Energy Symbiosis: Powering Compute for AI & Bitcoin Simultaneously | Bitcoin 2026

It's great to be here uh at Bitcoin conference. Uh excited for this panel. Uh we have a lot of uh um great great minds here on stage. So excited to talk about this uh this topic. My name is Jesse. I work at Bloxbridge Consulting. Uh we advise uh bit public bitcoin miners and ahc uh companies um in the space. Uh I wanted to start with just a brief uh introduction of my fellow panelists. Uh just a little bit of an introduction. Who are you? What do you do? and uh a fun fact about yourself. Uh we'll start with you, Billy. >> Yeah, for sure. Thank you, Jesse. Uh my name is Billy Boon. I'm with Simple Mining. We help investors get exposure to Bitcoin mining. We have 10 data centers in Iowa. We're really good at running computers and building facilities to run computers in Iowa. Uh interesting fact about myself. I uh I've had a lot of people come up to me and say, "Wow, you're way taller than I thought you would be." Cuz all these digital people I see online. And uh I've actually won a dunk contest before, so it helps to be tall for stuff like that. >> All right. Well, I'm Rick and uh my background is HPC, artificial intelligence. Back in the day, starting back in 2010, helping develop products for Nvidia and other other companies. Uh last 10 years, I've been doing nothing but liquid cooling. Uh doing consulting for different companies, helping bring different type of products to market for multiple companies. And uh fun fact about myself, I'm also a musician for most of my life and uh yeah, drummer. Uh David Turnis, director of flexible compute platforms at Se Power uh got really involved with the mining sector in 2018 as they started showing up in Niagara Falls and Buffalo. And from a grid reliability standpoint, we realized these are the greatest demand response assets that the grid has ever had. uh because they can curtail in minutes to seconds in the uh when you know when when power is constrained. Um so we're really excited for this this moment this discussion here. Uh fun fact uh I too uh well I should say uh people I I'm actually shorter than people expect me to be and I too am a musician so uh we have all of that in common. >> I love that convergence of uh of fun facts. Uh thank you. Um yeah so getting to the to the to the topic of uh of this panel um I think you know anybody who hasn't been living under a rock has has been able to see that there has been uh quite quite a trend in this past year year or so um Bitcoin miners are slowly or not so slowly uh transitioning into uh the AI HPC space there's obvious uh obviously a massive demand coming from uh from all of these u uh these new demands from um computational persp perspective uh chaptt entropic what have you. Um and so a lot of people just ask uh are these two worlds are these two computational loads actually compatible? Are they competing for the same power? Uh is there a way for them to work together or is it that one is going to displace the other one? And uh I think uh the three panelists to my left uh have a have a take on this and can talk a little bit about this. Maybe we can start with you uh Billy. Um are are they are they compatible? and they live together and coexist. >> Yeah. I mean, I would say that they generally cannot live together because they're competing for the same thing. I mean, the common denominator between an AI company and a Neocloud and a Bitcoin miner is they need physical space. They need land to host their equipment, their physical operations, and they need some sort of power source. And so, that's the common denominator between the two. But it starts to deviate directly from there in terms of of cost. Uh the the cost to build an AI facility is is 10 times the cost of running a Bitcoin mining facility. They're much less capex and opex sensitive. Like for example, Bitcoin miners can can afford maybe $40 a megawatt hour whereas an AI company can afford $80 $200 a megawatt hour. So they're they're much less cost sensitive and and ultimately if like what we're seeing on our end u from companies that want to invest in in training in inference training the models there's an unlimited amount of demand like the the issue is not the demand the issue is the physical constraints like getting a transformer ordered from China gets delayed for 6 months or making sure you get the right permits or approvals from the city council. those those are the issues. And so I think that ultimately they're going to they're going to be able to outbid, but at the end of the day, miners will probably end up getting pushed away from those places where they get outbid by AI companies and these NeoClouds. >> Uh do you agree with that, Rick? >> There's definitely valid points there. The concept of of uh you know, some of the some of the larger miners out there, of course, uh they're all starting to move towards the AI space. Uh now the problem with that is that their investors and everybody else uh they're already expecting their returns. They're expecting a certain amount of return. Their ROI, the amount of time that it takes, they're not going to be able to, in my opinion, turn them off immediately. They're going to have to make that payment. They're going to have to make that that capex return. >> You're saying the miners are not going to be able to turn off immediately? >> No. Because people are expecting their returns that that the investment for the companies that are actually providing that funding for those for those mining companies unless you're unless you just personally uh vested into those miners and you sell them off. And so there's going to need to be a little bit of shared space for a