Record Low Hash Price Changes Mining Economics for 2026 – BitPlanet

As the pressure on mining profitability continues, the utilization criteria for power assets are also being divided between mining and AI.

[Block Media BitPlanet] [Block Media=BitPlanet] The economics of Bitcoin mining are in a phase of structural change. We examined the changes in profitability indicators associated with Bitcoin mining.

EXECUTIVE SUMMARY

Key Conclusion: Structural Profitability Pressure Persists Despite August Rebound

Main Points The hash price is a key indicator of Bitcoin mining profitability. All three record low monthly averages were set in 2026: $30.37 in June, $31.21 in July, and $31.27 in March. The BIP-110 mandatory signaling that started on August 8 saw miner support at just 2.53%, finishing the first cycle at 0.00% and failing to activate. On August 22, during the difficulty adjustment, the difficulty dropped again by 1.31%.

Core Meaning The network hash rate (7-day moving average) slightly decreased after surpassing 1 ZH/s in early June but remains at a high level of around 900 EH/s. In contrast, the difficulty adjustments in 2026 have seen more decreases (10 times) than increases (7 times). Amid high network competition, the shutdown and restart of less profitable mining machines are recurring. The responses from publicly listed mining companies are also changing. MARA and CleanSpark have sold most of the Bitcoin they mined, and the cumulative announcement of AI and HPC contracts by listed mining companies has exceeded $70 billion according to CoinShares' Q1 tally. However, there are also cases where mining competitiveness is maintained based on low-cost electricity.

Key Monitoring Indicators Moving forward, we will focus on the direction and magnitude of the next difficulty adjustment (scheduled for September 5, estimated +0.67%), the difference between forward hash prices and spot prices, and the proportion of HPC revenue from listed mining companies in Q3. As of August 31, the average forward hash price for the next six months is $37.59. If the monthly average hash price stays above the lower bound of $37.89 for more than two months and difficulty increases, we will reconsider the judgment of structural profitability pressure.

01 Three Record Low Hash Prices Concentrated in 2026

Hash price refers to the expected mining revenue earned by a computing power of 1 PH/s over a day and is used as a representative profitability indicator in the mining industry. All three record low monthly averages were set in 2026: $30.37 in June, $31.21 in July, and $31.27 in March. The average monthly hash price in 2025 ranged from $37.89 to $59.38, with an average of $50.68. The lowest value in 2026 is about 20% lower than last year's lowest value (self-calculated).

The Hashrate Index analyzed that if the daily revenue per PH/s is around $32 in August, many mining companies may be around or below their break-even point depending on operating costs and equipment types. However, this is an analysis of the industry as a whole, and the actual cost structure varies by company.

MARA indicated a daily cost of $27.7 per PH/s in its Q2 shareholder letter. However, it is not clearly stated what items are included in this cost. According to the same document, the electricity purchase cost on its site was $38,690 per Bitcoin, which was about 54% of the average Bitcoin price of $71,325 during the same quarter (self-calculated). The net loss for the same quarter was $613 million, but this figure includes non-cash items such as a $343 million impairment loss on held Bitcoin, making direct comparison with operational profit and loss difficult. Therefore, it is hard to definitively determine MARA's break-even point based solely on these figures.

02 Hash Price at $30, Power Costs and Equipment Efficiency Determine Profitability

Analyzing that the hash price is around the industry average break-even point is not sufficient to assess the profitability of individual operators. Even with the same hash price, actual margins can vary significantly depending on equipment efficiency and power costs.

The Hashrate Index presented energy hash prices based on equipment efficiency as of August 17. This indicator shows the revenue that can be obtained when 1 MWh of power is used for mining. When the hash price is $31.89, the latest equipment under 14 J/TH earns $107 per MWh, 14-19 J/TH earns $79, 19-25 J/TH earns $59, and older generation equipment of 25-38 J/TH earns $41.

Based on this, we can proportionally convert the profits at hash prices of $30, $40, and $50, and apply power costs of 4, 6, and 8 cents per kWh to compare cash margins excluding electricity costs (self-calculated, Appendix ②). However, hosting costs, labor costs, and depreciation are not included, so actual profitability is likely lower than this.

When the hash price drops to $30, the profitability differences based on equipment efficiency become significant. Older generation equipment of 25-38 J/TH sees cash margins nearly disappear with a power cost of just 4 cents per kWh. The break-even power cost for 19-25 J/TH equipment is about 5.6 cents, while for 14-19 J/TH equipment, it is about 7.4 cents.

