By contrast, self-mining revenue fell approximately 66% year over year, from $62.42 million to $21.54 million. The business incurred $33.7 million in quarterly costs, resulting in a gross loss of approximately $12.17 million and a gross margin of -56%. The high-density colocation business, meanwhile, generated approximately $79.98 million in gross profit, with a gross margin of 59%.
The same portfolio of data center assets produced completely different financial results under the two business models. By mid-July 2026, Core Scientific had 437 MW of customer power capacity generating billable revenue, corresponding to approximately $635 million in annualized GAAP colocation revenue. It had contracted approximately 1.1 GW of customer power capacity, representing more than $24 billion in potential contract revenue.
This does not mean the transformation is complete. The company recorded $797.5 million in capital expenditures in the second quarter, far exceeding its revenue for the period, and a significant portion of its contracted capacity had not yet begun generating billable revenue. But the results at least demonstrate that a mining company’s financial center of gravity can shift within a very short period. A decline in Bitcoin production no longer necessarily means revenue must fall with it, provided that capacity previously used for mining can be converted into billable customer power capacity.
A Single Data Center Lease Can Be Worth Tens of Billions of Dollars
If Core Scientific illustrates the shift in revenue mix, TeraWulf and Hut 8 demonstrate how power capacity and data center development, construction, and operating capabilities can be repriced through long-term contracts.
In July 2026, TeraWulf signed a 20-year data center lease with Anthropic. The agreement covers the Justified Data campus in Hawesville, Kentucky, which is planned to provide approximately 401 MW of critical IT load for AI workloads. Delivery is expected to begin in the second half of 2027, with the full capacity scheduled to come online in early 2028. TeraWulf said the contract was expected to generate approximately $19 billion in revenue over its initial term and receive investment-grade credit support.
That same month, Hut 8 announced a second 15-year, $9.8 billion lease at its Beacon Point campus in Texas, adding 352 MW of IT capacity and bringing the same customer’s total contracted capacity at the campus to 704 MW. The base-term contract value at Beacon Point consequently reached $19.6 billion, even though the first data hall in the second phase is not expected to begin delivery until the second quarter of 2028.
These transactions reveal a new industry reality: for AI customers, the hardest resource to secure may not be GPUs, but large-scale power capacity that can come online within a defined timeframe. Chips can be purchased and servers can be deployed, but transmission lines, substations, land permits, and grid interconnections often take years to develop.
Bitcoin miners happen to have completed part of that work in advance. In pursuit of low-cost power, they have spent the past decade searching for sites near generation sources that can support large loads and allow rapid construction. Those assets were originally used to keep tens of thousands of ASIC miners running. Now that AI data centers face power bottlenecks, the same siting and energy capabilities have gained a new path to monetization.
Why the Same Power Capacity Commands a Higher Premium When Used for AI
The essence of the miners’ transformation is not simply unplugging mining machines and replacing them with GPUs.
Bitcoin mining can tolerate relatively high interruption rates. Miners can shut down voluntarily when electricity prices rise or the grid is under strain, and they can relocate equipment to another site. AI training and inference workloads, by contrast, require greater power stability, network bandwidth, cooling capacity, and system redundancy. High-density GPU racks also impose far more demanding data center design requirements than traditional mining facilities.
The premium does not come from reselling the same electricity at a higher price. It comes from converting power capacity into highly reliable critical IT load. What commands the premium is therefore not merely the possession of power, but a combination of four capabilities: energized capacity or capacity with definitive grid interconnection arrangements, engineering capabilities sufficient to complete construction on schedule, financing capabilities that can support enormous upfront investment, and customer credit strong enough to underpin long-term leases.
Long-term leases also change the risk profile of a mining company. Mining revenue depends on Bitcoin prices, network difficulty, and transaction fees, all of which can change daily. AI colocation revenue depends more heavily on contract duration, delivery schedules, and tenant performance. The former resembles commodity production, while the latter increasingly resembles the business of a data center developer or infrastructure asset operator.
This also explains why the market has begun to focus on who is leasing the data center capacity. A long-term lease backed by an investment-grade hyperscaler can help a miner obtain project financing at a lower cost. The same 100 MW of capacity may receive a very different valuation if the tenant is smaller, the lease term is shorter, or the credit quality is weaker.
