CoinShares Bitcoin Mining Report: Hashrate Prices Hit 5-Year Low, 20% of Older Mining Machines Are Now Losing Money

Author | James Butterfill

Compiled by Wu Shuo Blockchain

TL;DR: Key Takeaways from the 2026 Q1 Bitcoin Mining Report

• Profitability under extreme pressure: Q4 of 2025 became the most difficult quarter since the halving. Affected by the price correction of the coin and the high hash rate, the hash rate price once fell below $30/PH/day, hitting a five-year low . About 15-20% of the old mining machines in the entire network fell into losses.

• AI transformation is accelerating: Listed mining companies have cumulatively announced AI/HPC contracts exceeding $70 billion. The capital market is giving AI narratives extremely high premiums (valuation multiples of up to 12.3 times), and the industry is rapidly differentiating into "infrastructure providers" and "pure mining companies".

• Brief pullback in computing power: Affected by factors such as profit squeeze, winter power rationing and regulatory inspections, the total network computing power in Q4 dropped by about 10% from its peak. However, the model predicts that the industry will still be resilient and the total network computing power will rebound and climb to 1.8 ZH/s by the end of 2026.

• Cost and Debt Restructuring: AI construction has led to a surge in the overall cost per BTC for some hybrid mining companies (such as CIFR and WULF) and a huge amount of debt; in contrast, low-leverage mining companies such as CLSK and HIVE have demonstrated strong financial discipline and a pure mining cost advantage.

Key takeaway: The mining industry is undergoing a profound structural restructuring. If the price of Bitcoin fails to rebound to above $100,000 by 2026, high-cost miners will face accelerated exits (miner capitulation), while operators with extremely low energy costs or successful forays into AI will dominate the future capital markets. ( Related reading: Bitcoin mining companies accelerate their exit from the mining era, MARA sells off large amounts of coins to invest in AI )

I. Execution Summary

The fourth quarter of 2025 will be the most challenging quarter for Bitcoin miners since the halving in April 2024. A sharp correction in the price of Bitcoin (from an all-time high of around $124,500 in early October to around $86,000 at the end of December, a pullback of about 31%), coupled with near-historic network hashrate, has compressed the hashrate to its lowest point in five years.

In the fourth quarter of 2025, the weighted average cash cost for publicly traded mining companies to mine a single Bitcoin climbed to approximately $79,995.

This quarter highlighted three core themes:

Profitability under pressure : Hash price has fallen to approximately $36–38/PH/s/day, nearing or at the break-even point for many miners. Three consecutive mining difficulty reductions (the first consecutive reductions since July 2022) signaled "miner capitulation." Entering the first quarter, hash price further plummeted to $29/PH/s/day, meaning miners will face even more growing pains.

The AI/HPC transformation is accelerating : the divide between pure mining companies and infrastructure companies shifting towards AI is widening. Currently, the entire listed mining sector has announced AI/HPC (high-performance computing) contracts worth over $70 billion. WULF, CORZ, CIFR, and HUT are essentially evolving into data center operators that also mine Bitcoin.

Capital restructuring : Several mining companies have taken on massive debt to raise funds for AI infrastructure construction. IREN currently has $3.7 billion in convertible notes; WULF's total debt reaches $5.7 billion; and CIFR has issued $1.7 billion in senior secured notes. The industry's overall leverage has fundamentally altered its risk profile.

II. AI and Bitcoin Mining Compete for Rack Space

AI is continuously vying for rack space in many data centers, which in the long run could drive Bitcoin mining to shift to more intermittent and cheaper sources of electricity.

The migration of Bitcoin miners to AI and high-performance computing (HPC) is accelerating rapidly. According to recent company announcements, by the end of this year, listed mining companies may derive up to 70% of their revenue from AI, compared to approximately 30% currently. What began as a peripheral diversification strategy is increasingly becoming a core business.

Between 2025 and early 2026, Bitcoin mining companies signed numerous GPU co-location and cloud service agreements with hyperscalers, totaling over $70 billion. While most agreements planned for new data centers, cannibalization and closure of existing mining facilities are still highly likely. Therefore, as the capacity stipulated in these contracts gradually ramps up, the proportion of Bitcoin mining in these operators' revenue will significantly decline throughout 2026.

This shift is largely driven by economic considerations. With computing power prices hovering near cyclical lows, mining profit margins are shrinking, while AI infrastructure structurally offers higher and more stable returns. In this context, reallocating electricity and capital to high-performance computing (HPC) makes perfect sense, especially for operators with scalable energy and existing data center capabilities.

