Bitcoin Miners Are Becoming AI Infrastructure Companies. The Market Is Noticing.
Hive raised $115M for GPUs. Keel rebranded away from Bitcoin entirely. CoinShares projects publicly listed miners could generate 70% of revenues from AI by December. This is not a trend anymore. It is a structural shift, and the reason it makes sense has everything to do with what mining infrastructure actually is.
Something structural is happening in the Bitcoin mining industry, and the stock market is pricing it faster than most people realize.
On April 22, shares of HIVE Digital Technologies rose sharply after the company closed a $115 million private offering of exchangeable senior notes earmarked specifically for GPU purchases and data center development. On the same day, Keel Infrastructure, formerly known as Bitfarms, rebranded in February 2026 after declaring it was "no longer a Bitcoin company" and announced the sale of its Paraguay site and closed fresh capital for high-performance computing builds across North America.
Both stocks moved double digits. The market is not confused about what this means.
CoinShares projects that publicly listed Bitcoin miners could derive up to 70% of their revenues from AI and high-performance computing by December 2026, up from roughly 30% today. S&P Global estimates HIVE alone could see HPC revenue hit 15% of total revenue this year, up from 7% in 2024.
The question worth asking is not whether this pivot is happening. It clearly is. The question is why mining infrastructure translates so naturally into AI infrastructure, and what it means for operators who are already running the hardware.
The Infrastructure Is the Same Problem
AI data centers and Bitcoin mining operations share three fundamental requirements: access to large amounts of power, the ability to handle extreme heat density, and physical space to deploy and operate hardware at scale.
For most of the tech industry, those three things are the bottleneck. Power is constrained by grid access and permitting timelines that run years, not months. Heat density is a growing crisis: average data center rack power density has risen from 8 kW in 2021 to over 50 kW for AI workloads in 2026, according to reporting from Data Center World 2026. Liquid cooling, once a niche consideration, is now a baseline requirement for high-density AI systems.
For Bitcoin miners, none of those are novel problems. They are the job.
Mining operators have spent years negotiating power contracts at scale, building relationships with utilities, and locating facilities in jurisdictions with favorable energy costs and regulatory environments. They have built or sourced liquid cooling infrastructure because the efficiency math demanded it. They understand load management, uptime requirements, and the operational discipline of running dense hardware continuously.
When an AI company needs to deploy 100 MW of GPU compute and cannot get a grid connection for two years, a mining operator with existing power access and cooling infrastructure looks like a very attractive partner.
The Energy Constraint Is Real and Getting Worse
The scale of AI energy demand is difficult to overstate. The International Energy Agency projects global data center electricity consumption will exceed 1,000 TWh by the end of 2026, equivalent to Japan's entire annual electricity usage. That is up from roughly 415 TWh in 2024.
The five largest technology companies collectively exceeded $400 billion in capital expenditure in 2025, with that figure expected to jump 75% in 2026, according to S&P Global. Most of that capital is chasing physical infrastructure: land, power, cooling, and the engineering to put it all together.
The bottleneck is not money. It is access. CNN Business reported in April that up to 11 GW of announced data center capacity remains stalled without construction underway, with half of global projects delayed by power limitations and grid equipment shortages.
Mining operators who have already cleared that permitting and infrastructure hurdle are sitting on a resource that the AI industry cannot easily replicate on short timelines.
Who Is Making the Move and How
The approaches vary, but the direction is consistent.
HIVE Digital Technologies is the most aggressive example. The company operates roughly 300 MW of hydro-powered infrastructure in Paraguay and is treating that as the base for an AI expansion through 2027. The $115 million raise is going directly into GPU fleet build-out. HIVE is not abandoning Bitcoin mining. It is layering AI compute on top of existing infrastructure that was already paid for and operational.
"With Bitcoin's next cycle and AI demand surging, HIVE's dual engine model is positioned to capture both supercycles in real time, with cash flow from Bitcoin operations driving exponential HPC growth."
— Aydin Kilic, CEO, HIVE Digital Technologies (via TheStreet)
Keel Infrastructure (formerly Bitfarms) took a more definitive position. The February 2026 rebrand was accompanied by an explicit declaration that the company was repositioning as "infrastructure-first" for HPC and AI data centers. The Paraguay site sale was the first step in a broader portfolio rationalization. Keel is not running a hybrid model; it is completing a full transition.
Core Scientific has pursued a middle path, signing large-scale co-location agreements with AI and HPC customers while continuing to mine Bitcoin on its existing fleet. The model allows Core to generate HPC revenue from its power access without fully liquidating its mining operations.
Each approach reflects a different read on the same underlying thesis: mining infrastructure has strategic value to the AI industry that the market has not yet fully priced.
"We are no longer a Bitcoin company. We are an infrastructure-first owner and developer for HPC/AI data centers across North America."
— Ben Gagnon, CEO, Keel Infrastructure (via CoinDesk)
The Hydro Angle
Not all mining infrastructure translates equally. Air-cooled operations face a meaningful barrier: liquid cooling is now a prerequisite for the highest-density AI workloads, and retrofitting air-cooled facilities is expensive and operationally complex.
Hydro-cooled and immersion-cooled operations are significantly better positioned. The liquid cooling infrastructure is already in place. The heat management expertise is operational, not theoretical. And the power density that hydro-cooled mining achieves maps directly to what AI compute deployments require.
This is relevant for Ultra Labs specifically. We run 58 hydro-cooled Bitcoin miners in the Pacific Northwest on S21 XP Hydro hardware, with plans to quadruple that footprint over the next year. The liquid cooling infrastructure we operate is the same class of system that AI data center operators are now scrambling to build from scratch. We did not build it for AI. But the operational capability transfers directly, and the scale we are building toward puts us squarely in the range where AI co-location partnerships begin to make economic sense.
The Question for Operators: Pivot, Hybrid, or Stay the Course
The decision facing every mining operator right now is not simple.
A full pivot to AI infrastructure (the Keel model) makes sense if your power access and facility footprint are large enough to attract serious HPC customers, and if the AI revenue premium over Bitcoin mining margins is sufficient to justify the transition costs. At 50–100 MW scale, that math increasingly works.
A hybrid model (the HIVE and Core Scientific model) makes sense for operators who have excess power capacity or unutilized facility space, and who want to generate AI revenue without abandoning mining upside if the Bitcoin price moves significantly.
Staying the course makes sense for smaller operators who are optimizing for mining margins, running at high efficiency, and do not have the scale to attract enterprise HPC customers. Not every mining operation needs to become an AI company.
What does not make sense is ignoring the shift. The infrastructure premium that mining operators have built over years of operational experience is, for the first time, being valued by an industry outside Bitcoin. That changes the asset calculus for anyone thinking about capital allocation over a five-year horizon.
The Bigger Picture
Bitcoin mining was always fundamentally an infrastructure business. The compute was purpose-built, the economics were driven by energy efficiency, and the operators who survived multiple market cycles were the ones who treated it as an industrial operation rather than a speculative play.
AI infrastructure is the same business at a different scale and with a larger addressable market. The skills, the relationships, and the physical assets that mining operators have spent years building are exactly what the AI industry needs and cannot build fast enough.
The miners who recognized this early are being repriced by the market. The ones who act on it over the next 12 months will be better positioned regardless of where the Bitcoin price goes.
Ultra Labs is a US Bitcoin mining and technology company, operating hydro-cooled miners in the Pacific Northwest and building on Cardano and Midnight. Read more: Bitcoin Mining in 2026: Air, Hydro, Immersion, and the Bitaxe Revolution · BlackRock Is Testing Its Tokenized Fund on Cardano