little while until you're satisfied the people and you're not getting sued. Um but you know >> it's just liquidation. Like it'll come down to how fast they can liquidate it. Because if you could liquidate your entire fleet and you have a buyer for it, then you have probably investor like you have your investors who invested in funding your Bitcoin mining operation. But if you can get funding from someone else who can pay more money to reconvert it to an AI site and they're willing to help you liquidate your fleet and get it clean cut so that you can start putting GPUs in there, then that's where the difference will come in. And I think we're already starting to see that with pub codes that are making a complete 360°ree pivot. >> That is the sweet deal. >> Awesome. Um, I wanted to talk a little bit about uh specifically the coexistence or or the the AI HPC and Bitcoin mining. U but from a from a grid perspective and I know David has a has a lot of thoughts on that uh in that regard. Um so you you do a lot of that work when it comes to like the relationship between these data centers and these the Bitcoin mines and and the grid. Um, how are you thinking about this these two different types of loads when it comes to the you know air cuts but I know there there's some other power markets uh not just cuts that are that are interesting for them and the the grid monetization opportunities that uh that are there. Yeah, just to just to echo the the sentiments of um of Rick and Billy, uh certainly the publicly traded miners who have a fiduciary duty to you know for profitability. We are seeing a hard pivot to AI hosting there. The smaller and the midsize miners uh are taking a more a slower more measure measured approach. we are seeing more of a load shift but um still it's a hybrid load mining uh and then just starting to add some GPU um architecture there um from a grid standpoint if you look at Senate Bill 6 in OT which mandates all large loads over 75 megawatts to be flexible flexibility is on the mining side is actually really um valuable for you know if you're trying to get more capacity for your AI side of your house. Uh when the utility and the ISO can look at the flexibility lever that has like that h that does have value and then there's all of the uh demand response uh revenues. If we look at my state in New York ISO I mentioned this in an ear earlier panel you have access to ancillary services now with the deer model day ahead and real-time markets. So you can actually sell your megawws back in near real time when it's above your uh profitability strike price to mine and in some cases drive your blended cost per megawatt for the month into negative territory in the most expensive peak months. So we're kind of we're thinking about those kinds of models of having exposure to ancillary data in real-time markets fully automated response in seconds at the even being able to do regulation as the future state of an AI factory when we know we have all of the flexibility levers that we're going to talk about later. But um so far what we're seeing on our side is the uh the value of DR for the flexibility side on the mining side is keeping mining on these hybrid sites for now. >> Um yeah, I actually would like to to double click on that for for a second. You know, being at the Bitcoin conference, I'm sure a lot of people are kind of familiar with um with the the nature of Bitcoin as a as a very very flexible load uh that has been able to like be very responsive in particular in Texas, but elsewhere as well. Um and most people actually think that AI is much more it's much less flexible um than it might actually be and I know you have thoughts in that regard. I know you have these concepts of walls of batteries surrounding surrounding AI AI facilities. Uh I know that you have some thoughts on inference. So maybe you can share a little bit more about that and the flexibility of AI and the future in that. >> My favorite topic. So been preparing for this kind of convergence of these two loads for 2 years, 3 years I guess uh since chat GPTt uh took off. So the way we're envisioning flexibility for AI factories, um there's three different flexibility levers. Uh power util power, well, first of all, you have all of the transient peaks, right? Very spiky load that UPS used to be able to handle. Now we're thinking about sure super capacitors, uh solid stage capacitors, but battery storage, uh acting as a as a like a a load smoothing effect. Now you have like unstable utility power in PJM. People are complaining about uh unstable power. It's not pure enough to to to run a cheese, you know, a cheese making uh plant that's not going to be stable enough for your GPU architecture. So reason two for storage uh power quality from outside from the utility. Reason three regulation SB6 and Urkott having some kind of flexibility 50 hours a year maybe less. Um storage gives you that uh availability to deploy at scale. Our partners can bring hundreds of megawatts to storage to a site quickly. Um and then four, the value of energy arbitrage. Urkott is thinking about AI factories and miners, data centers as power platforms and even as future state utilities, their own uh storage, you know, like utility sales, storage and bring your own generation. Um, so that's how we're Oh, and then we can flex at the white at the uh at the rack also. You've probably seen uh software companies like Emerald AI, Benthouse and Super Micro, Mercury, Neuralatt, uh others. Um others I'm I'm I'm forgetting, but they can actually flex a GPU cluster running inference loads at scale without while without without stopping the workload and maintaining the SLA and the quality of service. So outside with