Nick Burley, VP of North America for Fog Hashing, used his keynote at Mining Disrupt Dallas 2025, recorded November 12–14, 2025, to make a blunt case: AI is taking energy capacity away from Bitcoin mining sites, and the operators who come out ahead will be the ones who can get compute running on a site fastest. Fog Hashing builds data centers, he said, and its pitch is a cooling and infrastructure layer that lets a site cash-flow on Bitcoin today and add AI capacity when a contract lands.
Burley opened by asking the room to applaud the Mining Disrupt team for pulling the event together in a year, then went straight to the market. The value of an energy site, he argued, is directly proportionate to the resources already in place that can be deployed right away. The time it takes to drop compute onto a site is, in his words, a very large concern in the industry right now.
Why AI Is Eating Bitcoin's Energy Capacity
Burley did not soften the premise. He described a market in which two industries are converging and the newer one is winning the fight for power.
AI is coming in and eating capacity in the Bitcoin market for energy sites.
Nick Burley · 0:47
He said people push back on the idea of taking a GPU cluster and dropping it into mining infrastructure. His answer was that Fog Hashing has run the models and builds data centers, and there is more compatibility between the two than people say. There are nuances in the parameters, he conceded, but the question he put to the room was about revenue and the cost basis to act on it.
Do you have an opportunity for three, four, five times revenue on your site?
Nick Burley · 1:51
Having something in situ that can take advantage of that opportunity matters because, as he put it, time is money and nobody wants to stop. Other regions in the world are scaling, he said, and keeping model performance on a linear trajectory means compute has to expand exponentially. He compared it to the doubling penny: by day thirty the total is in the millions, and that is the growth curve the industry is now on.
The Deployment Gap: Months of AI Negotiation vs. Plug-In Hashing
The infrastructure layers of the two industries are relatively close, Burley said; the differences sit in the technology and the market. Bitcoin is much easier to deploy. Units arrive on site, power goes in, and there is a customer ready to go as soon as you plug in. AI does not work that way. Negotiating an AI contract with a group can take months.
That gap is dangerous for an industry that runs, in his description, on a capital model of get the site, get it online, and understand your run costs over time. If an operator allocates millions of dollars toward infrastructure and the client side is not tied down, the result is what he called a massive bleed. The last thing you want, he said, is to deploy equipment on a site and then negotiate for 18 months on a contract.
Legacy Hardware Follows the Bitcoin Pattern
Burley expects the parallels between the two industries to show up less in the technology stack than in the market stack. In Bitcoin, legacy equipment moved to lower-cost jurisdictions. He expects the same trajectory in AI: depreciated hardware that still has an operational base to serve a function in the market, but not frontier training.
It's not going to be a circumstance where those units are going to be sitting in that data center.
Nick Burley · 4:56
Later in the talk he returned to the point. As legacy machines come offline, those GPUs are not going to be worth zero. A revenue stream built on easily deployable compute will not carry the same parameters on an inference load as on a training load, but it is still revenue, and the question for a site owner is how to bring the higher stream in quickly.
Tents, Talent, and the Supply Chain Pinch
Burley pointed to the shift from ten-year engineering projects to tents built to house billions of dollars of GPU compute as a sign of where the supply chain stands. The industry is running into its pinch points, he said: there are not enough trace people and not enough talent. He described one mega project with 4,000 people on site, everyone crawling over everyone else, each group running its own Gantt chart, where a single miss ripples through the whole schedule.
On the energy side, he said supply is struggling to match demand. People are talking about daisy-chaining Mack trucks together for power, while other jurisdictions put as much new load on their grids every month as North America manages in a year. The question he kept returning to was how to take what you have and maximize its revenue potential.
Chillers, Dry Coolers, and the Jackknife Site
Electrons travel copper and aluminum regardless of the circumstance, Burley said; the difference is on the cooling side, where the integration layer has to work across a regime that includes 115° temperatures and the other end of the range as well. Fog Hashing's answer is chiller units with dry coolers, which he said give a site the capability to serve both AI demand and hydro Bitcoin mining.
You then can add modules to the equipment layer like a chiller to make sure you hit the parameters needed for your AI side.
Nick Burley · 9:10
The sequence he described starts with a Bitcoin client: bring the unit on site, get it online, and start cash-flowing. When an opportunity to negotiate on capacity arrives, there is already something running, and modules such as a chiller can be added to hit the AI parameters. The temperatures are different and it is not easy, he said, but a tech stack with jackknife capability is more than a transformer on site and a hope that it works. It gives a site a chance to pull down load that would otherwise go to higher-cost jurisdictions.
Exponentials are exponentials. And when you see this line on a graph, there will be a pinch for power.
Nick Burley · 10:26
Burley closed with an invitation. Fog Hashing has clients looking for GPU space and clients looking for mining hosting, and he said the company can put site owners in contact with both and help build out the infrastructure. His warning was that a pinch for power is coming, and the question for every operator is how to capture that revenue stream in the smallest amount of time.
Bitcoin has moved since the session. On November 12, 2025, when it was recorded, Bitcoin traded at $104,994. On October 12, 2026, it was $83,419. The conversation about how mining sites take on AI load continues at Intelligence Disrupt 2027.