Nevertheless, this transformation has not been uniform.

• Some mining companies, such as IREN and Bitfarms, are actively repositioning themselves as HPC providers, essentially using mining as a bridge to AI infrastructure.
• Other mining companies, such as CleanSpark, continue to prioritize mining in the short term, using it to monetize their recently developed capacity while gradually expanding their presence in the AI ​​field.
• The third group remains committed to Bitcoin mining, but their operational methods are evolving. These operators are no longer pursuing hyperscale facilities, but instead focusing on the lowest-cost and often intermittent energy sources, such as stranded renewables or flare gas. For example, Marathon has deployed smaller, approximately 10-megawatt localized containerized sites at the edge of its energy grid. This type of configuration is well-suited for mining operations that can tolerate power outages, but is incompatible with AI workloads that require near-continuous, uninterrupted operation.

Load balancing is likely to remain a persistent segment within the mining industry. By providing demand-side flexibility to providers like the Texas Grid (ERCOT), miners can secure more favorable electricity rates. This role is likely to become increasingly important, although it may become more attractive to smaller, more specialized operators over time.

A key unresolved question is the sustainability of this AI-driven transformation. While the current economic climate is heavily biased towards AI, the mining business remains highly sensitive to Bitcoin prices. If mining profitability experiences a substantial recovery, some operators may reassess their capital allocation between the two businesses. In this sense, the current trend may not be a permanent transformation, but rather a result of relative returns.

In the long run, this could mean a shrinking of the pure-play mining companies, while hybrid infrastructure companies that cross over from mining to AI will become more prevalent. At the same time, new players may emerge to tap into niche markets vacated by established players, particularly in energy-constrained or highly flexible market segments.

The cost difference between Bitcoin mining infrastructure (approximately $700,000 to $1 million per megawatt) and AI infrastructure (approximately $8 million to $15 million per megawatt) is enormous, and this transitional opportunity is currently being realized on a large scale.

CORZ: Approximately 350 MW of high-performance computing (HPC) power is up and running, with approximately 200 MW billed. The contract with CoreWeave has expanded to $10.2 billion over 12 years. The goal is to have all 590 MW operational by early 2027.

• WULF : The Lake Mariner site has 39 MW of critical IT capacity online. Contracted HPC revenue totals $12.8 billion. Other facilities are progressing as planned by Q4 2026. The platform will expand to five locations with a total capacity of approximately 2.9 GW.
• CIFR : Partnering with Fortress Credit Advisors to develop the 300 MW Barber Lake site. Has entered into a multi-billion dollar Fluidstack deal (backed by Google). No revenue has been generated yet.
• IREN : Scale has expanded to over 10,900 NVIDIA GPUs. Childress Horizon Phase 1–4 expansion project (up to 200 MW of liquid-cooled GPUs). Q4 AI cloud service revenue reached $17.3 million.
• HUT : A $7 billion, 15-year, 245-megawatt lease agreement has been signed with Fluidstack at the River Bend campus in Louisiana, with the first data hall scheduled to open in early 2027.

The failure of the CORZ-CoreWeave merger (rejected by shareholders on October 30, 2025) highlights the tension between infrastructure value and equity value. CORZ subsequently restated its financial data due to the improper capitalization of assets committed to be demolished during the HPC conversion, demonstrating the complexity of its accounting practices.

Revenue contributions are still in their early stages but are growing: CORZ's hosted AI/HPC data centers accounted for 39% of its Q4 revenue; WULF's HPC business accounted for 27%; IREN's AI cloud business accounted for 9%; and HIVE's HPC business accounted for 5%. Although mining still dominates, it is clear that AI's revenue contribution will continue to grow across the board.

III. Total Network Computing Power

In late August 2025, the Bitcoin network reached a significant milestone, with its hashrate surpassing 1 ZH/s for the first time. In early October, the network hashrate peaked at approximately 1,160 EH/s.

However, a significant reversal occurred in the fourth quarter. The network hashrate dropped by approximately 10% from its October peak, falling to around 1,045 EH/s by the end of December (before further declining to 850 EH/s in early February before recovering), accompanied by three consecutive mining difficulty reductions—the first consecutive reductions since July 2022. This was primarily driven by the following factors:

The BTC price correction caused older mining rigs from the S19 era to fall below the break-even point (the break-even electricity price for the S19 XP dropped from approximately $0.12/kWh in December 2024 to approximately $0.077/kWh in December 2025).