the storage inside at the rack basically like hash rate optimization right for large uh inference clusters and then um we we have one of the true pioneers and and and innovators in the cooling space. I'd love to hear u uh Rick talk a little bit about flexibility in direct liquid cooling. >> Yeah and and that was going to be my my next uh my next question for Rick. um uh indeed in the cooling space and generally speaking in the buildout phase of these data centers. Um can you share some some thoughts in that regard on how uh on how you can build out these uh these sites with this this specific um topic in mind and how to you know adapt to specifically to AI or to Bitcoin mining or uh to have optionality for for the two. >> Sure. the uh this this is a deep talk but so I'm not I'm just going to do a flyby on it but uh honestly the the differences of trying to power these things has been made a point it's massively more expensive to do true power redundant power redundant data battery backup generation all all those different things that really make a true AI data center now if you start putting mining equipment inside of there how do you make that money back if you're if you're just you know in hard times if scraping by and you're not making the money that you want to be making or you're in the negative. Having that massive amount of money that you're spending because you have to have twice as twice as much power. They're both got to be lit up at the same time your kilowatt charge and everything else. It really wrecks the whole concept of Bitcoin um well of making money in Bitcoin. And so uh making things more efficient, making it smarter on the inside of the data center in the conversion going from a high voltage AC uh or high voltage DC directly to the to the system so you're not doing any conversions, there's no losses. Um using liquid cooling as an example is a massive massive right move in the right direction. Um getting your PUE, removing your water usage, all the things that actually cost money that take the dollar out of your pocket. These are all the things that really really start to hit home and start to start to generate the the real line sight of where this is how you really make money. It's dotting the eyes, crossing the tees of of the planning and knowing what you're doing from start to finish to make sure that at the end of the day you're actually making the money that you're expecting to make, that you're not losing that money, that you're not running hotter. As an example, uh a lot of people are still even running the old J Pros. If you go from a a 20° Celsius inlet temperature all the way to a 45 degree on inlet temperature if you're running it in Texas, you think that you're getting that, you know, whatever it is, 28 to 32 watts per terash, but you're not. As soon as you go above that 25° C, that watt that watt per terahash really starts to dra drastically change and you're not really making the money you're making. And so, uh, you got to you just take into account everything and, uh, you know, the proper cooling, the proper conversions, that sort of stuff. It really just does make a huge difference. >> I think one thing that I'll jump in on this is the way that I would look at it and how miners will come into play is related to the KPMG report that was published in 2024. It was Bitcoin mining as an ESG imperative. And the thesis there was that Bitcoin miners incentivize renewable energy buildout because they they don't have to have a 99.98% uptime like a lot of these these AI servers. they can be much more intermittent and much more flexible. And so when you go to a place like gridless for example, you go to a place where there's just water flowing off of a cliff and you have no idea if that project would make any sense, then you can almost run a pilot project with Bitcoin miners to build smallcale level of infrastructure to see if it even makes sense like is it does this even work? And that's where Bitcoin miners thrive because the cost of power is essentially zero. And then from there that can grow into potentially a much bigger site that eventually an AI company could take over because the infrastructure is already there. They have the networking there. They have all the components there that would make it a tier four, tier five data center, but it has to start somewhere. So it's almost seeding a site that may not be anything at this point in time, but eventually could grow into a bigger one. I love that you you just uh mentioned that KPMG uh report and I'm not going to call him out specifically but one of the guys who who worked on that is sitting right here at the audience. Um but uh >> welcome to predict the world is a market. Everything is a market. Get a 100% cash back up to $100 on your first predict bet if it loses. predict where everything is a market. I wanted to to go back uh for a second to uh Rick and and just ask you um what what do you think uh uh people building out the AI people involved in the AIHPC development uh are not learning from Bitcoin mining that they should be learning from when it comes to to the buildout and the and the cooling uh and building the racks. Um what are the lessons learn lessons learned that you think are missing there? Um and then I think you also have some thoughts when it comes to like specialization of hardware. Um and maybe you have some thoughts on that as well. Yep. Reality is the Bitcoin miners made the liquid cooling in the data center a reality. We were actually the pioneers in this space and you they would never in my opinion liquid cooling has been around forever. It's not until the very recently in the last two maybe two and a half years that you see Dell now