talk a little bit about that and I want to also expand that you know basically what we're seeing is that there's you you call them electrons there's competition for usage of of energy right even from the human side and I want to go into the digital infrastructure side but I want you to start with uh your comment

yeah so as people are kind of grappling with this notion of data centers more and more And in the Bitcoin mining space, we kind of had a head start because all the vitrial that you're seeing against data centers now, uh, if you go back five years, you know, that's what Bitcoin mining was was going up against. It was enough power to power small cities and etc, etc. There was a lot of uh altered information or just flatout wrong information in the uh kind of political zeitgeist. And now that's what you're seeing for data centers. So I think it's important to just think about uh what a data center actually is and a framework that I didn't invent.

I saw somebody else say it and and I thought it made a lot of sense is that it's akin to like an oil refinery. An oil refinery takes on crude and depending on the composition of that crude, they're going to bring out a lot of different distillates that have different values, right? So jet fuel, diesel, uh regular gasoline, kerosene, etc., etc. Um so the analogy to a data center is it's just taking on electricity.

It's taking on those electrons and depending on what type of data center it is, it's producing uh different byproducts. So AI HPC data centers are obviously producing intelligence and that's in more and more demand. Bitcoin mining data centers take those electrons on and they produce a monetary good. Uh there's other types of data centers like cloud data centers that are going to produce like essentially memory or knowledge of your organization or you as an individual. Uh and then there's other data centers like highfrequency traders rely on that uh produce communication essentially.

Uh so I think those data centers as refineries is a good way of thinking about it. Now where the analogy breaks down however is that crude uh oil refineries take on the same crude and they produce a lot of different things whereas data centers are purpose-built to produce one thing and they need a certain type of electricity input uh that that can produce that output. So for instance ahpc a training data center can exist anywhere because latency to a customer doesn't matter. You could theoretically have a training data center in Antarctica because you're just taking those electrons and the output is that model that you're creating and then you're taking that model and bring it somewhere else for use. Whereas an inference data center uh needs uh incredibly high amounts of uptime of electricity but also low latency to whoever's going to pay for that uh inference.

uh so I think it's a useful paradigm of thinking about data centers as uh just a refinery of what the output is uh that is is coming from them

and every grid electricity grid in the world is already running at a supply and demand imbalance. So you know how much I guess is uh is AI making this worse?

Yeah. So most people, and I was included in this, um, we don't really understand how the grid works and how fragile it is. You know, I was, uh, the quintessential guy that thought, you know, when I flip the switch, the lights come on. That was my understanding of electricity. But at a very basic level, uh, the generation that's being put on the grid has to equal the load or the offtake of electricity coming off the grid.

And that's measured in the frequency of the grid. Here in the United States, that's 60 hertz. In Europe and most of the world, it's 50. So, what happens if you have too much electricity is that breakers start breaking uh you know, things pop uh because there's there's too much voltage or the frequency is too high. Uh and if I've got any electric engineers in here, they're probably squirming because I'm this is layman's uh description of it.

But if you've got too little energy as compared to the demand, that's when you start getting rolling brown outs and blackouts. The former uh is bad because if I've got too much electricity, I have to curtail some of it. It's an easy problem to manage, but it's an expensive one. So that cost gets uh carried over to the rateayer, the tariff. you're going to increase that because if I'm paying a certain amount for my generation, but I'm selling less and less of it because I have to curtail.

Okay. Well, I'm just passing that on to the consumer. The latter problem though, the rolling brownouts and blackouts get you fired in a second. So, every grid operator, what they're doing is they're planning conservatively to make sure that they've got enough energy for their absolute edge cases. So, usually that's like peak summer, uh, middle of the morning or middle of the evening because that's when we use the most electricity.

I need enough generation to cover that time. And so, for the rest of the year when people are using AC less or uh other, you know, high electricity consumption, you've got a lot of excess electricity. So, what exacerbates this is the more and more renewables that we have. Renewables are great. you don't need fuel.

The fuel comes from the sun or the wind or what have you, but it's very unpredictable. So, the grid operator has to plan even more conservatively for those edge cases. Now, what AI data centers are doing is it's taking the unpredictability of our demand and it's exacerbating that by, you know, sometimes an order of magnitude because we're already unpredictable in our demand. what you know when are we going to all be watching TV when's a hot day that we're using a lot of uh air conditioning etc etc well now think about all that AI inference that's hitting the grid and it's a whipssaw of electricity demand so now it's it's taking that unpredictability and and ratcheting it up even more so when people talk the common um negative components about AI data centers Some of them are outright not true. Like so for instance the water thing.

