Skip to the content
MinerTerminal

Analysis: the figures on file tested in pairs across the miners, and each figure's spread across them

Follow a company

Pairs

10 pairs, 15 of 16 tests with a fitFormulas
How to read it
Each card is one question the brief asked of the figures: the scatter with the least-squares line, the full read-out, and the inputs. A pair on two sizes also carries the test on log10 of both.

Revenue against hashrate

n 10 of 18Formulas
Why this pair
Revenue is a share of network hashrate times the bitcoin price (spec 7.1), so realized hashrate should carry annualized revenue.

Annualized revenue against realized hashrate, 10 miners

USD per year against EH/s · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1Tier 2

Reading: The least-squares line accounts for 28 percent of the spread (R squared 0.28), with p 0.112 for its slope, on 10 points.

Annualized revenue against realized hashrate, 10 miners. The least-squares line accounts for 28 percent of the spread (R squared 0.28), with p 0.112 for its slope, on 10 points.
CompanyCoverage tierHashrate, realized (EH/s)Revenue, annualized from the latest reported period (USD per year)
BTBTTier 20.4$128.5M
BTDRTier 176.7$451.4M
CIFRTier 111.6$99.3M
CLSKTier 138.6$766.3M
CORZTier 115.7$656.8M
HUTTier 128.1$299.7M
IRENTier 123.2$707M
MARATier 170.3$699.5M
RIOTTier 144.4$696.9M
WULFTier 19.3$179.1M
Least-squares lineslope 5582723.880114, intercept 290762398.895966, fitted in levels
The statistics of annualized revenue on realized hashrate
n10
degrees of freedom8
slope5.58M95% interval -1.63M to 12.8M
intercept291M
SST651,000Ttotal sum of squares
SSR185,000Texplained by the fit
SSE466,000Tleft over
R squared0.28SSR over SST
t, slope1.79standard error 3.13M
p, slope0.112two-sided
Pearson r0.53p 0.112
Spearman rho0.60p 0.067
x mean31.8sd 25.7, variance 661, median 25.7
y mean468Msd 269M, variance 72,300T, median 554M

computed 2026-09-22 from data as of 2026-09-06, written to data/analysis/cross_section.json; formulas

Each company's two figures, with period, as-of and source (10 of 18)
Each company's hashrate, realized and revenue, annualized from the latest reported period, with period, as-of and source
BTBT0.4period not stated, as of 2026-06-30, source$128.5M (derived)quarter ended 2026-06-30, times 4, as of 2026-06-30, source
BTDR76.7July 2026, as of 2026-07-31, source$451.4M (derived)half-year ended 2025-06-30, times 2, as of 2025-06-30, source
CIFR11.6Q2 2026, as of 2026-08-04, source$99.3M (derived)quarter ended 2026-06-30, times 4, as of 2026-06-30, source
CLSK38.6July 2026, as of 2026-07-31, source$766.3Myear ended 2025-09-30, as of 2025-09-30, source
CORZ15.7FY2025, as of 2025-12-31, source$656.8M (derived)quarter ended 2026-06-30, times 4, as of 2026-06-30, source
HUT28.1Q2 2026, as of 2026-06-30, source$299.7M (derived)quarter ended 2026-06-30, times 4, as of 2026-06-30, source
IREN23.2FY26 year-end, as of 2026-06-30, source$707Myear ended 2026-06-30, as of 2026-06-30, source
MARA70.3Q2 2026, as of 2026-06-30, source$699.5M (derived)quarter ended 2026-06-30, times 4, as of 2026-06-30, source
RIOT44.4Q2 2026, as of 2026-06-30, source$696.9M (derived)quarter ended 2026-06-30, times 4, as of 2026-06-30, source
WULF9.3FY2025 year-end, as of 2025-12-31, source$179.1M (derived)quarter ended 2026-06-30, times 4, as of 2026-06-30, source

Not in the sample: HIVE (realized hashrate: status reported-not-comparable: This is installed hashrate, not realized: the filing says the total installed hashrate was approximately 25.2 EH/s, and the same quarter's release gives 24.0 EH/s as the quarter average. The Tier 1 companies report self-mining hashrate actually produced. The company's words: "As of July 31, 2026, the Company's total installed hashrate was approximately 25.2 EH/s with an implied efficiency of 16.5 J/TH". The Q1 FY2027 release gives 24.0 EH/s as the quarter average.); KEEL (realized hashrate: Not disclosed by the company in the documents read on 2026-09-06.); DMGGF (realized hashrate: Not disclosed by the company in the documents read on 2026-09-06.); ARBK (realized hashrate: status reported-not-comparable: Fleet capability, not realized hashrate: the 20-F says the fleet 'can generate up to 2.4 exahash per second'. It is also the oldest hero figure on the site, from the fiscal 2025 balance date. The company's words: "As of December 31, 2025, we had a fleet of approximately 22,600 machines mining Bitcoin that can generate up to 2.4 exahash per second of computing power."); VIP (realized hashrate: Not disclosed by the company in the documents read on 2026-09-06.); DGXX (realized hashrate: no research pack carries this company); ANY (realized hashrate: no research pack carries this company); BGDE (realized hashrate: no research pack carries this company).

