
Mining Companies' Transformation: The Most Valuable Asset Isn't Bitcoin, But Electricity
Editor | Wushuo Blockchain
TL;DR
· For a group of publicly traded Bitcoin mining companies pivoting to AI and high-performance computing (HPC), valuation logic is shifting from hashrate, Bitcoin production, and BTC holdings toward energized capacity, contracted IT load, and project delivery capabilities. Land, substations, and grid interconnections previously prepared for mining are becoming scarce assets in competition among AI data centers.
· Core Scientific’s second-quarter 2026 high-density hosting revenue reached $136.7 million, accounting for approximately 83% of total revenue; self-mining revenue fell year-over-year by about 66% to $21.54 million, recording a gross loss of approximately $12.17 million, demonstrating a pivot in its revenue focus from mining to AI infrastructure.
· TeraWulf signed a data center lease with Anthropic covering approximately 401 MW of critical IT load, valued at an estimated $19 billion; Hut 8 also secured a new data center lease adding 352 MW valued at $9.8 billion. However, most of this capacity will not be delivered until between 2027 and 2028.
· “Power is the most valuable asset” does not refer to planned gigawatts, but rather to infrastructure that can be energized on schedule, secure financing, build high-density data centers, and secure long-term leases from reliable customers. Based on data as of June 4, 2026, VanEck estimated that relevant companies had only delivered approximately 25% of leased capacity at that time, facing a near-term funding gap of roughly $50 billion.
In the past, the market evaluated Bitcoin mining companies based on three primary metrics: hashrate size, daily Bitcoin production, and BTC holdings on the balance sheet. Today, for a cohort of publicly listed miners pivoting to AI and HPC, this formula is no longer valid.
The new units of valuation are no longer just EH/s and BTC, but MW and GW. Investors are now asking: how much power capacity does a mining company control that is already energized or has clear interconnection arrangements, how much has been contracted to AI clients, and how much has been fully delivered and is already generating billable revenue.
Beneath this shift lies two vastly different industry realities. Revenue per unit of hashrate in Bitcoin mining continues to face pressure, while demand for power from AI data centers is rapidly growing. In April 2026, the International Energy Agency (IEA) updated its forecasts, projecting global data center electricity consumption to rise from approximately 485 TWh in 2025 to around 950 TWh by 2030, nearly doubling. The IEA had previously estimated that U.S. data centers would account for nearly half of new power demand between 2024 and 2030.
Power thus becomes a scarce asset linking both industries. Land, substations, transmission interconnections, and long-term power purchase agreements once prepared for Bitcoin miners are being repurposed into AI infrastructure. Mining companies are no longer solely producing Bitcoin through computation; they are also monetizing power capacity along with corresponding construction and operational capabilities through long-term data center leases.
The Old Formula for Mining Is Failing
Mining companies are pivoting to AI not primarily because the AI narrative is more compelling, but because profit margins in traditional mining operations are narrowing.
Following the 2024 Bitcoin halving, the block subsidy dropped to 3.125 BTC. Meanwhile, network hashrate and mining difficulty have remained elevated over the long term, continuously diluting the revenue allocated to each individual machine. Data from VanEck shows that as of mid-July 2026, Bitcoin Hashprice had fallen to approximately $30.6/PH/s/day, with average daily industry miner revenue over the past 30 days at roughly $28.5 million. Inefficient machines are now hovering around or below the breakeven point.
For mining companies, this means that even without further significant declines in Bitcoin prices, earnings could drop due to intensified competition. Machines require constant updates, electricity costs must remain sufficiently low, and any downtime, difficulty adjustments, or rising financing expenses could quickly erode profits.
More importantly, Bitcoin mining remains fundamentally a commoditized business. Computational power generated by machines of the same generation shows little differentiation, making it difficult for companies to secure long-term premiums purely based on branding. Revenue fluctuates in real-time alongside coin price, fees, and network difficulty. By contrast, once an AI data center secures a long-term lease with a highly creditworthy client, it gains the opportunity to convert power interconnection, data center construction, and operational capabilities into longer-duration, more predictable cash flows.
This is where mining companies begin reassessing their assets: the most valuable resource may not be the ASIC rigs in the server rooms, nor the unsold Bitcoin, but the power access rights secured behind the mining facility.
Core Scientific’s Financials Have Transformed
Core Scientific serves as the most direct example of this shift.