starting to offer liquid cooling, HP, everybody else is now doing liquid cooling. You know, we got these little specialties, the places I used to work for in the past, but it's only in the last two years that it's really come into play. And it's only because all the Bitcoin miners are finally going, "Yeah, I can make this happen. I can make a reality." They've proving it. We we have our our failures and we have our wins and we have everything in between, but we've actually made it a reality for the AI side of things. and you know having having the ability to to mix and match you know back to what you brought up before being able to mix and match some of the Bitcoin miners with the HBC GPUs and stuff like that. Intelloflux over there. They got a nice tank sitting over there you can take a look at. They have the what's miner product in there. They have the the AI side of things all in a single tank. If you're gonna buy this infrastructure, make sure that infrastructure is flexible. Make sure it's modular so that when you're making those change if you want if you're already running Bitcoin miners right now, you're trying to get into the AI space, buy the right product. Get something that allows you to flex and move into where you want to be, not today, not tomorrow, next year, the year after that. Make the right purchase right up front. Sometimes it costs a little bit more upfront, but in the long run, if you're playing the long game, that's where it counts. >> Um, and Billy, uh, I believe, um, Neil Young at some point said, uh, my my, hey, hey, Bitcoin mining is here to stay. Um, and I I tend to agree with that. Um, can you speak a little bit about, uh, um, kind of where you see the future of Bitcoin miners and kind of being this frontier um, frontier industry? um and possibly the the the owners of the infrastructure, those who actually control the land and have the transmission and the permits and all of that and uh AI and HPC being more like people who just are hosting there. Uh is that is that accurate? >> Yeah, I mean I think to your previous question, the one thing that that AI companies could learn from Bitcoin mining companies would be vertical integration. So owning owning the land and the power and the having control over the power purchase agreement that your operation is running on because I mean there's no shortage of mining companies that were running in a jurisdiction that wasn't favorable or they didn't have complete vertical integration or control over their power purchase agreement and it blows up their operation because their entire business is energy arbitrage. You your input of buying electricity and converting it into Bitcoin there has to be a spread there. So if there's there's not a spread or if your power purchase agreement changes or if the government imposes some sort of exorbitant tax then that blows up your entire operation. >> Thank you. Um I just wanted to use these last few minutes uh uh to hear from all of you a little bit more about uh what do you see happening in the next uh in the next couple of years when it comes to both Bitcoin mining and and AI HPC uh integration. You all have your own different uh specialties. Uh uh Billy, maybe you can talk a little bit more about hash rate, difficulty, and all of that. >> Yeah, I mean I I've I've written ad nauseium about how I think difficulty is eventually there's got to be some headwinds on difficulty. There's got to be some headwinds on on Bitmain coming out with another machine that's going to be a 30% efficiency improvement. I mean, they are closing in on zero. Like the reality is at this at this point in time you can't convert zero watts or zero jewels into some sort of terash. And so you go back whatever 2016 2017 you're at starting starting around 150 jewels per terash. That was what miners a mining machine was able to do was take in 150 watts or 150 jewels and turn it into a trillion hashes. And so now you have like the S23 which is like 9 and a half jewels per terahash. significant improvement from from one and it didn't just go from there. It was it was much more linear. But now like what they'll go from from 9 and a half to maybe they'll go to five. It's going to be a much it's going to be much harder when you're when you're getting into like quantum tunneling and making chips. The the size is just so much like two n you're working on two nanometer chips is like the competitive chip right now. As you start to get smaller and smaller and smaller, it's going to be harder to to make significant improvements like it was in the past. And so that should be a headwind on difficulty, which for those of you who don't know, difficulty is by far the variable that plays the biggest role in hash price, which is the revenue that miners could expect per unit of hash. So you have that headwind, and then you also have the headwind of competition from AI companies. Like before, there weren't these like massive hyperscaler NeoClouds that wanted gigawatts of power. They want the same thing that miners want. So they're going to be able to outbid us. They're going to be able to pay for money until Bitcoin goes to a million dollars a coin in a very short period of time. And so if they outbid us, where do the miners go? They're keeping the land and the power. They're not they're and they're going to put GPUs there. So where do the machines go? They're going to go to the Peran Basin where electricity prices are negative. They're going to go to hydroelectric dams where you can get 2 to 3 cents per kilowatt hour. They're going to go to places that have stable curtailment power purchase agreements that are demand responders that are strategic for utility providers in a local community. And