Yes, AI data centers use water for cooling but it's a closed loop system. So you fill up the water and then it just gets recycled over and over and over again. Like if you look at the statistics of water usage AI data centers, it's like a incredibly small fraction of golf courses, almond farming, you know, things that we just, you know, take for for granted. One of the ones that's partially true though is making electricity prices go up for your residential tariff pair. And the reason why that's true is not because they're taking electricity that would otherwise go to those consumers.

It's they're making that supply demand imbalance worse. So they have to curtail more energy which then gets passed to every uh tariff payer. And that's why retail uh uh get hates it and they kind of have a point kind of don't.

So you bring up the interesting point about um my mind goes to propaganda because I think there's two sides to the AI now political football. And so who you know who's standing to benefit from the rhetoric on either side um for promotion of of AI I guess which is probably couched in um ushering in new economic growth and development in the United States and then what's the other side? So uh my background before I got into Bitcoin mining I worked for the government for almost 20 years. And uh this might come off as a conspiracy theorist take. I don't mean it to be, but there is a lot of linkages between foreign money and anti-data center campaigns.

And if you think about it, it it I don't think it comes off as very tinfoil hatish to say that a technology that could unlock untold amounts of productivity uh that our adversaries would have an interest in us at least slow rolling that technology out. And what I find interesting about the anti-data center um campaigns is that when you talk to most people, if you just ask a couple of critical questions of okay, are you using AI? And most people will say yes, I'm using AI. So if you don't want data centers in your backyard, but you are using AI, you're essentially saying my area is too good for it, but I want it to exist somewhere because I'm using it. I'm I'm using the tokens that are being produced by these data centers.

Uh I just don't want it near me and make my electricity tariffs go up. Um otherwise I think today is so politicized people will run with a narrative without really thinking through it. Um, but I don't without getting in the risk of like this or that either side of the aisle, I think anybody would see that this technology is coming no matter what. You can't have something so productive and that it's not going to come. Uh, so it's really just more uh misguided when people kind of fall prey to that line of thinking.

I'm not even saying that they're wrong. I'm saying that they're fighting a losing battle. Like it's coming either way. these data centers will will they'll find a way. Now, I think one of the useful things that's going to come from it is that AI data centers will have to become better consumers of electricity where they don't exacerbate that supply and demand imbalance.

The reason why that hasn't happened yet is because they don't have an economic incentive to do it. Bitcoin miners like are super super attuned to our electricity tariff. like moving a half a cent is like a seismic event to us. There's so much money in uh the revenue from AI HPC compute that they don't care. They're a price insensitive buyer.

But what is going to force them into being a better electricity customer for the grid and the utility is all the push back that you're seeing. So I'm not in agreement with it, but that is one I think beneficial byproduct of that. I think from what you just said though, it made me realize that we're now stacking narratives. There are narratives on narratives being built which would effectively prevent this productivity and thereby you know the economic development and growth unlock that the data centers in AI would bring because if there's a currently a bottleneck to the power generation necessary to not only fulfill the needs of this new tech but also to fulfill the base load power generation demand of residents and businesses in places like Massachusetts which is maybe something we'll get into um people have been led to believe that certain forms of energy production is bad. And so when you're fueling one political narrative that's connected to the environment and energy production to the negativity uh that's being promoted around data centers, it doesn't help the cause, right?

So how do you get around how do you counter that narrative? um because there's so much influence and information being flooded on the airwaves that people are consuming and again not checking not doing any deeper research. You know, how are you actually going to counter those narratives to achieve positive outcomes?

Yeah, it's a good question. I I don't honestly know the answer to that. Um I do believe though that AI um we exist in this kind of weird time with AI where the the opinion of AI is excessively negative but that's a that's a western phenomenon like they've done polling about optimism of AI throughout the world and in China for instance it's through the roof it's super popular here there's a universally negative feeling about AI and I think part of that is because the purveyors of AI like uh anthropic and open AI have been some of their own worst marketers of proliferating this message of it's going to take all your jobs. Uh you know there was the famous story just yesterday of the fourmonth employee at uh Anthropic who quit and then wrote an op-ed saying uh it's going to kill us all. like uh I've never seen an industry that's been so terrible at messaging for their own industry.