Market cap against BTC held

n 8 of 18Formulas
Why this pair
Whether the market values a miner by the bitcoin on its balance sheet.

Bitcoin held against market capitalization, 8 miners

BTC against USD · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1

Reading: The least-squares line accounts for 7 percent of the spread (R squared 0.07), with p 0.538 for its slope, on 8 points.

Bitcoin held against market capitalization, 8 miners. The least-squares line accounts for 7 percent of the spread (R squared 0.07), with p 0.538 for its slope, on 8 points.
CompanyCoverage tierMarket capitalization (USD)Bitcoin held (BTC)
BTDRTier 1$3.36B257
CIFRTier 1$7.36B646
CLSKTier 1$3.26B13,931
CORZTier 1$5.75B848
HUTTier 1$11.53B17,316
IRENTier 1$17.61B0
MARATier 1$4.37B35,577
RIOTTier 1$8.18B11,380
Least-squares lineslope -6.59065e-7, intercept 15054.482532, fitted in levels
The statistics of bitcoin held on market capitalization
n8
degrees of freedom6
slope-0.00000065995% interval -0.00000313 to 0.00000181
intercept15.1K
SST1.09Btotal sum of squares
SSR72.4Mexplained by the fit
SSE1.02Bleft over
R squared0.07SSR over SST
t, slope-0.65standard error 0.00000101
p, slope0.538two-sided
Pearson r-0.26p 0.538
Spearman rho-0.24p 0.570
x mean7.68Bsd 4.88B, variance 23,800,000T, median 6.56B
y mean9.99Ksd 12.5K, variance 156M, median 6.11K

computed 2026-09-22 from data as of 2026-09-06, written to data/analysis/cross_section.json; formulas

Each company's two figures, with period, as-of and source (8 of 18)
Each company's market capitalization and bitcoin held, with period, as-of and source
BTDR$3.36B (computed)spot, as of 2026-09-04, source257July 2026, as of 2026-07-31, source
CIFR$7.36B (computed)spot, as of 2026-09-04, source646Q2 2026, as of 2026-06-30, source
CLSK$3.26B (computed)spot, as of 2026-09-04, source13,931July 2026, as of 2026-07-31, source
CORZ$5.75B (computed)spot, as of 2026-09-04, source848Q2 2026, as of 2026-06-30, source
HUT$11.53B (computed)spot, as of 2026-09-04, source17,316Q2 2026, as of 2026-06-30, source
IREN$17.61B (computed)spot, as of 2026-09-04, source0FY26 year-end, as of 2026-06-30, source
MARA$4.37B (computed)spot, as of 2026-09-04, source35,577Q2 2026, as of 2026-06-30, source
RIOT$8.18B (computed)spot, as of 2026-09-04, source11,380Q2 2026, as of 2026-06-30, source

Not in the sample: WULF (bitcoin held: BTC count not disclosed; digital assets carried at $133 thousand fair value at 2026-06-30, so holdings are de minimis. 10-Q Note 1: 'Bitcoin earned as consideration is routinely converted to U.S. dollars.'); BTBT (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $83.6M, debt $336.2M, and 360.6M shares are on file; the price is the term that is missing.); HIVE (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $208M, debt $234.9M, and 274.3M shares are on file; the price is the term that is missing.); KEEL (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $768.9M, debt $1.03B, and 617.6M shares are on file; the price is the term that is missing.); DMGGF (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. None of cash, debt or the share count is on file; the price is the term that is missing. DMG files nothing with the SEC and trades over the counter, so its figures are graded as company releases.); ARBK (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $2.2M, debt $40.2M, and 28.9B shares are on file; the price is the term that is missing. Argo trades as ADSs worth 2,160 ordinary shares each, so an ADS price multiplied by the ordinary share count would overstate market capitalization by that ratio; the figure is left null rather than computed on a mismatched pair.); VIP (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $3.2M, debt $39.5M, and 18M shares are on file; the price is the term that is missing.); DGXX (market capitalization: no research pack carries this company); ANY (market capitalization: no research pack carries this company); BGDE (market capitalization: no research pack carries this company).

Market cap against EV/EH

n 6 of 18Formulas
Why this pair
Whether larger miners trade at a higher or lower price per exahash than smaller ones.

EV per EH/s against market capitalization, 6 miners

USD per EH/s against USD · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1

Reading: The least-squares line accounts for 38 percent of the spread (R squared 0.38), with p 0.190 for its slope, on 6 points.

EV per EH/s against market capitalization, 6 miners. The least-squares line accounts for 38 percent of the spread (R squared 0.38), with p 0.190 for its slope, on 6 points.
CompanyCoverage tierMarket capitalization (USD)EV per EH/s (USD per EH/s)
BTDRTier 1$3.36B$61.9M
CIFRTier 1$7.36B$1.1B
CLSKTier 1$3.26B$125.4M
HUTTier 1$11.53B$677.4M
MARATier 1$4.37B$91M
RIOTTier 1$8.18B$191.5M
Least-squares lineslope 0.078634, intercept -127427215.901797, fitted in levels

6 points: the test is computed but has little power, and one company can move it.