If the hash price rises to $40, older generation equipment can still leave a margin of about $11 per MWh at a power cost of 4 cents, but it returns to a loss starting from 6 cents. Ultimately, the operators with the most pressure at this low point are those with less efficient older generation equipment and those bearing high power costs. This aligns with CoinShares' estimate that 15-20% of older generation equipment is in a loss state.

Comparing MARA's actual operational metrics allows for a more detailed view of its profitability levels. MARA reported an electricity cost of $0.04 per kWh and a daily cost of $27.7 per PH/s. Assuming an equipment efficiency of 17.5 J/TH, approximately 0.42 MWh of power is needed to operate 1 PH/s for a day. If the electricity cost is 4 cents, the daily electricity expense would be about $16.8 per PH/s.

Subtracting the electricity cost from $27.7 leaves approximately $11. However, since the detailed components of the $27.7 cost are not disclosed, we cannot definitively conclude that this difference is entirely operational costs excluding electricity. When simply comparing the total cost of $27.7 to the hash price of $31.89 at that time, about $4.2 remains per PH/s per day.

This indicates that even large operators with low-cost electricity did not have significant margin capacity at this low point. The higher the electricity cost or the lower the equipment efficiency, the greater the likelihood of cash-based losses.

Mining Revenue Rebounds in July, Recovering Only 16% of June's Decline
According to Newhedge.io, monthly mining revenue increased from $947.26 million in April to $1.086 billion in May. In June, it dropped to $836.41 million, a 23% decrease from the previous month, and in July, it slightly rebounded to $875.35 million. The increase of $38.94 million in July accounted for only about 16% of the $249.59 million decline in June (self-calculated). Despite the rebound in July, most of the decline in June was not recovered.

In June, the hash price fell to an all-time low, leading to shutdowns primarily among less profitable miners, and the difficulty also decreased by 10.09%.

In August, the weekly transaction fees accounted for only 0.69% to 0.77% of the total block rewards. This is a minimal level to compensate for the deterioration in mining profitability. Currently, most mining revenue comes from block subsidies, meaning it is heavily influenced by fluctuations in Bitcoin prices.

Difficulty Adjustments in 2026: 10 Out of 17 Decreases
Difficulty adjustments are procedures that adjust mining difficulty according to changes in the network hash rate approximately every two weeks (2,016 blocks). A decrease in difficulty indicates that the hash rate input into the network during that period has decreased. However, it is difficult to determine whether this is due to deteriorating mining profitability or power market factors such as reduced operational capacity due to increased summer electricity demand based solely on difficulty changes.

As of August 22, 2026, there have been a total of 17 difficulty adjustments, of which 10 were decreases and 7 were increases. The difficulty dropped from around 148.25T at the beginning of the year to 125.81T, a decrease of about 15.1% compared to just before the first adjustment. On June 13, it fell by 10.09% to a yearly low of 124.93T, then rebounded by 7.2% to 133.87T on June 26 (self-calculated). Subsequently, on July 11, it decreased by 5.00% (127.17T), on July 25 by 0.74% (126.23T), on August 8 it increased by 0.99% (127.48T), and on August 22 it decreased by 1.31% (125.81T). As of August 22, the difficulty is 0.7% higher than the yearly low recorded on June 13.

It is difficult to explain the decrease in difficulty solely by the deterioration in mining profitability. According to analysis by index operators, the decline in June was influenced by shutdowns due to the all-time low hash price and seasonal power reductions, while the decline on July 11 was significantly impacted by operational reductions in response to peak demand on the Texas power grid (ERCOT). While it has been confirmed that the deterioration in mining profitability was one of the causes of the decrease in hash rate, it is not the sole factor explaining the overall decline in difficulty.

BIP-110: Support from Miners Only 2.53%, Activation Failed
BIP-110 is a soft fork proposal aimed at limiting the recording of non-financial data on the Bitcoin blockchain. It includes provisions to limit output scripts to 34 bytes and OP_RETURN outputs to 83 bytes. It entered the mandatory signaling phase starting from block 961,632 on August 8, 2026.

Out of 2,016 blocks prior to the mandatory signaling, only 51 blocks sent support signals for BIP-110, accounting for just 2.53% of the total. This is significantly below the 55% needed for early activation, and no major mining pools publicly expressed support. Some nodes applying BIP-110 formed a separate chain but soon lagged behind the main chain.

During the first cycle of mandatory signaling, the 477th difficulty cycle saw no support signals among the 2,016 blocks, resulting in a signaling rate of 0.00%, and ultimately it was not activated. The highest support rate during the previous voluntary signaling period (cycles 465-476) was also 2.53%.