Rising energy costs in winter and curtailment by ERCOT (Texas Electric Reliability Commission) have led to a sharp increase in unprofitable mining time during November and December.

Regulatory action has been restarted in Xinjiang, China (the crackdown in December 2025 restricted mining operations, although this capacity was not permanently transferred).

Despite a short-term decline, the Bitcoin network still added approximately 300 EH/s of hashrate throughout 2025. As of this writing, the network hashrate has remained largely at the level of the end of 2025, at approximately 1,020 EH/s.

While the recent hashrate pullback may seem worrying, a logarithmic scale reveals it to be far less severe than China's 2021 mining ban. This is more a result of cyclical and weather factors than a sign of a more serious crisis for the industry. The subsequent strong rebound in hashrate also highlights that many miners still consider mining an economically viable business activity.

Based on our previously detailed piecewise prediction model, we now expect the total network hashrate to reach 1.8 Zetahash (ZH/s) by the end of 2026 and 2 Zetahash (ZH/s) by the end of March 2027, a month later than our previous prediction.

Geographic Shift in Computing Power: The top three countries (the United States, China, and Russia) control approximately 68% of global computing power. The United States' market share increased by about 2 percentage points quarter-over-quarter (QoQ). Driven by mining companies such as HIVE (Paraguay's 300 MW project) and BTDR (Ethiopia's 40 MW project), emerging markets like Paraguay, Ethiopia, and Oman have successfully entered the global top ten.

IV. Dynamics of Computing Power Prices

The hash price (a metric that determines a miner's revenue per unit of hash power) peaked at approximately $63/PH/s/day in July 2025 and then continued to decline throughout the fourth quarter. By November, it had fallen to approximately $35–37/PH/s/day, a five-year low at the time. A brief rebound to approximately $38–40 occurred in late December and early January, but this was short-lived, and the hash price collapsed further in the first quarter of 2026, falling to approximately $28–30/PH/s/day by early March, a post-halving low.

This decline was caused by a combination of factors: record mining difficulty (reaching a peak of 155.97T after a 6.31% increase on October 29), depressed Bitcoin prices (down about 31% from the all-time high in October), and extremely low transaction fee revenue (consistently below 1% of the total block reward, with an average fee of about 0.018 BTC per block).

This has created the most challenging profit environment since the halving in April 2024. With an average industrial electricity price of $0.05/kWh (or $0.077/kWh for the S19 XP), miners operating mid-generation mining rigs (such as the S19j Pro-class rig with an efficiency ratio of approximately 29.5 J/TH) were already operating well below break-even by the end of the year, and the situation is expected to worsen further into 2026.

Latest forecast: The deterioration of the hashrate pricing environment has exceeded our previous expectations. It briefly touched approximately $28/PH/s/day in late February, before recovering to around $30–35 at the time of writing. At current levels, miners operating mid-generation mining rigs need electricity prices below $0.05/kWh to remain cash-profitable, while the latest generation models (efficiency ratio below 15 J/TH) can still maintain considerable profit margins under typical industrial electricity prices. For hashrate prices to continue recovering to above $40, the price of Bitcoin needs to rebound to $100,000 by the end of the year, and its price increase must outpace the continued growth of the entire network hashrate.

Unless there is a substantial rebound in Bitcoin prices, we expect high-cost operators to face further "miner capitulation" in the first half of 2026. The current mining economics are insufficient to stimulate a large-scale hardware upgrade cycle. Hashrate prices must first fall further, forcing enough outdated capacity and operators to shut down, thereby reducing the overall network hashrate and mining difficulty. This would provide an entry point for new Bitcoin miners or sufficient incentive for upgrading existing nodes. However, despite the relentless squeeze on profit margins, the network hashrate has shown remarkable resilience. This is likely supported by a combination of factors: including state-backed mining activities driven by strategic rather than purely economic benefits; operators with access to extremely cheap or stranded electricity; and ASIC manufacturers integrating unsold inventory into their own facilities to maintain order commitments with foundries like TSMC and Samsung.

The growing pains in the mining industry have triggered a massive sell-off and capitulation by miners. Listed mining companies have collectively reduced their BTC holdings by more than 15,000 BTC from their peak. Core Scientific alone sold approximately 1,900 BTC (about $175 million) in January and plans to liquidate almost all of its remaining holdings in the first quarter of 2026; Bitdeer wiped out its BTC holdings in February; and Riot sold 1,818 BTC (about $162 million) in December 2025.