The statistics of EV per EH/s on market capitalization
n6
degrees of freedom4
slope0.078695% interval -0.0599 to 0.217
intercept-127M
SST865,000Ttotal sum of squares
SSR331,000Texplained by the fit
SSE534,000Tleft over
R squared0.38SSR over SST
t, slope1.58standard error 0.0499
p, slope0.190two-sided
Pearson r0.62p 0.190
Spearman rho0.66p 0.156
x mean6.34Bsd 3.27B, variance 10,700,000T, median 5.87B
y mean371Msd 416M, variance 173,000T, median 158M

computed 2026-09-22 from data as of 2026-09-06, written to data/analysis/cross_section.json; formulas

Each company's two figures, with period, as-of and source (6 of 18)
Each company's market capitalization and ev per eh/s, with period, as-of and source
BTDR$3.36B (computed)spot, as of 2026-09-04, source$61.9M (computed)spot, as of 2026-09-04, source
CIFR$7.36B (computed)spot, as of 2026-09-04, source$1.1B (computed)spot, as of 2026-09-04, source
CLSK$3.26B (computed)spot, as of 2026-09-04, source$125.4M (computed)spot, as of 2026-09-04, source
HUT$11.53B (computed)spot, as of 2026-09-04, source$677.4M (computed)spot, as of 2026-09-04, source
MARA$4.37B (computed)spot, as of 2026-09-04, source$91M (computed)spot, as of 2026-09-04, source
RIOT$8.18B (computed)spot, as of 2026-09-04, source$191.5M (computed)spot, as of 2026-09-04, source

Not in the sample: CORZ (EV per EH/s: status computed-not-comparable: all inputs present; BTC holdings not netted (shown separately); CAVEAT: restricted cash is not netted in EV (spec formula uses cash and equivalents); see the dossier cash note for the restricted balance; Not comparable: HPC/AI revenue is 83% of quarterly revenue, so a hashrate denominator or a mining-cost run-rate does not describe this company); IREN (EV per EH/s: status computed-not-comparable: all inputs present; BTC holdings not netted (shown separately); CAVEAT: restricted cash is not netted in EV (spec formula uses cash and equivalents); see the dossier cash note for the restricted balance; Not comparable: HPC/AI revenue is 51% of quarterly revenue, so a hashrate denominator or a mining-cost run-rate does not describe this company); WULF (EV per EH/s: status computed-not-comparable: all inputs present; BTC holdings not netted (shown separately); CAVEAT: restricted cash is not netted in EV (spec formula uses cash and equivalents); see the dossier cash note for the restricted balance; Not comparable: HPC/AI revenue is 71% of quarterly revenue, so a hashrate denominator or a mining-cost run-rate does not describe this company); BTBT (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $83.6M, debt $336.2M, and 360.6M shares are on file; the price is the term that is missing.); HIVE (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $208M, debt $234.9M, and 274.3M shares are on file; the price is the term that is missing.); KEEL (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $768.9M, debt $1.03B, and 617.6M shares are on file; the price is the term that is missing.); DMGGF (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. None of cash, debt or the share count is on file; the price is the term that is missing. DMG files nothing with the SEC and trades over the counter, so its figures are graded as company releases.); ARBK (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $2.2M, debt $40.2M, and 28.9B shares are on file; the price is the term that is missing. Argo trades as ADSs worth 2,160 ordinary shares each, so an ADS price multiplied by the ordinary share count would overstate market capitalization by that ratio; the figure is left null rather than computed on a mismatched pair.); VIP (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $3.2M, debt $39.5M, and 18M shares are on file; the price is the term that is missing.); DGXX (market capitalization: no research pack carries this company); ANY (market capitalization: no research pack carries this company); BGDE (market capitalization: no research pack carries this company).

Market cap against hashrate

n 9 of 18Formulas
Why this pair
Whether the market values a miner by the hashrate it runs.

Realized hashrate against market capitalization, 9 miners

EH/s against USD · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1

Reading: The least-squares line accounts for 22 percent of the spread (R squared 0.22), with p 0.203 for its slope, on 9 points.

Realized hashrate against market capitalization, 9 miners. The least-squares line accounts for 22 percent of the spread (R squared 0.22), with p 0.203 for its slope, on 9 points.
CompanyCoverage tierMarket capitalization (USD)Hashrate, realized (EH/s)
BTDRTier 1$3.36B76.7
CIFRTier 1$7.36B11.6
CLSKTier 1$3.26B38.6
CORZTier 1$5.75B15.7
HUTTier 1$11.53B28.1
IRENTier 1$17.61B23.2
MARATier 1$4.37B70.3
RIOTTier 1$8.18B44.4
WULFTier 1$8.24B9.3
Least-squares lineslope -2.52652e-9, intercept 54.877352, fitted in levels
The statistics of realized hashrate on market capitalization
n9
degrees of freedom7
slope-0.0000000025395% interval -0.00000000678 to 0.00000000173
intercept54.9
SST4.85Ktotal sum of squares
SSR1.07Kexplained by the fit
SSE3.79Kleft over
R squared0.22SSR over SST
t, slope-1.40standard error 0.0000000018
p, slope0.203two-sided
Pearson r-0.47p 0.203
Spearman rho-0.50p 0.170
x mean7.74Bsd 4.57B, variance 20,900,000T, median 7.36B
y mean35.3sd 24.6, variance 607, median 28.1

computed 2026-09-22 from data as of 2026-09-06, written to data/analysis/cross_section.json; formulas

