CoinShares Bitcoin Mining Report — Brief Q3 Earnings Update | by CointimeSG

From CoinShares Research Blog by James Butterfill

Authors: James Butterfill — Head of Research & Max Shannon — Research Analyst

The weighted average cash cost to produce one bitcoin among listed miners rose to ~US$55,950 in Q3 from ~US$49,500 in Q2, a 13% increase, with mining remaining profitable at current prices of $100k. Including non-cash costs such as depreciation and stock-based compensation, the average cost increases to US$106,000.

The increase in cost of production overall is driven by the network hashrate outpacing public miner hashrate growth, therefore reducing their share of the network’s mined bitcoin as they become a smaller percentage of the overall network. There are a few factors at play:

The AI craze has diverted capital away from scaling mining operations. Whilst miners are well suited to satisfy the needs of hyperscalers who need substantial clean energy in the pipeline, this does distract miners from scaling their mining operations whilst network hashrate continues to grow regardless.
Some miners have also been distracted by the HODL premium, whereby miners issue convertible notes at no or very low coupon rates to which they proceed to buy bitcoin as a Treasury Reserve asset, instead of, again, using the capital to further scale operations. This is also in conjunction with the fact that the ROIC of buying and HODLing bitcoin is better than mining bitcoin, as shown in our last quarterly report.
Many public miners operate in Texas, where higher summer power costs further increase the cost of production, as well as curtailment capabilities further reducing the gross bitcoin mined (although improving the electricity cost per bitcoin, if done well).

Cost to Mine (USD)

Cost of revenues (electricity) represents the self-mining cost of revenues, mostly electricity cost.
SG&A eliminates non-cash expenses, such as stock-based compensation, one-time payments, and is netted of non-cash costs such as stock-based compensation and depreciation, as well as pro-rata based on the self-mining revenues as a portion of total revenues.
Taxes are calculated using valuation allowances to determine income tax expenses and benefits, wherever possible, as well as pro-rata based on the self-mining revenues as a portion of total revenues.
Interest expense accounts solely for interest on debt, excluding lease expenses and other finance costs, where possible, although is offset by interest income, where applicable. As well as pro-rata based on the self-mining revenues as a portion of total revenues.
Stock-based Compensation is derived from P&L/Cash Flows, as well as pro-rata based on the self-mining revenues as a portion of total revenues.
Depreciation and amortisation (D&A) is pro-rata based on the self-mining revenues as a portion of total revenues.

Cormint — a private miner — which was once the second lowest cash-cost miner, has now increased to the third highest cash-cost miner. This was because of a significant spike in their power costs for the quarter, with a ~$20m realised loss in their hedging strategy hurting their P&L. The Q3 summer is typically Cormint’s worst performing quarter given the market dynamics in Texas power markets. Despite this Cormint has been CoinShares’ lowest cost miner for three quarters in a row, therefore, we suspect a return to the mean is likely in Q4, and full year costs to be market-leading given YTD power costs. But it highlights the challenges of effectively hedging in the power market.

Terawulf, which was the 5th lowest cash-cost miner in Q2, has moved to third lowest cash-cost miner in Q3 because of a 20% decrease in cash costs quarter-over-quarter because their interest expense decreased 92% quarter-over-quarter from $5.3m to $0.409m. This relates primarily to the borrowings under the Loan, Guaranty and Security Agreement (the “LGSA”) with Wilmington Trust (the “Term Loans”), which had an original maturity date of December 1, 2024 and was fully repaid in July 2024 ahead of maturity.

Marathon was previously the fourth lowest cash-cost miner for Q2, but has now moved to first place. This is because they are one of only four miners that have increased the number of bitcoin mined quarter-over-quarter (albeit only 1%), alongside Terawulf, Riot and Bitfarms. Marathon also benefitted from a large tax benefit that primarily arises from the release of the valuation allowance on deferred tax assets, driven by the increase in bitcoin’s fair value and positive forecasts for its future value.

Riot has moved down the order one spot to the seventh lowest cash-cost miner despite their improved operations as seen by their increased uptime (11%), hashrate (52%), efficiency (-6%) and bitcoin mined (31%), respectively QoQ.

Outlook for Q4 results & 2025

Q4 hashprice has increased marginally, driven by a surging $100k Bitcoin price post-election. This development should temporarily relieve miners from the suppressed hashprice resulting from halved block rewards in the first full quarter post-halving.

2025 could bring some interesting and profitable AI opportunities, our prediction is from TeraWulf and Cipher due to their relationships and partnerships with energy companies, as well as their large GWs and clean energy. However, it may take some time for the income from these deals to reflect in the P&L.

On the cost-of-production front, we anticipate that upward pressures will persist due to a combination of factors:

Miners are diversifying their strategies, such as HODLing Bitcoin or investing in AI, which reduces their own hashrate growth and, consequently, Bitcoin production.
Miners continue to purchase land and access to power but struggle to secure low prices due to competition from more profitable hyperscalers, who are bidding up land and power assets.
A high correlation between machine prices and Bitcoin prices means that an increase in Bitcoin’s price will likely drive up machine costs, thereby increasing capital expenditure, as well as raising depreciation expenses in upcoming quarters.
Looser, more liquid debt and convertible note markets will likely encourage miners to issue more debt, increasing interest expenses. We do not expect all miners to secure zero-coupon convertible debt.
We also foresee potential unfavorable outcomes, such as Chapter 11 filings or insolvency, for some public miners like Argo if Bitcoin prices decline. This risk is heightened by negative shareholder equity and the inability to raise capital as fundraising windows close.

Conclusion

The CoinShares Q3 Bitcoin Mining Report reveals a moderate rise in the average cash cost of mining bitcoin to ~$55,950, a 13% increase from Q2, with total costs, including non-cash expenses, reaching ~$106,000. Public miners face declining shares of mined bitcoin due to the network’s growing hashrate amidst distractions like AI infrastructure or bitcoin HODL strategies. Regional factors, such as Texas’s high summer power costs, further challenge profitability and increase cost of production measures.

Among individual miners, Terawulf became the lowest-cost producer after significantly reducing debt expenses as Cormint learned lessons from their failed power hedge strategy, while Riot and Marathon both increased their bitcoin production quarter-over-quarter. Next quarter should provide a temporary relief in depressed hashprices, yet it is important for miners to carry on focusing on reducing expenses, mainly power costs, to survive lower network profitability into the future.