We believe it's not unrealistic to assume the Bitcoin price will rebound to the $100,000 mark; if that price is reached, the hashrate price would rise to $37/PH/s/day. If the price remains below $80,000 for the remainder of the year, and assuming mining difficulty continues to rise, we predict the hashrate price will continue to decline. However, in this scenario, the actual trajectory could be different: as miners shut down unprofitable rigs, the network hashrate could decrease further, making the hashrate price more likely to stabilize. If we see the price begin to test the all-time high of $126,000, the hashrate price could surge to $59/PH/s/day.

The drop in computing power prices has far exceeded our predictions, although we believe this is a temporary phenomenon caused by the recent decline in cryptocurrency prices and expect it to gradually stabilize in the range of $30 to $40 per PH per day.

Current hashrate prices have made it unprofitable to continue operating multiple mining rigs. At the current hashrate price level of $30/PH/day, any mining rig with performance lower than the S19 XP and facing electricity prices of $0.06/kWh (6 cents/kWh) or higher is operating at a loss — we estimate that this segment of equipment accounts for approximately 15% to 20% of the global active mining rig fleet.

V. Mining Cost Analysis

1. Overview

The table below shows the cost per BTC for all mining companies included in the study in the fourth quarter of 2025. All data are denominated in USD for the cost of mining a single BTC and are allocated to self-mining operations using the revenue-share methodology described in the appendix.

Key observations:

The construction of AI/HPC is distorting the headline cost-per-BTC metric for hybrid operators. Debt, selling and administrative expenses (SG&A), and depreciation and amortization (D&A) resulting from AI infrastructure construction are being spread across a shrinking BTC production base, thus inflating the apparent headline cost-per-BTC figure. For companies like WULF, CORZ, and CIFR, their all-in cost increasingly reflects the economics of transitioning into data center operators, rather than simply the economics of Bitcoin mining.

Compared to the second quarter of 2025, the industry-wide electricity costs have seen a substantial and significant increase. This reflects the increased mining difficulty across the network, which diluted the yield per BTC, rising energy costs during the winter, and the decline in BTC price.

Depreciation and amortization (D&A) is the largest component of non-cash costs and varies significantly depending on the company's depreciation policy. MARA's $136,000/BTC and CIFR's $88,000/BTC are outliers (because MARA has a large mining fleet; while CIFR uses a 3-year depreciation assumption).

Stock-based compensation (SBC, equity incentive expense) remains a significant differentiator. HUT's $48,500/BTC (primarily one-time awards to the CEO/CSO) and CORZ's $35,500/BTC are outliers. BTDR ($3,900/BTC) and CLSK ($6,700/BTC) demonstrate the most stringent financial discipline.

Interest costs are currently having a significant impact on several mining companies. WULF ($145,000/BTC), CIFR ($56,000/BTC), and BTDR ($16,000/BTC) are burdened with massive debt. In contrast, HIVE ($320/BTC) and CLSK ($830/BTC) have extremely low leverage and a significant structural advantage.

2. Details of each company

MARA (MARA Holdings)

BTC produced: 2,011

Total cost: $153,040/BTC

Cash cost (before tax): $103,605/BTC

In the fourth quarter, MARA produced 2,011 BTC, remaining the largest publicly traded mining company by output. As of the end of December, the company's energized hashrate reached 53.2 EH/s (a 15% increase this quarter), but due to increased network difficulty, daily output dropped to approximately 21.9 BTC, lower than in previous quarters.

Its electricity cost of $64,703/BTC is in the middle range among its peers, reflecting its diverse geographical distribution and heavy reliance on third-party hosting (of the total electricity cost of $130.1 million, third-party hosting accounted for $79.4 million). Its depreciation and amortization (D&A) is the highest among its peers at $136,166/BTC, reflecting its massive mining rig scale (full fiscal year D&A was $772.8 million).

The apparent comprehensive cost was significantly distorted by a $183.4 million income tax gain, stemming from a fair value adjustment to BTC holdings in accordance with ASU 2023–08 accounting standards. Excluding this non-operating gain, the comprehensive cost soared to $240,407. In the fourth quarter, MARA maintained its “Hold-on-Demand (HODL)” strategy, not selling any BTC and keeping 7,377 BTC in third-party lending arrangements. However, the company had already begun softening this stance in the third quarter of 2025, allowing the sale of newly mined BTC to fund operations. In its 10-K filing dated March 2, 2026, MARA further expanded this policy, authorizing the sale of all 53,822 BTC in its balance sheet reserves. This shift was partly due to pressure on its $350 million Bitcoin-collateralized credit facility—the loan-to-value (LTV) ratio of which climbed to approximately 87% as BTC fell toward $68,000 in early 2026. This marks a substantial departure from the full HODL strategy adopted in July 2024.