Each company's two figures, with period, as-of and source (9 of 18)
Each company's market capitalization and hashrate, realized, with period, as-of and source
BTDR$3.36B (computed)spot, as of 2026-09-04, source76.7July 2026, as of 2026-07-31, source
CIFR$7.36B (computed)spot, as of 2026-09-04, source11.6Q2 2026, as of 2026-08-04, source
CLSK$3.26B (computed)spot, as of 2026-09-04, source38.6July 2026, as of 2026-07-31, source
CORZ$5.75B (computed)spot, as of 2026-09-04, source15.7FY2025, as of 2025-12-31, source
HUT$11.53B (computed)spot, as of 2026-09-04, source28.1Q2 2026, as of 2026-06-30, source
IREN$17.61B (computed)spot, as of 2026-09-04, source23.2FY26 year-end, as of 2026-06-30, source
MARA$4.37B (computed)spot, as of 2026-09-04, source70.3Q2 2026, as of 2026-06-30, source
RIOT$8.18B (computed)spot, as of 2026-09-04, source44.4Q2 2026, as of 2026-06-30, source
WULF$8.24B (computed)spot, as of 2026-09-04, source9.3FY2025 year-end, as of 2025-12-31, source

Not in the sample: BTBT (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $83.6M, debt $336.2M, and 360.6M shares are on file; the price is the term that is missing.); HIVE (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $208M, debt $234.9M, and 274.3M shares are on file; the price is the term that is missing.); KEEL (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $768.9M, debt $1.03B, and 617.6M shares are on file; the price is the term that is missing.); DMGGF (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. None of cash, debt or the share count is on file; the price is the term that is missing. DMG files nothing with the SEC and trades over the counter, so its figures are graded as company releases.); ARBK (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $2.2M, debt $40.2M, and 28.9B shares are on file; the price is the term that is missing. Argo trades as ADSs worth 2,160 ordinary shares each, so an ADS price multiplied by the ordinary share count would overstate market capitalization by that ratio; the figure is left null rather than computed on a mismatched pair.); VIP (market capitalization: This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $3.2M, debt $39.5M, and 18M shares are on file; the price is the term that is missing.); DGXX (market capitalization: no research pack carries this company); ANY (market capitalization: no research pack carries this company); BGDE (market capitalization: no research pack carries this company).

Margin against all-in cost

n 2 of 18Formulas
Why this pair
Whether a higher all-in cost per bitcoin shows up as a thinner gross margin in the filed accounts.

Gross margin against all-in cost per bitcoin, 2 miners

percent of revenue against USD per BTC · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1

Reading: 2 points are too few for a least-squares fit, so the points are drawn without a line.

Gross margin against all-in cost per bitcoin, 2 miners. 2 points are too few for a least-squares fit, so the points are drawn without a line.
CompanyCoverage tierAll-in cost per bitcoin (USD per BTC)Gross margin (percent of revenue)
CLSKTier 1$96,277+55.2%
WULFTier 1$95,611+72.3%

2 points: no test is computed under 3.

The statistics of gross margin on all-in cost per bitcoin
n2
fitnot computed
x mean95.9Ksd 471, variance 222K, median 95.9K
y mean63.8sd 12.1, variance 146, median 63.8

computed 2026-09-22 from data as of 2026-09-06, written to data/analysis/cross_section.json; formulas

Each company's two figures, with period, as-of and source (2 of 18)
Each company's all-in cost per bitcoin and gross margin, with period, as-of and source
CLSK$96,277Q3 FY2026, as of 2026-06-30, source+55.2% (derived)year ended 2025-09-30, as of 2025-09-30, source
WULF$95,611FY2025, as of 2025-12-31, source+72.3% (derived)quarter ended 2026-06-30, as of 2026-06-30, source

Not in the sample: MARA (all-in cost per bitcoin: not disclosed as a per-BTC all-in figure. The company reports cost per petahash per day ($27.7 in Q2 2026, down 4% from $28.7) instead. Derivable inputs: purchased energy $48.8 million, O&M $26.9 million, hosting and other energy $69.2 million, D&A $174.7 million against 2,422 BTC.); RIOT (gross margin: no cost of revenue tagged for the quarter ended 2026-06-30); CORZ (all-in cost per bitcoin: Not disclosed. Derived here for reference only, not a company figure: 80.9M USD / about 695 BTC = about 116,400 USD per BTC (segment cost incl. depreciation, excl. SG&A).); IREN (all-in cost per bitcoin: Not disclosed. Derived FY26 cost of revenue per BTC is about $33,400 (see direct cost note); no depreciation-inclusive figure is published.); HUT (all-in cost per bitcoin: Not disclosed by Hut 8 or ABTC (checked Q2 2026 releases and 10-Q).); CIFR (all-in cost per bitcoin: Not disclosed; Company reports no all-in cost metric.); BTDR (all-in cost per bitcoin: Not disclosed. Derived all-in cost of revenue = ($171.0M + $28.8M) / 2,694 BTC = ~$74,200 per BTC including depreciation and SBC in cost of revenue (derived here, not a company figure).); BTBT (all-in cost per bitcoin: Not disclosed by the company in the documents read on 2026-09-06.); HIVE (all-in cost per bitcoin: Not disclosed by the company in the documents read on 2026-09-06.); KEEL (all-in cost per bitcoin: Not disclosed by the company in the documents read on 2026-09-06.); DMGGF (all-in cost per bitcoin: Not disclosed by the company in the documents read on 2026-09-06.); ARBK (all-in cost per bitcoin: Not disclosed by the company in the documents read on 2026-09-06.); VIP (all-in cost per bitcoin: Not disclosed by the company in the documents read on 2026-09-06.); DGXX (all-in cost per bitcoin: no research pack carries this company); ANY (all-in cost per bitcoin: no research pack carries this company); BGDE (all-in cost per bitcoin: no research pack carries this company).