so yeah, I mean that would be my two cents on it. >> It all goes back to power for sure. But inside inside the actual data center um Rick um next few years you're you're you're working some on some exciting stuff. I don't know if you can disclose all of the details, but um what do the next few years look like uh in terms of efficiency and racks and cooling? >> It's true. the uh you know making these chips smaller and smaller making them more efficient more effective creating more terra hash for less wattage that's going to keep on happening over and over again but the technology is changing um from the AI AI side specifically uh I don't know if I should talk tell a little story about um you know to break break a concept open but as an example you have you have the the Nvidia chips which are really fantastic they're great the 100 200 300 uh all the stuff that's coming out. Uh they're fantastic. They're super fast. They do everything top to bottom. Literally, if you want to do inference, if you want to do training, you want to do whatever you want to do to it, it does it. And it does a fantastic job at it, but at a price. You're talking about a 500,000 700,000 server. Now, there are different ways of actually getting to that same spot where where you're making those those tokens at the end that are worth the dollar, same as hashes. And uh as an example, if uh if if my mom were picking up the kids for ice cream and go and eat groceries, she'd come in her vehicle and she'd pick them up, she'd go get some ice cream. She go and pick up some groceries for pick up. You say, "Oh, real nice to her son. Going to get two weeks of groceries, put in the back of Minivan or whatever it's going to be. Drops us off. She leaves and then she goes to NASCAR. And at NASCAR, she gets into a race. She jumps in that vehicle and she wins NASCAR with the same vehicle. This woman is tireless. She now goes jumps on a plane with that same vehicle, shipping it over there because she's going to go to Dubai. She's going to win the trophy truck race with the exact same vehicle. And then the next day, she's going to come and she's going to pick up the kids again and go grocery shopping again. That would be one expensive vehicle. That would be one of the most expensive vehicles you could ever think. Can you imagine? That is what Nvidia's done. And it's a great chip. It's a very expensive chip. Now the problem is and that's not a problem. Inference running llama as an example. There are specific tasks, specific things in the driver, specific things in the firmware, spec specific things that the way it actually communicates on on the interconnect on the motherboard and everything else that occurs. That's all going to change. If uh if if one one vehicle was the best vehicle to do everything all the time, then Ford, Chevy, nobody would have a different vehicle. it would be one vehicle from that manufacturer but it doesn't it doesn't work like that because there are tasks specific tasks that this vehicle is good for that that vehicle is good for something else and that I believe is going to be coming instead of having these dedicated chips where it's it's a onesizefits-all I believe that that's that's going to be coming in the very near future and yeah it is something I am working on >> getting increased Nvidia competition is what >> massive massive more increase >> love it thank you Um, love to see grandma driving at NASCAR. I love the analogy. Uh, David, final words. Um, and and if you can talk a little bit about the the next few years on your end when it comes to grid and flexibility. Uh, and if you also want to talk a little bit about the the recent announcement you had in February with the few other companies, uh, that would be great. >> Yeah, I mean I I get excited about this topic. Look at all the innovation the the other panelists have have pointed at direct liquid cooling immersion miners the the wildcatting like sense of the the the spirit of innovation on the mining side will continue to move to center to AI factory uh development and design because miners understand uh energy in a very sophisticated way. You have traditional data center operators, your coloss, who are really real estate investment trusts. Um, energy never had to know about it. They pass through energy to their tenants as a straight pass through. This is a major advantage in my opinion. Miners will as we as we meet in the middle with AI, miners have a clear uh clear innovation advantage and I'm really excited to see what's going to happen between now today and 2030. uh because we have to get through till to to to then until we can build more power plants, right? So flexibility, demand response, flexibility as a as a design feature and and energy arbitrage and demand response give you freedom of location. We can you know DR gives you that 4 cent power and various uh markets where DR exists including New York which people tend to avoid sometimes negative energy pricing. So just the understanding of energy arbitrage, flexibility, bringing the spirit of innovation to center with AI factory design. U I'm really excited to see what's going to happen and I think we'll I think we're going to solve this capacity problem that way with this group of people. >> Love the optimism. Thank you. Thank you everybody for uh for sharing your thoughts and thank you for the audience for being here. Right. >> Thank you Jess. Thank you very much. Every year this community comes together to celebrate, to debate, to build what comes next. And every year the stage gets bigger. Sound money center stage. So where do you go to celebrate the next chapter in Bitcoin history? You come home. Nashville, July 2027. Thank you very