In addition, the company announced a partnership with Starwood Capital in AI and HPC data centers, and acquired a 64% stake in Exaion for $174.5 million in February 2026, indicating that it is accelerating its diversification beyond pure mining.

IREN (IREN Limited)

BTC produced: 1,664

Total cost: $140,441/BTC

Cash cost: $58,462/BTC

Thanks to a favorable electricity agreement at its Childress, Texas facility and $1.8 million in demand response revenue in the fourth quarter, IREN achieved the lowest electricity cost per BTC at just $34,325. Its installed hashrate reached 46 EH/s, with a cluster efficiency ratio of approximately 15 W/T.

Its share-based compensation expense (SBC) was $31,717/BTC, ranking second among its peers (Q4 SBC was $58.2 million, a 7.3-fold increase year-over-year, primarily driven by options with a $75 strike price and a large vesting of restricted stock units (RSUs)). Payroll taxes associated with the SBC increased actual cash costs by $6.8 million. Depreciation and amortization (D&A) nearly tripled year-over-year to $99.2 million, reflecting the expansion of the Childress project.

IREN carries a total of $3.7 billion in convertible notes across five series (2029–2033), making it the heaviest indebted among its peers by face value. However, its interest expenses are manageable due to low coupon rates (2.75%–3.50%). $111.8 million in debt conversion inducement fees (non-cash) and $182.5 million in deferred income tax benefits are excluded from the cost analysis. Its AI cloud service revenue reached $17.3 million (9% of total revenue), while the Horizon Phase 1–4 GPU expansion project (up to 200 megawatts) is under construction.

CLSK (CleanSpark)

BTC produced: 1,821

Total cost: $118,932/BTC

Cash cost (before tax): $71,188/BTC

CleanSpark demonstrates exceptional operational discipline. Its sales and administrative expenses (SG&A) are $17,848/BTC, and its equity incentive expenses (SBC) are $6,662/BTC, among the lowest in its industry. A 100% allocation ratio (pure mining, no hosting/HPC revenue) simplifies cost analysis.

Electricity costs were $52,463/BTC, up from $44,679 in the second quarter, reflecting increased mining difficulty. With an installed capacity of approximately 50 EH/s, the cluster efficiency of approximately 16 W/T remains industry-leading. Depreciation and amortization (D&A) were $58,381/BTC, roughly in line with peers. Interest expense was extremely low ($830/BTC), reflecting its low-leverage balance sheet.

New CEO Matt Schultz (who will succeed Zach Bradford in August 2025) stated that computing power could potentially climb to approximately 60 EH/s if market conditions allow. The company is exploring equipment vendor diversification to reduce its reliance on Bitmain. No concrete AI/HPC plans have been announced yet, although management has hinted at the possibility of monetizing data center assets located near metropolitan areas (Georgia facilities). Note: CLSK's fiscal year ends September 30th, meaning the current data pertains to the first quarter of its fiscal year 2026.

RIOT (Riot Platforms)

BTC produced: 1,324

Total cost: $170,366/BTC

Cash cost (before tax): $102,538/BTC

Riot produced 1,324 BTC with an average deployment hashrate of 31.5 EH/s. The $9.9 million ERCOT demand response credit in the fourth quarter (totaling $56.7 million for the full fiscal year) significantly benefited its electricity costs of $49,196/BTC, effectively offsetting its total electricity costs.

Selling and administrative expenses (SG&A) reached $31,534/BTC, among the highest in the industry, reflecting corporate administrative costs and development expenditures for the 1 GW Corsicana project. Share-based compensation expenses (SBC) were also high at $21,586/BTC. Depreciation and amortization (D&A) were $66,900/BTC, reflecting ongoing investment in mining equipment. As of December 31, the company held 17,722 BTC (worth over $1.5 billion at closing prices).

Riot's strategic focus is on Project Corsicana, with 600 megawatts designated for AI workloads. While this represents a significant long-term opportunity, Q4 revenue was still largely driven by mining operations. The total site capacity of 1 gigawatt makes Riot one of the largest single-site facility operators in North America.