Margin against direct cost

n 4 of 18Formulas
Why this pair
The same question on the direct cost, which more companies report.

Gross margin against direct cost per bitcoin, 4 miners

percent of revenue against USD per BTC · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1

Reading: The least-squares line accounts for 0 percent of the spread (R squared 0.00), with p 0.977 for its slope, on 4 points.

Gross margin against direct cost per bitcoin, 4 miners. The least-squares line accounts for 0 percent of the spread (R squared 0.00), with p 0.977 for its slope, on 4 points.
CompanyCoverage tierDirect cost per bitcoin (USD per BTC)Gross margin (percent of revenue)
CLSKTier 1$44,406+55.2%
HUTTier 1$36,500+64.1%
IRENTier 1$26,259+68.9%
WULFTier 1$53,681+72.3%
Least-squares lineslope 0.000014443, intercept 64.560578, fitted in levels

4 points: the test is computed but has little power, and one company can move it.

The statistics of gross margin on direct cost per bitcoin
n4
degrees of freedom2
slope0.000014495% interval -0.00192 to 0.00195
intercept64.6
SST165total sum of squares
SSR0.085explained by the fit
SSE165left over
R squared0.00SSR over SST
t, slope0.03standard error 0.00045
p, slope0.977two-sided
Pearson r0.02p 0.977
Spearman rho0.20p 0.800
x mean40.2Ksd 11.7K, variance 136M, median 40.5K
y mean65.1sd 7.41, variance 55, median 66.5

computed 2026-09-22 from data as of 2026-09-06, written to data/analysis/cross_section.json; formulas

Each company's two figures, with period, as-of and source (4 of 18)
Each company's direct cost per bitcoin and gross margin, with period, as-of and source
CLSK$44,406Q3 FY2026, as of 2026-06-30, source+55.2% (derived)year ended 2025-09-30, as of 2025-09-30, source
HUT$36,500Q2 2026, as of 2026-06-30, source+64.1% (derived)quarter ended 2026-06-30, as of 2026-06-30, source
IREN$26,259June 2025, as of 2025-06-30, source+68.9% (derived)year ended 2026-06-30, as of 2026-06-30, source
WULF$53,681FY2025, as of 2025-12-31, source+72.3% (derived)quarter ended 2026-06-30, as of 2026-06-30, source

Not in the sample: MARA (gross margin: no cost of revenue tagged for the quarter ended 2026-06-30); RIOT (gross margin: no cost of revenue tagged for the quarter ended 2026-06-30); CORZ (direct cost per bitcoin: Not disclosed as a per-coin figure. Derived here for reference only, not a company figure: H1 2026 self-mining power fees 45.1M USD / about 695 BTC = about 64,900 USD per BTC (power only).); CIFR (direct cost per bitcoin: Not disclosed. Derived only: Q2 2026 cost of revenue $15.0M (primarily Odessa power) / ~346 BTC = ~$43,500 per BTC before depreciation and SG&A (derived here, not a company figure).); BTDR (direct cost per bitcoin: Not disclosed. Derived electricity-only cost = ($84.7M + $13.3M) / 2,694 BTC = ~$36,400 per BTC (derived here, excludes co-mining hosting fees in 'other costs').); BTBT (direct cost per bitcoin: Not disclosed by the company in the documents read on 2026-09-06.); HIVE (direct cost per bitcoin: Not disclosed by the company in the documents read on 2026-09-06.); KEEL (direct cost per bitcoin: Not disclosed by the company in the documents read on 2026-09-06.); DMGGF (direct cost per bitcoin: Not disclosed by the company in the documents read on 2026-09-06.); ARBK (direct cost per bitcoin: Not disclosed by the company in the documents read on 2026-09-06.); VIP (direct cost per bitcoin: Not disclosed by the company in the documents read on 2026-09-06.); DGXX (direct cost per bitcoin: no research pack carries this company); ANY (direct cost per bitcoin: no research pack carries this company); BGDE (direct cost per bitcoin: no research pack carries this company).

Direct cost against J/TH

n 6 of 18Formulas
Why this pair
Fewer joules per terahash should mean less power bought per bitcoin mined.

