Each published estimate against the model at the same date
| Date | Source | Measure | Published | Model band (p10 to p90) | Model average | Against the band |
|---|---|---|---|---|---|---|
| September 24, 2026 | JPMorgan | all-in cost | $84,948 | $43,488 to $68,107 | $54,416 | above the band |
| June 30, 2026 | JPMorgan | all-in cost | $78,000 | $46,638 to $70,720 | $57,405 | above the band |
| June 30, 2026 | CoinShares | cash costs | $75,500 | $46,638 to $70,720 | $57,405 | above the band |
| March 22, 2026 | Capriole | electricity only | $46,426 | $50,900 to $79,453 | $63,011 | below the band |
| March 22, 2026 | Capriole | all-in cost | $58,032 | $50,900 to $79,453 | $63,011 | inside the band, −8% vs the average |
| December 31, 2025 | CoinShares | cash costs | $79,995 | $46,680 to $78,643 | $61,012 | above the band |
The band runs from the 10th to the 90th percentile of the modelled fleet: the cheapest machines on the cheapest power at the bottom, the oldest machines on the dearest power at the top. All figures are in US dollars per bitcoin.
Why the figures differ
The model estimates the whole network: every machine, weighted by its share of the hashrate. CoinShares averages the costs reported by listed mining companies, which carry overheads the model does not see (staff, offices, financing). JPMorgan does not detail its method in the press coverage of its notes. Capriole publishes two lines: the electricity cost alone, and a full production cost. The last column of the table says where each figure falls against the band on its date.
Sources
- The Block, 24/09/2026: JPMorgan, production cost around $85,000
- ForkLog, 25/09/2026: JPMorgan chart, cost of production $84,948
- Bitcoin Magazine, 19/06/2026: JPMorgan, all-in cost about $78,000
- ForkLog, 16/09/2026: CoinShares Q2 2026, about $75,500
- KuCoin News, 13/04/2026: Capriole on 22/03/2026, $46,426 electricity and $58,032 full cost