CORZ (Core Scientific)

BTC produced: 421

Total cost: $168,693/BTC

Cash cost: $110,282/BTC

The fourth quarter marked a milestone for CORZ's transition to AI/HPC. Hosting revenue reached $31.3 million (39% of total revenue, up from $8.5 million in the fourth quarter of 2024). Proprietary mining revenue decreased year-over-year from $79.9 million to $42.2 million, as capacity was deliberately shifted to the HPC sector.

Lower BTC production (421 BTC) drove up metrics per BTC. Sales and administrative expenses (SG&A) of $47,510/BTC and share-based compensation expenses (SBC) of $35,506/BTC were among the highest in the industry, reflecting corporate administrative costs and the costs of the failed CoreWeave merger. A fleet efficiency ratio of approximately 24.7 W/T lagged behind peers (15–18 W/T), resulting in electricity costs as high as $66,720/BTC.

While the failed CoreWeave merger (October 30, 2025) has introduced uncertainty, execution continues: approximately 350 MW are energized, and approximately 200 MW are billed, with a target of 590 MW fully operational by early 2027 (contract value of $10.2 billion over 12 years). Due to improper capitalization of assets committed to be dismantled during the HPC conversion, the company made a significant restatement of its financial data for 2024–2025, resulting in a change of auditor to KPMG and the assessment that internal controls were ineffective. Depreciation and amortization (D&A) was $17,701/BTC, the lowest among its peers, reflecting in part the impairment of assets following the restatement.

WULF (TeraWulf)

BTC produced: 262

Total cost: $471,841/BTC

Cash cost: $384,517/BTC

Important note: WULF's cost per BTC is not comparable to that of pure mining peers.

The company has fundamentally transformed into an AI/HPC infrastructure enterprise, maintaining only a shrinking mining operation. The 262 BTC mined this quarter were generated along with $9.7 million in HPC rental revenue.

Fourth-quarter mining revenue declined 40% quarter-over-quarter (QoQ) to $26.1 million. HPC rental revenue increased 35% quarter-over-quarter to $9.7 million (representing 27% of total fourth-quarter revenue). Total revenue for fiscal year 2025 is $168.5 million, with the HPC business contributing $16.9 million.

The extremely high overall costs reflect the following factors: interest at $144,974/BTC (total debt of $5.7 billion: including $2.5 billion in convertible notes and $3.2 billion in senior secured notes from WULF Compute); selling and administrative expenses (SG&A) of $167,221/BTC (primarily due to workforce expansion and milestone compensation); and depreciation and amortization (D&A) of $77,217/BTC (for new HPC infrastructure). By the end of 2025, the company's cash reserves reached $3.7 billion (previously $274 million), reflecting substantial capital formation. It currently has 522 MW of contracted capacity, involving $12.8 billion in long-term customer contracts.

CIFR (Cipher Digital)

BTC produced: 591

Total cost: $231,980/BTC

Cash cost: $103,516/BTC

CIFR has the second highest overall cost (excluding WULF), primarily driven by depreciation and amortization (D&A) of up to $87,768/BTC (based on a 3-year useful life assumption adopted in 2024) and interest expense of $56,445/BTC.

A surge in interest rates was a defining characteristic of CIFR's fourth quarter: its issuance of $1.733 billion in 7.125% senior secured notes in November 2025 led to a spike in interest expenses to $33.4 million in the fourth quarter, compared to just $3.2 million in total interest for the first nine months. Electricity costs were highly competitive at $41,047/BTC (the PPC at the Odessa site is approximately $0.028/kWh). Share-based compensation expenses (SBC) were high at $40,695/BTC and were categorized under "Compensation and Benefits" rather than "Selling and Administrative Expenses (SG&A)" (an unusual reporting method).

Significant asset impairments incurred in the fourth quarter (US$45.3 million impairment of Odessa mining rigs, US$96.1 million impairment of Black Pearl, and US$29.4 million loss on disposal) were excluded from the cost analysis. The company changed its name to Cipher Digital Inc. on February 20, 2026. In high-performance computing (HPC), the 300 MW Barber Lake site (in partnership with Fortress) and the Fluidstack protocol (backed by Google) have laid the foundation for CIFR's diversification, although they have not yet generated revenue.

HUT (Hut 8 Corp.)

BTC produced: 719

Total cost: $160,402/BTC

Cash cost: $50,332/BTC

The overall cost of the Hut 8 surface appears competitive, but it needs to be interpreted carefully due to the presence of several one-off items.