Direct cost per bitcoin against fleet efficiency in J/TH, 6 miners

USD per BTC against J/TH · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1

Reading: The least-squares line accounts for 37 percent of the spread (R squared 0.37), with p 0.199 for its slope, on 6 points.

Direct cost per bitcoin against fleet efficiency in J/TH, 6 miners. The least-squares line accounts for 37 percent of the spread (R squared 0.37), with p 0.199 for its slope, on 6 points.
CompanyCoverage tierFleet efficiency (J/TH)Direct cost per bitcoin (USD per BTC)
CLSKTier 115.8$44,406
HUTTier 114.1$36,500
IRENTier 115.0$26,259
MARATier 118.6$38,690
RIOTTier 120.2$49,912
WULFTier 117.5$53,681
Least-squares lineslope 2615.106992, intercept -2533.471262, fitted in levels

6 points: the test is computed but has little power, and one company can move it.

The statistics of direct cost per bitcoin on fleet efficiency in J/TH
n6
degrees of freedom4
slope2.62K95% interval -2.1K to 7.33K
intercept-2.53K
SST493Mtotal sum of squares
SSR183Mexplained by the fit
SSE310Mleft over
R squared0.37SSR over SST
t, slope1.54standard error 1.7K
p, slope0.199two-sided
Pearson r0.61p 0.199
Spearman rho0.66p 0.156
x mean16.9sd 2.31, variance 5.36, median 16.7
y mean41.6Ksd 9.93K, variance 98.5M, median 41.5K

computed 2026-09-22 from data as of 2026-09-06, written to data/analysis/cross_section.json; formulas

Each company's two figures, with period, as-of and source (6 of 18)
Each company's fleet efficiency and direct cost per bitcoin, with period, as-of and source
CLSK15.8Q3 FY2026, as of 2026-06-30, source$44,406Q3 FY2026, as of 2026-06-30, source
HUT14.1Q2 2026, as of 2026-06-30, source$36,500Q2 2026, as of 2026-06-30, source
IREN15.0June 2025, as of 2025-06-30, source$26,259June 2025, as of 2025-06-30, source
MARA18.6FY2025, as of 2025-12-31, source$38,690Q2 2026, as of 2026-06-30, source
RIOT20.2Q1 2026, as of 2026-03-31, source$49,912Q2 2026, as of 2026-06-30, source
WULF17.5FY2025 year-end, as of 2025-12-31, source$53,681FY2025, as of 2025-12-31, source

Not in the sample: CORZ (fleet efficiency in J/TH: not disclosed in the FY2025 10-K or 2026 10-Qs); CIFR (direct cost per bitcoin: Not disclosed. Derived only: Q2 2026 cost of revenue $15.0M (primarily Odessa power) / ~346 BTC = ~$43,500 per BTC before depreciation and SG&A (derived here, not a company figure).); BTDR (direct cost per bitcoin: Not disclosed. Derived electricity-only cost = ($84.7M + $13.3M) / 2,694 BTC = ~$36,400 per BTC (derived here, excludes co-mining hosting fees in 'other costs').); BTBT (fleet efficiency in J/TH: Not disclosed by the company in the documents read on 2026-09-06.); HIVE (fleet efficiency in J/TH: status reported-not-comparable: Efficiency implied from the nameplate rather than fleet-weighted from production: the filing says 'an implied efficiency of 16.5 J/TH, based on the nameplat[e]'. The company's words: "total installed hashrate was approximately 25.2 EH/s with an implied efficiency of 16.5 J/TH, based on the nameplat[e]"); KEEL (fleet efficiency in J/TH: Not disclosed by the company in the documents read on 2026-09-06.); DMGGF (fleet efficiency in J/TH: Not disclosed by the company in the documents read on 2026-09-06.); ARBK (fleet efficiency in J/TH: Not disclosed by the company in the documents read on 2026-09-06.); VIP (fleet efficiency in J/TH: Not disclosed by the company in the documents read on 2026-09-06.); DGXX (fleet efficiency in J/TH: no research pack carries this company); ANY (fleet efficiency in J/TH: no research pack carries this company); BGDE (fleet efficiency in J/TH: no research pack carries this company).

BTC held against hashrate

n 8 of 18Formulas
Why this pair
Whether the miners that run the most hashrate also hold the most bitcoin.

Bitcoin held against realized hashrate, 8 miners

BTC against EH/s · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1

Reading: The least-squares line accounts for 22 percent of the spread (R squared 0.22), with p 0.243 for its slope, on 8 points.