In Q2 2026, the company reported total revenue of $164.2 million, representing a year-over-year increase of approximately 109%. High-density hosting revenue reached $136.7 million, accounting for roughly 83% of total revenue, compared to just $10.6 million in the same period last year.
Conversely, self-mining revenue declined from $62.42 million in the prior-year period to $21.54 million, down approximately 66% year-over-year. Costs for this segment totaled $33.7 million for the quarter, resulting in a gross loss of roughly $12.17 million and a gross margin of -56%. Meanwhile, high-density hosting generated approximately $79.98 million in gross profit, yielding a 59% gross margin.
The same portfolio of data center assets yields entirely different financial outcomes under these two business models. As of mid-July 2026, Core Scientific had begun billing for 437 MW of customer power capacity, corresponding to an annualized GAAP hosting revenue of approximately $635 million; its contracted customer power capacity stands at roughly 1.1 GW, with potential contract revenue exceeding $24 billion.
This does not signify that the transformation is complete. The company’s Q2 capital expenditures reached $797.5 million, far outstripping concurrent revenue, and a significant portion of contracted capacity remains unbilled. Nevertheless, it demonstrates that a mining company’s financial focus can shift rapidly: declining Bitcoin production no longer necessarily translates to proportional revenue drops, provided that capacity originally used for mining can be converted into billable customer power capacity.
A Single Data Center Lease Can Be Worth Billions
If Core Scientific illustrates changes in revenue structure, then TeraWulf and Hut 8 demonstrate how power capacity and data center construction and operational capabilities are being repriced through long-term contracts.
In July 2026, TeraWulf signed a 20-year data center lease with Anthropic. The agreement covers its Justified Data campus in Hawesville, Kentucky, which plans to deliver approximately 401 MW of critical IT capacity for AI workloads. Delivery is expected to begin in the second half of 2027, with full operations targeted for early 2028. TeraWulf stated that the contract is projected to generate approximately $19 billion in contractual revenue during its initial term, backed by anticipated investment-grade credit support.
Later that month, Hut 8 announced a second 15-year lease worth $9.8 billion at its Beacon Point campus in Texas, adding 352 MW of IT capacity and bringing the total contracted footprint for that single client at the site to 704 MW. Consequently, base-term contract values at Beacon Point have reached $19.6 billion, though the first Phase II data hall is not expected to begin deliveries until Q2 2028.
These deals reflect a new industry reality: from the perspective of AI clients, the scarcest resource may not be GPUs, but large-scale power available to come online within defined timelines. Chips can be procured, servers deployed, but transmission lines, substations, land permits, and grid interconnections often take years to secure.
Bitcoin mining companies happen to have completed portions of this groundwork in advance. In pursuit of cheap power, they have spent the last decade scouting locations close to generation sources, capable of supporting massive loads, and permitting rapid construction. Previously, these assets served to keep tens of thousands of ASIC rigs running continuously; when AI data centers face power bottlenecks, the exact same location selection and energy capabilities unlock new monetization channels.
Why Does the Same Power Capacity Command Higher Premiums Serving AI?
The essence of a mining company’s transition does not involve simply unplugging mining rigs and plugging in GPUs.
Bitcoin mining can tolerate relatively high downtime rates. Operators can voluntarily shut down during price spikes or grid stress, or relocate equipment to other facilities. AI training and inference workloads, however, demand higher power stability, network bandwidth, cooling capacity, and system redundancy. High-density GPU racks impose significantly greater requirements on data center design than traditional mining setups.
The premium here does not stem from markups on reselling kilowatt-hours, but from converting power capacity into highly reliable critical IT load. Therefore, what truly commands the premium is not merely “access to power,” but a combination of four capabilities: energized capacity or clear interconnection arrangements, engineering prowess capable of meeting construction milestones, financial backing able to absorb massive upfront investments, and tenant creditworthiness sufficient to secure long-term leases.
Long-term leases also alter risk structures for mining firms. Mining revenue fluctuates almost daily based on Bitcoin price, network difficulty, and transaction fees; AI hosting revenue depends primarily on contract terms, delivery schedules, and tenant performance. The former resembles commodity production, while the latter increasingly mirrors data center development or infrastructure asset operations.
This explains why the market is increasingly scrutinizing “who is leasing this data center capacity.” A long-term lease backed by an investment-grade hyperscale cloud provider can help mining companies secure project financing at lower costs. The same 100 MW of capacity will command a markedly different valuation if the lessee is smaller, the lease term is shorter, or the credit profile is weaker.
