•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
The AI boom has increased demand for compute exponentially, pushing operators to secure cutting-edge infrastructure and high-efficiency technology to support grid resilience. As a result, the way electricity is converted into digital services is being reshaped—drawing parallels to an earlier playbook developed by Bitcoin miners.
This third installment in a series argues that Bitcoin mining and AI data centers rely on the same underlying physical system: chips, power, cooling, and infrastructure. The key difference is how companies position themselves within that system, creating a spectrum of business models that ranges from outsourcing to full vertical integration.
At the most basic level, participation in both Bitcoin mining and AI computing begins with deploying hardware. In Bitcoin mining, this typically means owning ASIC machines—specialized chips designed to perform a single task—and placing them into facilities operated by others to generate Bitcoin. In AI computing, the equivalent is deploying GPUs into data centers used to train models and run customer workloads.
In both cases, companies own the machines but not the infrastructure. That infrastructure is supplied by colocation operators, which provide power, cooling, and physical space to run compute at scale. The article notes that colocation is increasingly viewed not just as a hosting function, but as a way to monetize power and infrastructure.
Examples cited include Bitcoin colocation models where hardware ownership is separated from infrastructure operation. The article mentions ABTC deploying miners hosted through parent infrastructure operators such as Hut 8, and Cango operating fleets hosted in facilities managed by Bitmain.
It says a similar structure is emerging in AI. Fluidstack is described as deploying GPU clusters through partnerships with infrastructure providers including Cipher and TeraWulf, using existing power-connected sites to serve AI workloads without owning the underlying facilities. Bitdeer is described as developing AI colocation capacity at scale, including a planned 180 MW facility in Tydal, Norway that is in negotiations with a potential tenant.
As AI demand accelerates and power becomes the limiting factor, the article argues that infrastructure with existing grid access is becoming especially valuable. It adds that many sites originally built for Bitcoin mining are positioned to support AI and high-performance computing workloads, with an expected shift over time—particularly across the U.S. and Europe.
As companies move up the stack, the next step is owning the physical environment itself. At this stage, operators build or acquire facilities such as data centers, substations, and cooling systems rather than relying on third-party hosting.
The article says infrastructure ownership can change operations by allowing operators to control power costs, optimize performance, and reduce dependency on external providers. It also emphasizes that the value increasingly lies in the power connections attached to these facilities.
To illustrate the repurposing of industrial assets, the article cites Alcoa exploring the sale of idled aluminum smelter sites to digital asset firms such as NYDIG, and Century Aluminum selling a Kentucky facility to TeraWulf, which is pivoting toward AI and high-performance computing. It notes that many such sites faced shutdown due to offshoring of high-paying industry jobs, but share a critical feature: they are already connected to the energy grid at scale.
In this framing, interconnection—often described as the hardest and slowest part of building new infrastructure—becomes a valuable asset. The article argues this is bringing technical roles back to the United States and accelerating the country’s position in technology and innovation.
The article argues that even grid-connected infrastructure is finite. It says the number of industrial sites with high-capacity interconnections is limited and much of it has already been identified or repurposed by major industries. As computing demand accelerates—especially from AI—the constraint shifts from where infrastructure exists to whether the grid itself can keep up.
It adds that major power markets are responding to the complexity of connecting large loads. Regulators are revisiting how large energy users are integrated into the system, with the article citing adjustments in regions such as PJM and ERCOT and new rules and proposals governing how large-load data centers connect, how costs are allocated, and how reliability is maintained.
To address these challenges, the article says some operators are moving beyond the grid altogether. It points to a partnership between Amazon and Talen Energy, where data center infrastructure is being developed alongside nuclear generation capacity. While AWS is described as not owning the power asset outright, the structure is presented as aligning compute with a dedicated energy supply.
The article also describes similar concepts in Bitcoin mining. It mentions New West Data using power from flared gas at oil production sites to energize Bitcoin miners for additional cash flow. It also cites Greenidge Generation as the first power plant described as directly participating in Bitcoin mining in 2020, reviving an asset that would have been shut down due to lack of competitiveness in the power market.
In AI computing, the article says developers are increasingly partnering with—or building alongside—power generation assets, including natural gas, nuclear, and renewable energy. It frames the “bring your own power” model as turning electricity from a cost center into a strategic advantage by stabilizing pricing, ensuring availability, and aligning compute capacity with energy supply.
As an example from Bitcoin, the article cites Bitfarms, which historically operated as a self-mining business owning infrastructure and deploying its own computational power. After acquiring Stronghold, it says Bitfarms moved upstream into power generation, gaining direct control over energy assets, later rebranding to Keel Infrastructure to support a broader model for AI and high-performance computing workloads.
At the highest end of the spectrum, the article says some operators can control nearly every component of the compute system, including power generation, infrastructure, hardware deployment, and even chip design.
In AI computing, it notes that hyperscalers such as Amazon Web Services, Microsoft, and Google are developing custom chips, securing long-term energy supply, and building large-scale data center campuses tailored to their workloads. In Bitcoin mining, it says this model is already taking shape.
The article cites Canaan, described as the earliest Bitcoin ASIC designers with its Avalon brand, expanding beyond hardware into operating its own mining infrastructure. It says Canaan scaled proprietary computing power by deploying its own machines across sites it controls directly or through joint ventures, and that it deepened its strategy by acquiring Cipher Digital’s 49% stake in Texas joint ventures with WindHQ, a wind electric power generator.
It also describes Bitdeer as expanding from cloud mining and proprietary operations into greater control over infrastructure, scaling its exclusive computing power to around 70 EH/s. The article says Bitdeer has also moved into power generation, including acquiring land and a license for a 101 MW permitted plant in Canada.
At the same time, the article says Bitdeer is extending into AI processing by deploying its own GPU infrastructure for AI cloud services and exploring high-performance computing colocation opportunities with tenants. It argues that this dual expansion—up the stack into power and across into AI workloads—reflects a shift in objectives from efficiency to access.
The article concludes that Bitcoin mining and AI data centers are not separate industries so much as participants in a shared system with multiple points of entry. It says they differ in workloads and customers, but structurally operate along a continuum of ownership—from asset-light deployment to infrastructure ownership, to securing power directly, and ultimately to full vertical integration.
Crucially, it says these positions are not fixed. Companies reposition themselves by moving up the stack to gain control or across it to capture new sources of consumer demand. The article describes a strategy that combines securing a power contract with monetizing it through proprietary Bitcoin mining power while retrofitting infrastructure for higher-margin AI computing colocation.
It adds that Bitcoin miners began solving these problems early, and that AI companies are arriving at similar conclusions. The key difference, it says, is no longer the system itself, but how each company chooses to navigate it.
The article states that the next installment will examine how these models are beginning to converge and what that means for the future of energy, compute, and capital.
Premium gym chains are entering a “golden era” that is ending or already in decline, as rising operating costs collide with shifting consumer preferences toward more flexible, community-based ways to exercise. Long-term memberships are shrinking, margins are pressured by higher rents and facility expenses, and competition from smaller, more personalized…