Bitcoin held against realized hashrate, 8 miners. The least-squares line accounts for 22 percent of the spread (R squared 0.22), with p 0.243 for its slope, on 8 points.
CompanyCoverage tierHashrate, realized (EH/s)Bitcoin held (BTC)
BTDRTier 176.7257
CIFRTier 111.6646
CLSKTier 138.613,931
CORZTier 115.7848
HUTTier 128.117,316
IRENTier 123.20
MARATier 170.335,577
RIOTTier 144.411,380
Least-squares lineslope 241.241049, intercept 688.501541, fitted in levels
The statistics of bitcoin held on realized hashrate
n8
degrees of freedom6
slope24195% interval -215 to 697
intercept689
SST1.09Btotal sum of squares
SSR238Mexplained by the fit
SSE853Mleft over
R squared0.22SSR over SST
t, slope1.29standard error 186
p, slope0.243two-sided
Pearson r0.47p 0.243
Spearman rho0.29p 0.493
x mean38.6sd 24.2, variance 584, median 33.4
y mean9.99Ksd 12.5K, variance 156M, median 6.11K

computed 2026-09-22 from data as of 2026-09-06, written to data/analysis/cross_section.json; formulas

Each company's two figures, with period, as-of and source (8 of 18)
Each company's hashrate, realized and bitcoin held, with period, as-of and source
BTDR76.7July 2026, as of 2026-07-31, source257July 2026, as of 2026-07-31, source
CIFR11.6Q2 2026, as of 2026-08-04, source646Q2 2026, as of 2026-06-30, source
CLSK38.6July 2026, as of 2026-07-31, source13,931July 2026, as of 2026-07-31, source
CORZ15.7FY2025, as of 2025-12-31, source848Q2 2026, as of 2026-06-30, source
HUT28.1Q2 2026, as of 2026-06-30, source17,316Q2 2026, as of 2026-06-30, source
IREN23.2FY26 year-end, as of 2026-06-30, source0FY26 year-end, as of 2026-06-30, source
MARA70.3Q2 2026, as of 2026-06-30, source35,577Q2 2026, as of 2026-06-30, source
RIOT44.4Q2 2026, as of 2026-06-30, source11,380Q2 2026, as of 2026-06-30, source

Not in the sample: WULF (bitcoin held: BTC count not disclosed; digital assets carried at $133 thousand fair value at 2026-06-30, so holdings are de minimis. 10-Q Note 1: 'Bitcoin earned as consideration is routinely converted to U.S. dollars.'); BTBT (bitcoin held: Not disclosed by the company in the documents read on 2026-09-06.); HIVE (realized hashrate: status reported-not-comparable: This is installed hashrate, not realized: the filing says the total installed hashrate was approximately 25.2 EH/s, and the same quarter's release gives 24.0 EH/s as the quarter average. The Tier 1 companies report self-mining hashrate actually produced. The company's words: "As of July 31, 2026, the Company's total installed hashrate was approximately 25.2 EH/s with an implied efficiency of 16.5 J/TH". The Q1 FY2027 release gives 24.0 EH/s as the quarter average.); KEEL (realized hashrate: Not disclosed by the company in the documents read on 2026-09-06.); DMGGF (realized hashrate: Not disclosed by the company in the documents read on 2026-09-06.); ARBK (realized hashrate: status reported-not-comparable: Fleet capability, not realized hashrate: the 20-F says the fleet 'can generate up to 2.4 exahash per second'. It is also the oldest hero figure on the site, from the fiscal 2025 balance date. The company's words: "As of December 31, 2025, we had a fleet of approximately 22,600 machines mining Bitcoin that can generate up to 2.4 exahash per second of computing power."); VIP (realized hashrate: Not disclosed by the company in the documents read on 2026-09-06.); DGXX (realized hashrate: no research pack carries this company); ANY (realized hashrate: no research pack carries this company); BGDE (realized hashrate: no research pack carries this company).

J/TH against revenue per MW

n 8 of 18Formulas
Why this pair
Whether a more efficient fleet earns more from each megawatt.

Fleet efficiency in J/TH against annualized revenue per MW, 8 miners

J/TH against USD per MW per year · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1

Reading: The least-squares line accounts for 2 percent of the spread (R squared 0.02), with p 0.760 for its slope, on 8 points.

Fleet efficiency in J/TH against annualized revenue per MW, 8 miners. The least-squares line accounts for 2 percent of the spread (R squared 0.02), with p 0.760 for its slope, on 8 points.
CompanyCoverage tierRevenue per MW, annualized (USD per MW per year)Fleet efficiency (J/TH)
BTDRTier 1$522K15.8
CIFRTier 1$480K17.2
CLSKTier 1$683K15.8
HUTTier 1$422K14.1
IRENTier 1$678K15.0
MARATier 1$532K18.6
RIOTTier 1$539K20.2
WULFTier 1$1.76M17.5
Least-squares lineslope 5.94063e-7, intercept 16.358256, fitted in levels
The statistics of fleet efficiency in J/TH on annualized revenue per MW
n8
degrees of freedom6
slope0.00000059495% interval -0.00000396 to 0.00000514
intercept16.4
SST28total sum of squares
SSR0.468explained by the fit
SSE27.5left over
R squared0.02SSR over SST
t, slope0.32standard error 0.00000186
p, slope0.760two-sided
Pearson r0.13p 0.760
Spearman rho0.29p 0.490
x mean702Ksd 435K, variance 189B, median 536K
y mean16.8sd 2, variance 4, median 16.5

computed 2026-09-22 from data as of 2026-09-06, written to data/analysis/cross_section.json; formulas