Wall Street Is Now Pricing in MW, Not BTC
In a June 2026 valuation framework for mining companies’ AI infrastructure, VanEck utilizes Gross Energized Power as the primary metric for comparing corporate enterprise value. Citing data as of June 4, 2026, VanEck noted that companies securing AI or HPC leases generally trade at valuation multiples exceeding 10x under this metric; firms with virtually no contracted capacity that still mainly showcase future power reserves trade at roughly 2x to 6x. These multiples are not P/E, EV/Revenue, or EV/EBITDA.
This indicates the market is distinguishing between various capacity stages: planned capacity remains in reserve or conceptual development; locked power capacity has secured power agreements or interconnection arrangements but may not yet be energized; energized capacity possesses actual power delivery conditions; and delivered/billed capacity has been handed over to clients and is generating revenue. The four cannot be treated equivalently.
As more projects come online, valuation standards will continue to evolve. Markets initially reward miners possessing power resources, then reward those securing contracts, before ultimately reverting to more traditional questions: whether projects are delivered on schedule and within budget, how much cash flow each MW generates, and whether capital returns cover financing costs.
Bitcoin miners are consequently diverging. Companies like Core Scientific, TeraWulf, Hut 8, and Cipher are increasingly resembling data center development and operations platforms; MARA and CleanSpark maintain higher exposure to Bitcoin mining. The so-called “mining stocks” sector may soon cease to be a single investable theme valuated under one unified logic.
The Greatest Risk: Treating Power Blueprints as Realized Revenue
The AI transition raises the ceiling for mining company valuations but introduces execution hurdles far surpassing those of traditional mining.
Based on data as of June 4, 2026, VanEck estimated that relevant companies had delivered only about 25% of leased capacity at that time, leaving a near-term funding gap between required capital expenditures and existing cash of approximately $50 billion. This estimate excludes future operating cash flows and proceeds from selling or pledging BTC. VanEck simultaneously estimated that long-term CapEx needs for these companies approach $221 billion, though it did not classify this as a “long-term funding gap.” The massive GW figures and contract totals frequently cited in press releases still correspond to construction plans slated for 2027 or even post-2028.
This means a multi-billion-dollar contract cannot be directly equated to equivalent current revenue. Projects remain vulnerable to grid upgrades, equipment delivery delays, construction costs, financing conditions, regulatory permits, and community pushback. For mining companies lacking high-density data center construction experience, any delay or cost overrun could simultaneously cripple cash flow and depress market valuations.
Customer concentration poses another risk. Long-term leases boost revenue visibility but may allow a single tenant to dictate the fate of an entire campus or even the company itself. If AI infrastructure spending cools, tenants cut CapEx, or next-generation chips alter data center designs, miners could find themselves stuck with heavy assets built for specific clients that are difficult to quickly repurpose.
Furthermore, the transition itself demands massive capital. Mining companies can raise funds through equity issuances, convertible bonds, project loans, and customer deposits, but these methods may also trigger equity dilution, higher leverage, and complex financing covenants. Controlling power resources is merely the entry ticket; the ultimate return depends on the ability to develop those resources into billable assets at reasonable costs.
Mining Companies Are Evolving Into Power Infrastructure Firms
Bitcoin mining has not lost its value. It can still rapidly monetize power and provide high-elasticity returns during bullish market cycles. Unlike traditional data centers, mining farms can actively curtail loads, participate in grid demand response, and generate transitional revenue for uncontracted power capacity.
Yet for a cohort of publicly traded miners, Bitcoin is shifting from a sole core business to merely one monetization avenue for power infrastructure. Rigs can be replaced, mined Bitcoin sold, but what remains genuinely hard to replicate is the acquired land, interconnection qualifications, transmission facilities, and large-scale power arrangements.
Thus, “power is the most valuable asset” requires a crucial caveat: it refers not to GW figures drafted in master plans, nor merely to cheap electricity, but to power that can be interconnected on schedule, secured financing, enable high-density data center construction, and be leased long-term by creditworthy clients.
Mining companies previously scoured for cheap power to produce more Bitcoin; today, they are beginning to monetize scarce interconnection resources and the time value of infrastructure through data center leases. Bitcoin mining is thus entering a new fork: some companies will continue betting on coin prices and hashrate cycles, while others may completely shed the “miner” label to become novel power real estate developers for the AI era.













