L’évolution du minage de Bitcoin vers l’IA | Bitcoin 2026

In 2022, uh we were building a project, a data center project in Florida. Everything was going great. Um who doesn't like a project in Florida, right? But then late September, uh I get my phone start blowing up. It's our customer calling us. Hurricane Ian hits mainland. So in the room, who who remembers? Because I sure do. We're talking category 5 160 mi wind about $112 billion in damage. So next morning speaking to our customer um it's chaos there people everywhere on boats on seedos trying to help their neighbors. We all understand it's a crisis and priority is making sure everybody's safe and in good care. But here's the thing. The world doesn't stop. It doesn't stop for for much usually, right? Even in these crisis. So there's still projects to complete. Show must go on. That's what AI demands. So can your field construction and operation keep going in a hurricane? Your construction's halfway through. What survives? How many months of delays and repair? Well, we have some technical difficulty here. I don't have any notes on uh screen anymore, guys. Where's my notes? Where is my notes, guys? All right, let's keep on going. So, if your operation can so if operational, can it keep running? Is it rated for the area where you deploy? Do you have redundancy and concurrently maintainable system to meet your customers expectation? And when I say hurricane, I don't just mean a literal hurricane, right? Like I'm talking about rain, mud, dust. We're talking extreme hot and cold weather pattern that we're all facing on construction site. So I can't jeez there's there I can't see my note guys so I'm not sure what's going on in the back but so what about supply chain what about interruption in supply chain interruption in water supplies I've seen project crippled for less than that, right? And here's the thing. During the hurricane, we never stopped. We never stopped. We kept going. By the time they were doing cleanup, we were deploying phase two of the project. Guys, my slides are not moving. I can't see my notes. So, my company Jesus My company infra my company builds infrastructure since 2011. So we've been part of the Bitcoin industry since 2019 and I've got to see firsthand the [snorts] bunch of the the the group of high achievers that we it's made of. And you find land, you find power, you put compute online faster than anyone. [snorts] And we've built lots of projects together which I'm very proud of. But AI demands more. We We're not just serving the blockchain anymore, guys. I don't see my notes. Like, this is I'm going to I can't. Sorry about that, guys. So, the good enough mentality and basic tier one infrastructure is simply incompatible to an AI transition. We'll need to follow strict guideline and operation process to make this project a success. So as we pivot and diversify to AI, let's consider four main areas of challenges. We're talking about speed, scale, flexibility, and quality. So, we're going to touch base on that quite a bit here in the next bit. So, speed, traditional data center are slow. We're talking 24 to 36 months of infrastructure to build them, right? That's without considering bad weather, supply chain, labor shortage, complex permitting. AI moved too fast for this. So if we follow Nvidia's GPU road map, your infrastructure will be a dinosaur completely outdated by the time it's constructed. Scale stick build data center project are tying up large amount of capital upfront inflexible right it's large building infrastructure with complete with fixed centralized power and cooling systems. This is all based on the available GP at the time of the design, right? So, we're talking about three years ago. So, this is really something that doesn't allow you to grow in phases and and and grow as as your customer demand comes in. Flexibility. So, traditional data centers are inflexible in nature. uh you think you could retrofit new generation of GPU as they get released but you have to think twice on that. So BTC minor upgrades follow the having every four years, right? So in between you're going to have minimal hash rate upgrades and and you know very minimal changes on the hardware. You don't have to retrofit your system as you go. But AIGPU evolves every 12 months. And we're not talking simple optimization here. We're talking doubling the density of the compute, potentially changing the form factor, and you're going to be shifting in voltage as well. So those are major changes. At this rate, you would be in a constant retrofit pattern. So you need to get ahead of the hardware cycle. You need infrastructure that allows flexibility for AI densification. Let's talk quality. We all know that the infrastructure quality in the Bitcoin mining industry varies a lot, right? Maybe a little too much. A lot a lot of time it's capex constraint or the good enough mentality out there. But now we have a whole new set of rules to go. AI customer will have different but really strict expectation. So we're talking quick deployment, no single point failure, concurrently maintainable systems, highest quality component. We're going to go with we'll have to follow Nvidia reference guideline on top of that, which is not easy. This is a completely different ballgame. So here's the response to our challenges. Modular, a fully integrated modular system that includes everything from utility connection through your white space, your cooling plant, your power generation, power transformation. You're going to have your control room, your warehousing, a true modular AI factory built in a controlled factory environment. So, let's just let's bring this back to the hurricane story for a second. The only way we maintain operation during the hurricane and delivered a project on time is a high quality pre-engineered modular system engineered and built for the environment that was deployed in. So in that case hurricane rated. So moving forward let's think hurricane wind loads high elevation operation seismic rating. So manufacturing facility with lean processes and quality process will always win the race. So speed, scale, flexibility and quality. So collectively let's start thinking modular. That's how you control all the variables. Now let's address a few of our challenges. So lot of problem out there with this but the speed let's forget 24 to 36 months right this is giving me anxiety just thinking about it let's keep this number in mind it's 12 months and under this is how you're going to attract customer and this is how you're going to make your project successful so modular infrastructure transform data center delivery from a construction project to a product line with predict predictable able supply chain and manufacturing schedule. Quite different than your construction. It allows parallel parallel path between site construction and manufacturing activity. That's how you're moving fast. And the only path to AI ready infrastructure at the speed the market demands is modular. So field construction simply can't keep pace. Grow in phases, scale as you grow. No