Um but there are just certain things that exist in loggerheads that make me believe like it's a fatal comp plea that this is coming. You can't have a $2 trillion IPO and yet have a backlog of data centers. Like one of those things has to break. And I'm willing to believe that people will eventually vote with their wallet. And if you have a two trillion IPO and there's a lot of just universal investment in that kind of stuff, the political will will get behind building a data center.

So again, I I think that's AI can be a real force for good because prior to AI, we if you looked at electricity generation in the country, it had really styied for a couple of decades there, which has put us behind the the power curve on and being ready for this huge consumer of electricity. But I'm of the belief that uh we should have as much energy as we can produce. Energy abundance leads to productivity. There's been a ton of economic studies that show that very thing. If you want to see the productivity of a country, look at their energy produced and electricity provided to their industry and and retail uh customers.

And AI has already inspired this like explosion of electricity generation in the country. And a lot of that when they couldn't tie it to the grid is now being tied uh what's called behind the meter. So meaning like instead of grid tying it, you just build a power plant right next to your load, which in this case is the data center. And uh if you're graphing the electricity generation, we were flat for a couple decades. in the last few years were were going like that again.

So I don't know how the political ideas behind AI change, but I do believe that as more and more people see the productivity that comes from it and like the amazing use cases, curing diseases, solving mathematical problems that have stumped us for centuries, you know, promoting interstellar travel, all these things are happening right now. The more that those messages shine through, the second or third order effect of that is people will see any electricity generation is good electricity uh generation. And as far as like quote unquote dirty sources, I think the market plays a role in like phasing those out. You know, regulation against coal, uh, for instance, um, there it's never going to be outright banned. we're too reliant on it, but it becomes regulatory uhly like harder and harder and therefore more expensive to have a quote unquote clean burning coal plant and then eventually the technology of you know best plus solar is cheaper and that's what really drives like a renewable energy revolution not us saying you know wagging a finger in somebody's face you know the the market the economy has a way of of taking care of that problem itself.

Let's talk about the 10% problem. So compute is currently concentrated in fewer than 10% of the world's countries. So does that change over time in adoption?

I think it does, but again I don't know what the sticking point like what the unlock there is. So if you really drill down into it, they haven't actually figured out how to monetize AI yet. And there's a lot of theories about where the most of the value acrru. Um right now there's a ton of value acrewing to people building data centers, you know, so people going out finding land purchase options, working with utilities or building generation, etc., etc.

And then the other value is really acrewing to the frontier labs uh layer. But there's a lot of um debate on how profitable that layer is and how long that will be profitable because like most of the compute of those frontier labs is still in the free layer even with all the revenue that Anthropic has made since they were the first to really decide that coding was the the real unlock. So there's a lot of people that think AI applications like the application layer uh will be where most the value uh acrru to but until they figure out exactly how to monetize it at scale it's really only a western phenomenon because uh if you can't really figure out how to monetize it here you definitely cannot figure out how to monetize it in the developing world. So until they figure that out, it won't that compute layer won't go to the developing world because that latency to that customer base doesn't matter.

But once they figure out, oh, this is how you really really make money from AI, that will drive where the compute goes because that latency is is always going to matter. That's a physics problem.

And I guess closer to your uh wheelhouse, does AI kill Bitcoin mining? So, uh, truth and lending, I I inserted that question in there because I wanted to, uh, address that because, um, I've heard a lot of people say, you know, Bitcoin mining is is going to get killed by AI because they're chasing the same electricity. Uh, and that's just not true. So if you think about AI, it measures downtime, acceptable downtime in, you know, hours if not minutes on an annualized basis. Um, it also needs incredibly high uh bandwidth um internet, both upload and download to make that work.

Tons of tokens, tons of packets going back and forth. Uh so it's an incredibly non-flexible buyer of electricity that can only be placed certain places where you've got that fiber optic connection. Um you know so here in America you'll see the data centers kind of strewn about the interstate lines because that's where we laid a lot of our really heavy uh fiber optic connections. That's why places like Dallas are hubs of of AI HPC compute, but even random places like South Dakota because there's an interstate that cuts across it. They've got good energy.