Wall Street Is Beginning to Price Miners by MW, Not BTC
In an AI infrastructure valuation framework for miners published in June 2026, VanEck used EV/Gross Energized Power as its primary metric for comparing mining company valuations. Based on data as of June 4, 2026, it found that companies with signed AI or HPC leases generally traded at multiples above 10x on this basis, while companies with little to no contracted capacity that were still relying primarily on their prospective power pipelines traded at approximately 2x to 6x. These multiples were not price-to-earnings, EV/Revenue, or EV/EBITDA multiples.
This means the market is distinguishing between capacity at different stages. Planned capacity remains part of a project pipeline or development concept. Secured power capacity has obtained a power agreement or grid interconnection arrangement, but may not yet be energized. Energized capacity is physically capable of receiving power. Delivered and billable capacity has been handed over to the customer and has begun generating revenue. The four categories cannot be valued as equivalent.
As more projects enter operation, valuation standards will continue to evolve. The market first rewards miners that control power resources, then those that sign contracts, and ultimately returns to more conventional questions: Were projects delivered on time and on budget? How much cash flow does each MW generate? Can returns on capital cover financing costs?
This shift is also driving a divergence among Bitcoin miners. Companies such as Core Scientific, TeraWulf, Hut 8, and Cipher increasingly resemble data center development and operating platforms, while MARA and CleanSpark continue to maintain greater exposure to Bitcoin mining. In the future, so-called “mining stocks” may no longer constitute a sector that can be valued using a single framework.
The Greatest Risk Is Treating a Power Blueprint as Realized Revenue
The AI pivot has raised the valuation ceiling for mining companies, but it has also created execution hurdles far higher than those in traditional mining.
Based on data as of June 4, 2026, VanEck estimated that the relevant companies had delivered only approximately 25% of their leased capacity at the time, with a funding gap of approximately $50 billion between near-term capital expenditure requirements and cash on hand. The estimate did not include future operating cash flow or funds that could be raised by selling or pledging BTC. VanEck also estimated that the companies faced nearly $221 billion in long-term capital expenditure requirements, but it did not define this amount as a “long-term funding gap.” Much of the GW capacity and aggregate contract value announced in press releases still corresponds to construction scheduled for 2027, 2028, or later.
This means a multibillion-dollar contract cannot be treated as an equivalent amount of current revenue. Projects may still be affected by grid upgrades, equipment deliveries, construction costs, financing conditions, regulatory approvals, and community opposition. For miners without experience building high-density data centers, any delay or cost overrun could damage both cash flow and market valuation.
Customer concentration is another risk. Long-term leases can improve revenue visibility, but they can also allow a single customer to determine the future of an entire campus or even an entire company. If investment in AI infrastructure slows, tenants reduce capital expenditures, or a new generation of chips changes data center design, miners may find that capital-intensive assets built for a specific customer cannot be repurposed quickly.
The transformation itself also requires enormous amounts of capital. Miners can raise funds through equity issuance, convertible bonds, project loans, and customer prepayments, but these methods may also result in shareholder dilution, higher leverage, and complex financing constraints. Controlling power resources is only the entry ticket. The ultimate return depends on whether those resources can be converted into billable assets at a reasonable cost.
Mining Companies Are Becoming Power Infrastructure Companies
Bitcoin mining has not lost its value. It can still monetize electricity quickly and provide highly leveraged returns when Bitcoin prices rise. Unlike traditional data centers, mining facilities can also curtail load voluntarily, participate in grid demand-response programs, and generate transitional revenue from power capacity that has not yet been contracted to another customer.
For a group of publicly listed miners, however, Bitcoin is shifting from the sole core business to one of several ways to monetize power infrastructure. Mining machines can be replaced and mined Bitcoin can be sold. What is truly difficult to replicate is secured land, grid interconnection rights, transmission infrastructure, and large-scale power arrangements.
The claim that “power is the most valuable asset” therefore requires an important qualification. It does not refer to GW capacity that exists only in a development plan, or simply to cheap electricity. It refers to power that can be connected to the grid on schedule, financed, built into a high-density data center, and leased over the long term to creditworthy customers.
Miners once sought cheap electricity to produce more Bitcoin. Now, through data center leases, they are beginning to monetize scarce grid interconnection resources and the time value embedded in infrastructure. Bitcoin mining has therefore reached a new fork in the road: some companies will continue to bet on Bitcoin prices and hash-rate cycles, while others may shed the “miner” label altogether and become a new class of power-backed real estate developers for the AI era.