guesswork, no overbuilding, no stranded capital and large building infrastructure, right? You start generating returns under 12 months. The scalability of a modular system is simply unmatched. So scale block by block without disrupting uptime and operation of your previous phases. Let's talk flexibility again. Keep your modular design flexible. You can't allow hardware densification to set you back with retrofit all the time. We're talking adaptable power cooling systems that can follow Nvidia's GPU roadmap. Power draws are leaping from 130 to 600 kW plus. This is incredible. Voltage are shifting from 415 AC to 800 volt DC. So inflexible infrastructure doesn't age. It it literally expires. So modular flexibility isn't an option anymore. It's how you avoid building tomorrow's outdated data centers. So quality building in a controlled factory environment allows for quality control factory acceptance testing and then you could follow all of these process and executed them uh consistently again mitigating surprises on site potentially costing you on quality and overall delivery of your project. But let me tell you this. Well, no actually now quality comes at a price. Typical BTC infrastructure range between 150 to 500K per megawatt. So this is what you see typically right a factory starts around 7.5 million per megawatt. So let me tell you this is a true this is the cost of an AI transition. Again, I don't want to sound redundant, but it it is a a completely different ballgame. This is not the same industry uh together, but there's ways to transition properly. So, let's go through the digital twin of our octopod solution as an example of a modular solution out there. So, octopods from Intellex and we have a digital twin we could play right now. So this is a full block of one of our solution. This is a true start to finish modular AI factory. So let's walk in through the first piece. This is a connecting module where you're going to bring all your medium voltage power distribution system through uh busway. This is where you house all your fire suppression system and control panels for it. And this is how you interconnect the next block when you grow uh your facilities. So we're going to go through a contained hot [snorts] aisle in the IT hall. This is where we have our craw. As you can see, they're a fan wall. What they do is they take care of all the residual heat that you have in your DLC cooling racks. And then they'll take care of all your lower density network racking and storage that don't have liquid cooling. So, let's let's turn around here and go right into one of our cold aisle. So, this is where we have um all of our facility water coming through. This is our ICDU, which is a distributed pump system. You're going to have a one:1 ratio with the IT rack with your GPUs and the CRA. And then this is all N plus1 redundancy. Again, as you can see, this is a facility, right? This is not containerized component. We're talking about a true AI factory that you could walk through, install, and operate your day-to-day operation and access every piece of equipment. Modular always brings repeatable, right? So, this is a second data hall mirror from the other side overall. in one of our factory we're going to have up to four of those which we're going to give you 8.8 8 megawatt of critical load it load. And how do we cool this? We're going to go through one of our cooling plant. What about our cinematic music that we had over that? Sounded really good, guys. So, the cooling plant is going to have four levels. [music] So we're going through here. The plant is made uh with chiller units on the first floor. So it's a six to make five configuration and it gives you all the uh water chiller cooling uh necessary to cool all your IT loads. We're going to go through in the middle where we have our utility modules where we have all our pump systems and secondary level is going to have other utilities like power distribution which we're going to go through after this. Again, you'll see the repeatable modules on the other side. It's again a mirror image from the other side. [music] You're going to have the three other chillers on this side. And we're going to walk outside and take a spin around to take a look at the overall cooling plant from the outside. We're going to go look at level three and four [music] of it. As you can see, very um large industrial modular, no no containerized component. All structural steel rated for the environment it's deployed in. So these are the dry coolers that we have on top. They work in parallel with the chiller units and they have the same configuration. So you have a six [music] to make five configuration. We're going to swoop around. Let's go take a look at the second level of just above our chillers in our IT hall. [music] Again, you need access to all your components. This is why we have walkways on the second floor. This is where we host all of our HVAC plenum. Uh we're going to have all our busway [music] system, tap boxes, uh cable tray for your networking. As you can see on the tap box here for uh the DLC [music] rack under four levels of cable tray for networking and fiber cable. Welcome to predict. The world is a [music] market. Everything is a market. Every headline moves the line. Every moment is your market. Call the moves. Bet on your instinct, your prediction, your edge. Dual bits predict where everything is a market. As you can see again, access to every piece of equipment. We're talking platforms to do all your preventative maintenance daily. And let's take a look at one last piece here. This is our mega box. This is where we do the power transformation from medium voltage to low voltage distribution which going to feed the data center from there. And that's just a quick overview of a modular data center which is the octtopod from Intelloflex. Uh it's an idea of how you could actually achieve uh this properly. So, uh, it doesn't matter how many times we're going through it, it's always nice to see this digital twin. Um, it just really puts it in perspective. So, here's what I want everyone to remember from this. Um, speed under 12 months, right? Scale. It's block by block by block. Don't strand your capital. Flexibility. Stay ahead of the hardware curve. In quality, build in a controlled manufacturing environment. So if you follow these four words, um it will guide you toward a successful AI transition or build out, right? And remember, modular wins the race every time. So speed, scale, flexibility, and quality. Thank you so much for listening. Uh truly looking forward to see you all on the floor. We're in booth 334 right across over here. We're displaying a lot of solutions. Um you could take the digital twin for a spin. Um thank you and have a great rest of the conference. [music] Every year this community comes together to celebrate, [music] to debate, to build what comes next. [music] And every year the stage [music] gets bigger. Sound money [music] center stage. So where do you go to celebrate the next chapter in Bitcoin history? You come home. [music] Nashville July 2027.