Voila, you've got data centers. Um, Bitcoin mining, as I kind of alluded to in the very beginning with my uh refinery analogy, uh it's kind of like uh I've heard people call it the dung beetle of like uh compute because we'll take energy anywhere. And for me, kind of my second Eureka moment when it came to Bitcoin was this idea that since humans discovered and harnessed electricity, it's only valuable near people and because electricity dissipates over space and time and transmission is really really expensive. So you want to have generation as close to your customers as possible with Bitcoin mining though because you're selling you're essentially exporting your electricity over the internet. The Bitcoin network is a buyer of first and last resort of your electricity.

That means if I can get a low latency internet connection anywhere like that there is electricity I can monetize that electricity. So that means that uh from a geographic perspective I can go anywhere whereas AI HBC data centers are very restricted on where they can go geographically. They're also need extremely firm electricity because they have to be up basically all the time. So they need firm grid power and then very firm backup power. Whereas I will gladly take 80% uptime if it means I pay half the rate of electricity.

Uh so there's always going to be a market for that type of compute. And the more that AI explodes, ironically, the more the addressable market for Bitcoin mining increases because utilities will need more and more uh flexible load. And just to put a point on that, uh Duke University did a study a couple years ago where they said if you know 1% of all the uh electricity offtake in America could be curtailable and curtailable according to when the utility needed them to be curtailable, it would unlock a further 10% of capacity on the grid. But the problem is is that a hospital, a steel plant, a factory, they never want to be told when or when they uh can use electricity. Like it's just not possible.

And the small incentives that the utility offers in a demand response program isn't enough to make up for the bottom line loss that they're going to incur. Uh which makes, you know, flexible load, something like this, that much more important. the uh it by 2030 a stat here US data centers will consume more electricity than all energyintensive manufacturing combined that's including aluminum steel cement and chemicals and will drive almost half of the US demand growth um you know you've got the hyperscalers and the uh the AI companies that are building uh huge rollouts of of power generation but who's going to support uh the base load generation needs of I other industry and of humans.

So again, I believe that the market is going to find a way to make all this work because loaded in that statement is a bunch of incentives like people that can figure out how to solve that problem will acrue a ton of value either individually or to a company. So it's it's probably not going to be one solution. It's going to be a myriad of things. It's going to be things like better and cheaper battery technology. It's going to be things like, you know, building more and more electricity uh behind the meter.

Uh, you know, there's all sorts of yet to be proven but theoretically possible technologies like um data centers in space or solar capture in space and then beam down uh to Earth. All these things sound incredibly sci-fi, but loaded in that statement is a ton of incentive to figure that stuff out. Uh, and so I think the market will will figure it out. Uh, because they're incentivized to do so.

Very expensive. In a former life, I spent a lot of time researching space-based solar power and um, people have to pay for that and there's a lot of research in kind of NASA, Japan, Germany. Um but this is this is fascinating and I appreciate you bringing um fiber into the conversation too because I think as we think about the broader digital infrastructure stack these are not monoliths blockchain AI cyber security fiber optics even uh essential public infrastructure like water. they now all fit into an equation where as we just discussed humans, businesses and uh compute are now competing for all these resources to to operate. Um we've got a couple minutes left.

What didn't we cover? Um any sort of finer points you want to hit?

I mean I think obviously when you come to conferences like this everything is is focused on the application layer. We're just talking about quantum. We were talking about agents. All these things are really important, but they all have a realworld tie that we kind of uh abstract out of our brain. Um, and there's a debt on all this stuff that that comes due.

And we eventually reach a point where and we've actually already reached that with a lot of inference. I don't know how much you guys have seen it where you log on to inference and it's like, hey, they're throttling back usage or it's down today or something like that. that's just a that's a real world constraint of not enough electrons reaching GPUs to uh to push that. So I think this is the layer that um not a lot of people are thinking about but um there's a ton of money and brain power going into figuring out how to solve the electricity aspect especially as it comes to uh data centers. Um, I'd say the last uh other thing that it's worth mentioning when it comes to renewable energy, I talked about best and batteries being a solution, but there's a really big economic problem that needs to be solved there where the majority of the the money that battery operators make is by doing arbitrage.

They buy electricity when it's cheap and they sell it when it's expensive. But the problem is that's self-cannibalizing because the more that that happens, the market becomes efficient and that just undercuts your your margin. So there's not one silver bullet. It's figuring out all this stuff in in concert.

No, it's complicated, fascinating, will require a lot of capital, time, and commitment. But thank you for being with us uh today at Boston Blockchain Week. This is a fascinating conversation. Uh please join me in uh giving Jake a round of applause. Thanks.

Thanks, man.