No significant changes were observed in the main chain tracked by the dashboard due to BIP-110 signaling. However, the separate chain applying BIP-110 was not included in the dashboard's tracking scope, making it difficult to determine the overall hash power movement based solely on this data.

MARA and CleanSpark's BTC Disposal Amidst Mining Profitability Pressure; AI and HPC Contracts Surpass $70 Billion
As the pressure on mining profitability increases, the responses from publicly listed mining companies are becoming more diverse. Movements to dispose of mined Bitcoin and the expansion of AI and high-performance computing (HPC) contracts are occurring simultaneously.

Bitcoin Disposal. MARA mined 2,422 BTC in the second quarter and sold 2,213 BTC at an average price of $73,078, which is about 91% of its mining output (self-calculation)[10]. Revenue for the same quarter was $174.9 million, a 27% decrease compared to the same period last year, and its Bitcoin holdings decreased by 29% to 35,577 BTC[10].

CleanSpark disposed of 579 BTC at an average price of $66,133 out of 586 BTC mined in July, which is about 99% of its mining output (self-calculation)[9]. However, 350 BTC of this was due to the exercise of call options, with the discretionary spot sales amounting to 229 BTC, about 39% of the monthly mining output (self-calculation)[9].

Both companies did not retain a significant portion of the Bitcoin mined during the period, but the nature of their disposals differed. MARA focused on spot sales, while CleanSpark had more than half of its disposals as deliveries from call options. Therefore, it is difficult to view these two cases as cashing out of the same nature.

Expansion of AI·HPC Contracts. CleanSpark signed a 20-year triple net lease for a 175 MW facility at its Sandersville campus in Georgia in July. According to the company's announcement, the contract revenue is approximately $6.6 billion, which could reach up to $11.6 billion if the extension option is exercised[9].

The total announced amount of AI·HPC contracts by listed mining companies has exceeded $70 billion according to CoinShares' report for the first quarter of 2026[11]. This figure does not only refer to contracts newly signed in the first quarter of 2026 but also includes contracts announced up to that point, including those from 2025. The same report estimated the infrastructure construction costs for mining at $700,000 to $1 million per MW, while for AI, it is estimated at $8 million to $15 million[11]. This indicates that transitioning to AI·HPC requires significantly larger initial capital than mining.

Subsequent large contracts continued. Riot Platforms announced a 20-year lease for a 191 MW facility on August 10. The company only referred to the tenant as a 'leading frontier AI research lab', with the base contract revenue being approximately $9.1 billion, and up to $16.1 billion if the extension option is exercised[16]. The tenant being Anthropic is based on subsequent reports rather than the company's announcement[17].

The EV/NTM Sales multiple for mining companies with HPC contracts is 12.3 times, more than double that of pure mining companies at 5.9 times (as of Q4 2025)[11]. However, this cannot be directly interpreted as a difference in actual profitability between the two businesses. The market's expectations for AI·HPC contracts may have been reflected in the corporate valuation. CoinShares also pointed out that for this evaluation to be justified, actual contract execution must be supported[11].

CoinShares interpreted this business transition as a choice based on economic viability. While the hash price remains low, pressuring mining profitability, AI infrastructure is expected to yield relatively high and stable returns[11]. However, since most contracts are predicated on the construction of new data centers, there is also a possibility that some existing mining facilities may be reduced or closed[11].

Reports covering the CoinShares report explained that large borrowings and Bitcoin sales are being utilized as funding sources for the transition to AI·HPC[18]. Therefore, there is a possibility that the disposal of Bitcoin and the expansion of AI·HPC contracts are occurring within the same business transition process. However, it has not been confirmed through disclosures where the proceeds from individual companies' Bitcoin disposals were actually used (Appendix ①).

However, not all large mining companies are focusing solely on the AI·HPC transition. MARA proposed a strategy to maintain mining competitiveness based on low-cost power while ensuring the same cost competitiveness even if some facilities are converted for other uses[12]. This involves allocating power assets to the most economically viable uses between mining and AI. Therefore, it is difficult to conclude that a high rate of Bitcoin disposal necessarily indicates a reduction in mining operations or a transition to AI·HPC.

Price Rebound in Late August, Still Early to Consider a Trend Change {#rps-9}

The price of Bitcoin rose by 19.6% from $64,135 on August 18 to $76,712 on August 21 (based on each query point, self-calculation)[19][20]. On August 31, it was $78,532[22], and the drop from the all-time high in October 2025 has also decreased to 38.8%[5].