Its share-based compensation expense (SBC) was the highest in the industry at $48,527/BTC, primarily driven by stock awards (2.3 million restricted stock units (RSUs) and performance-based stock units (PSUs)) to be awarded to the CEO and CSO in November 2025. Fourth-quarter SBC was $39.7 million, compared to only $18.1 million for the first nine months – a ratio of 2.2x. Normalizing the SBC would significantly reduce its overall costs.

Due to the $17.8 million Canadian Harmonized Sales Tax (HST) refund received in December 2025, its general and administrative expenses (G&A, excluding SBC) of $7,413/BTC appear artificially low. A normalized G&A would be closer to approximately $30,000/BTC. The depreciation and amortization (D&A) of $48,621/BTC is a consolidated financial statement figure; mining-related D&A is actually lower (due to approximately 74% of plant and equipment (PP&E) being mining-related). The interest expense of $6,840/BTC reflects its total debt of approximately $411 million (including 15.25% TZRC debt, 9% Coinbase debt, and 8% Coatue convertible bonds).

Thanks to Bitmain's mining rigs at the Vega site (with a hashrate of 14.86 EH/s), its BTC production climbed to 719 from 578 in the third quarter. The company currently holds 15,679 BTC (worth approximately $1.37 billion). Its complex business structure (including four business units, the ABTC subsidiary, and inter-company offsets) makes clear cost attribution extremely challenging. The $78.2 million in income tax gains (reversal of deferred income tax) in the fourth quarter has been excluded from the calculation.

BTDR (Bitdeer Technologies Group)

BTC produced: 1,673

Total cost: $118,188/BTC

Cash cost: $87,144/BTC

Bitdeer's overall cost is highly competitive among its peers, although this reflects, to some extent, International Financial Reporting Standards (IFRS) practices and multi-segment revenue (SEALMINER mining machine sales revenue of $23.4 million and HPC/AI revenue of $2.3 million). Average electricity costs rose to $46/MWh from $43/MWh in the third quarter.

The most notable issue was the depreciation policy change in the fourth quarter: management shortened the lifespan of mining rigs, causing depreciation and amortization (D&A) in cost of revenue (CoR) for proprietary mining to more than double sequentially (from $31.2 million to $63.9 million), despite a roughly 60% increase in hashrate. Proprietary mining gross margin plummeted from 27.7% in the third quarter to 3.6%. This was purely an accounting consequence, not a deterioration in operating conditions.

Under IFRS reporting practices, D&A and SBC are bundled into the cost of revenue (CoR), complicating comparisons with peers using US GAAP. The $16,306/BTC interest expense reflects approximately $1 billion in convertible notes and related-party borrowings. BTDR's proprietary ASIC chip strategy (the SEALMINER A2 with an energy efficiency ratio of 16.5 W/T, and the soon-to-be-mass-produced A3 with an energy efficiency ratio of 9.7 W/T) is a significant competitive advantage, substantially reducing capital expenditure per TH of hashrate (capex/TH) compared to purchasing mining rigs from Bitmain.

HIVE (HIVE Digital Technologies)

BTC produced: 884

Total cost: $144,321/BTC

Cash cost: $75,274/BTC

HIVE mined 884 BTC in the fourth quarter (its third fiscal quarter ending December 31), with production showing a significant increase driven by expansion in Paraguay. The fleet efficiency ratio improved from 21 W/T to 18.5 W/T.

HIVE's electricity cost of $65,368/BTC is the highest among its peers (excluding WULF), driven by a forward-looking accounting change: HIVE capitalized $41.3 million in non-refundable Paraguayan Value Added Tax (VAT) in Plant and Equipment (PP&E) and expensed $5.5 million in electricity VAT through operating expenses (opex). This accounting treatment, compared to its peers, simultaneously increased its D&A and electricity costs.

Its selling and administrative expenses (SG&A) of $9,054/BTC are among the lowest. Share-based compensation expenses (SBC) of $7,501/BTC are moderate (corresponding to RSUs issued at CAD 7.30 in October 2025). Interest expense of only $320/BTC is the lowest among its peers—HIIVE's total debt of only $13.8 million provides a significant structural advantage. During the quarter, the 100 MW Valenzuela facility became operational; HIVE now has a total of 300 MW of ANDE power purchase agreements in Paraguay.

The company faces contingent VAT liabilities of approximately $79.2 million (stemming from an assessment of its Bikupa subsidiary by the Swedish tax authorities, currently under appeal). Its use of 2,079 BTC with repurchase options to pay for equipment deposits is an unusual capital management tactic.