The hash price also rebounded. According to Hashrate Index, it rose by 20.41% from $31.80 on August 18 to $38.29 on August 22, reaching its highest level since May[6]. On August 31, it was $39.36, and the average for the month of August was $34.63, which is below the lower end of 2025 at $37.89[22]. The comparison periods for Bitcoin price and hash price differ by one day due to differences in the reference dates.

Monthly mining revenue also rebounded. According to Newhedge.io, August mining revenue was $1.00839 billion, about 15% higher than July's $875.35 million (self-calculation)[23]. However, transaction fees among this were only $7.19 million, accounting for 0.7% of the total[23].

This rebound can also be interpreted as premature to view the deterioration of mining profitability as a structural trend. In the past, there have been instances where the hash price fell near the breakeven point and then recovered. This year, mining revenue also recovered from the low in March to April-May[1][15], but it did not last more than two months and fell back to an all-time low in June[1][15].

CoinShares believes that the recovery of the hash price will be significantly influenced by the price of Bitcoin, estimating that the proportion of older generation mining machines that fall into losses at a hash price of around $30 is 15-20%[11]. This indicates that a considerable number of mining machines can continue to operate even at the $30 level.

As of August 31, the forward market reflected an average hash price of $37.59 for the next six months, significantly up from $30.67 before the rebound on August 17[8][22]. This is close to the lower end of 2025 at $37.89. The period of the recent rebound has also only been about two weeks. Therefore, it is premature to conclude that the structural pressure on mining profitability has been alleviated based solely on the recent rebound.

CONCLUSION

Conclusion: Structural Pressure on Mining Profitability and Diversification of Business Models

Bitplanet Research Lab views the deterioration of mining profitability in 2026 as closer to structural pressure than a temporary fluctuation. All three records for the lowest average monthly hash prices in history were set this year[1], and out of 17 difficulty adjustments up to August 22, 10 were decreases[5]. There has been a movement to dispose of mined Bitcoin on a large scale at both MARA and CleanSpark[9][10].

In contrast, in late August, the price of Bitcoin and hash price rebounded quickly. CoinShares estimated that a significant number of mining machines, excluding 15-20% of older generation equipment, could operate even at a hash price level of around $30. This is a signal contrary to the judgment of structural pressure. However, the recent rebound has not continued long enough to change the monthly average trend, and the recovery since the low point in March of this year has also been reversed after two months, making it too early to change the current judgment.

Currently, there are two conditions to reconsider the judgment. First, if the structural pressure on mining profitability actually eases. If the monthly average hash price exceeds $37.89 for more than two months, and the difficulty continues to rise, the judgment of structural pressure will be reconsidered.

The reason for using $37.89 as a benchmark is that 2025 is the first complete year after the halving, and a lower monthly average hash price has only appeared in 2026. If the hash price exceeds this level for more than two months and the difficulty also rises, it can be seen as a signal that mining profitability is recovering to 2025 levels, beyond a short-term rebound.

Second, the judgment regarding the diversification of business models will be reconsidered if cancellations or delays in AI and HPC contracts accumulate, or if the hash price recovers to 2025 levels, leading to a clear improvement in the profitability of mining itself.

If the current low hash price continues, mining companies with high electricity costs or low equipment efficiency may face greater pressure. In contrast, operators who secure low-cost electricity or long-term HPC contracts have relatively more options. In the current revenue structure, where the proportion of transaction fees is less than 1%, the key variables that operators can directly control are electricity costs, equipment efficiency, and the utilization of power assets.

Conflict of Numbers Processing: When different figures are confirmed, both sides are presented together. In cases where a reference value needs to be selected, priority is given to the disclosure of the range of estimates over the recognition of the media, the clarity of the query timing, and the possibility of separating detailed items, with the basis for selection specified in the text.

Self-Generated Figures: Self-generated figures are presented individually in Appendix ② with each formula, reference date, and limitations.

Legal and Regulatory Interpretation: Data from regulatory agencies is quoted only within the scope of explaining general structural classifications and applicability, and is not to be broadly interpreted as official classifications or approvals for specific products. Explanations of product structures and regulatory interpretations are separated into distinct paragraphs.

Conflict of Interest Disclosure: The same disclosure statement is published in a fixed position at the top of all reports.

Publication Information

Published by: Bitplanet Research Lab | Written by: Kim Tae-won | Reviewed by: Kim Soo-young | Publication Date: 2026-09-02

This material is prepared for informational purposes and is not an investment solicitation.

Comments 0