BITF (Bitfarms)

An update will be provided after Bitfarms releases its Q4 earnings report.

VI. Performance and Valuation of Mining Companies' Stocks

In the fourth quarter, the valuation premium for AI/HPC continued to widen. Currently, mining companies that have secured HPC contracts have an enterprise value to future twelve-month sales ratio (EV/NTM sales) of 12.3x, while pure mining companies have only 5.9x. The decline in BTC price in the fourth quarter (a 31% pullback from its all-time high) created a double headwind: it not only reduced mining revenue but also significantly diminished the value of BTC holdings in mining companies' coffers.

The valuation discount experienced by CORZ after the failed merger (likely due to hedge fund liquidation) contrasts sharply with the valuation premium enjoyed by WULF, CIFR, and HUT. Currently, short interest across the sector is high; as of this writing, MARA's short interest represents approximately 30% of its outstanding shares.

The industry has fundamentally diverged into "infrastructure companies" (such as WULF, CORZ, CIFR, and HUT) and "mining companies" (such as MARA, CLSK, RIOT, and HIVE). Whether these AI-driven high valuation multiples are reasonable ultimately depends on the companies' execution capabilities: not all announced agreements can be translated into actual operational infrastructure, and the capital requirements behind them remain extremely large.

VII. First Quarter of 2026 and Future Outlook

1. The recovery of hashrate prices depends on the price of BTC: With the price of BTC around $70,000 and the hashrate price around $30/PH/day, many mid-generation mining rigs are already at or below the break-even point. If the price continues to fall below $70,000, it could trigger a larger-scale "miner capitulation," but this, in turn, would benefit the survivors by reducing mining difficulty and the overall network hashrate.

2. Deployment of next-generation hardware: Bitmain's S23 series and SEALMINER A3 (both with energy efficiency ratios below 10 J/TH) are expected to be deployed on a large scale in the first half of 2026, which will further widen the energy efficiency gap and accelerate the replacement cycle of mining rigs.

3. Inflection Point for AI/HPC Revenue: CORZ aims for full delivery of its 590 MW CoreWeave project by early 2027. WULF's Lake Mariner site expansion continues. The market will closely watch whether contracted revenue translates into actual billing and whether profit margins can reach the target of over 85%.

4. The divergence in leverage ratios is creating a catalyst for mergers and acquisitions: Mining companies with healthy balance sheets and good liquidity (such as HIVE and CLSK) are likely to become acquirers. However, even CLSK has already taken on a considerable amount of convertible debt ($1.15 billion, zero interest rate) to fund its transformation into AI infrastructure.

5. Shifting Geographic Distribution and Regulatory Environment: The US continues to expand its market share. Paraguay and Ethiopia are emerging as mining hubs. Enforcement actions in Xinjiang, China, may prompt a shift of computing power overseas. Texas SB 6 (signed in June 2025) introduces new requirements for the power loads of large-scale mining and data center connections to ERCOT, including mandatory remote power-off capabilities.

6. Industry Consolidation: We expect more M&A activity in 2026. The energy efficiency gap between top-tier fleets (energy efficiency ratio of approximately 15 W/T) and lagging fleets (energy efficiency ratio of approximately 25 W/T or higher) is already large enough that directly acquiring high-efficiency capacity may be less costly than upgrading aging operating facilities.

Appendix: Methodology

Denominator: The amount of BTC mined by our own mining operations this quarter.

Allocation Ratio: Self-operated mining revenue / Total revenue. This ratio applies to selling and administrative expenses (SG&A), depreciation and amortization (D&A), share-based compensation expenses (SBC), interest, and taxes.

The total cost per BTC = electricity cost (after deducting power curtailment compensation) + SG&A (excluding SBC) + D&A + net interest + income tax + SBC — where applicable, all projects are allocated proportionally to mining revenue.

Cash Cost per BTC = Cost of Revenue (excluding D&A) + SG&A (excluding SBC) + Net Interest + Income Tax — all items are allocated proportionally.

Electricity costs: Excluding power curtailment/demand response credit lines. This cost analysis excludes asset impairment, fair value revaluation, and non-operating items (e.g., BTC revaluation gains/losses, fair value changes of derivatives, debt conversion inducement costs).

Units of measurement: Unless otherwise stated, all figures are in thousands of US dollars. Financial data not denominated in US dollars have been converted to the quarterly average exchange rate.