Each company's two figures, with period, as-of and source (8 of 18)
Each company's revenue per mw, annualized and fleet efficiency, with period, as-of and source
BTDR$522K (computed)Q2 2026, as of 2026-06-30, source15.8Q2 2026, as of 2026-06-30, source
CIFR$480K (computed)Q2 2026, as of 2026-06-30, source17.2Q2 2026, as of 2026-08-04, source
CLSK$683K (computed)Q3 FY2026, as of 2026-06-30, source15.8Q3 FY2026, as of 2026-06-30, source
HUT$422K (computed)Q2 2026, as of 2026-06-30, source14.1Q2 2026, as of 2026-06-30, source
IREN$678K (computed)Q4 FY26 (quarter ended June 30, 2026), as of 2026-06-30, source15.0June 2025, as of 2025-06-30, source
MARA$532K (computed)Q2 2026, as of 2026-06-30, source18.6FY2025, as of 2025-12-31, source
RIOT$539K (computed)Q2 2026, as of 2026-06-30, source20.2Q1 2026, as of 2026-03-31, source
WULF$1.76M (computed)Q2 2026, as of 2026-06-30, source17.5FY2025 year-end, as of 2025-12-31, source

Not in the sample: CORZ (fleet efficiency in J/TH: not disclosed in the FY2025 10-K or 2026 10-Qs); BTBT (annualized revenue per MW: Revenue per MW is annualized quarterly revenue over energized megawatts, and this company discloses no energized megawatts: it states 12.55 MW of nameplate or contracted capacity instead (hosted capacity at Merkle Standard, as of 2026-06-30), and the peers' denominator is energized capacity, so computing on nameplate would change what the metric means while keeping its label.); HIVE (fleet efficiency in J/TH: status reported-not-comparable: Efficiency implied from the nameplate rather than fleet-weighted from production: the filing says 'an implied efficiency of 16.5 J/TH, based on the nameplat[e]'. The company's words: "total installed hashrate was approximately 25.2 EH/s with an implied efficiency of 16.5 J/TH, based on the nameplat[e]"); KEEL (annualized revenue per MW: Revenue per MW is annualized quarterly revenue over energized megawatts, and this company discloses no energized megawatts.); DMGGF (annualized revenue per MW: Revenue per MW is annualized quarterly revenue over energized megawatts, and this company discloses no quarterly revenue, and no energized megawatts: it states 75 MW of nameplate or contracted capacity instead (as of 2026-03-06), and the peers' denominator is energized capacity, so computing on nameplate would change what the metric means while keeping its label.); ARBK (annualized revenue per MW: Revenue per MW is annualized quarterly revenue over energized megawatts, and this company discloses no energized megawatts.); VIP (fleet efficiency in J/TH: Not disclosed by the company in the documents read on 2026-09-06.); DGXX (annualized revenue per MW: no research pack carries this company); ANY (annualized revenue per MW: no research pack carries this company); BGDE (annualized revenue per MW: no research pack carries this company).

XBRL count against BTC held

n 5 of 18Formulas
Why this pair
The research pass's bitcoin count against the XBRL tag for the same thing: a check on the two sources, not a finding about miners.

The XBRL bitcoin count against bitcoin held, 5 miners

units against BTC · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1Tier 2

Reading: The least-squares line accounts for 53 percent of the spread (R squared 0.53), with p 0.161 for its slope, on 5 points.

The XBRL bitcoin count against bitcoin held, 5 miners. The least-squares line accounts for 53 percent of the spread (R squared 0.53), with p 0.161 for its slope, on 5 points.
CompanyCoverage tierBitcoin held (BTC)Bitcoin held, as tagged in XBRL (units)
CLSKTier 113,9311,641
HUTTier 117,31617,316
IRENTier 100
KEELTier 21,8612,261
RIOTTier 111,38011,380
Least-squares lineslope 0.720079, intercept 112.623711, fitted in levels

5 points: the test is computed but has little power, and one company can move it.

The statistics of the XBRL bitcoin count on bitcoin held
n5
degrees of freedom3
slope0.7295% interval -0.518 to 1.96
intercept113
SST225Mtotal sum of squares
SSR120Mexplained by the fit
SSE105Mleft over
R squared0.53SSR over SST
t, slope1.85standard error 0.389
p, slope0.161two-sided
Pearson r0.73p 0.161
Spearman rho0.70p 0.188
x mean8.9Ksd 7.6K, variance 57.8M, median 11.4K
y mean6.52Ksd 7.49K, variance 56.2M, median 2.26K

computed 2026-09-22 from data as of 2026-09-06, written to data/analysis/cross_section.json; formulas

Each company's two figures, with period, as-of and source (5 of 18)
Each company's bitcoin held and bitcoin held, as tagged in xbrl, with period, as-of and source
CLSK13,931July 2026, as of 2026-07-31, source1,641at 2026-03-31, as of 2026-03-31, source
HUT17,316Q2 2026, as of 2026-06-30, source17,316at 2026-06-30, as of 2026-06-30, source
IREN0FY26 year-end, as of 2026-06-30, source0at 2026-06-30, as of 2026-06-30, source
KEEL1,861period not stated, as of 2026-08-07, source2,261at 2026-06-30, as of 2026-06-30, source
RIOT11,380Q2 2026, as of 2026-06-30, source11,380at 2026-06-30, as of 2026-06-30, source

Not in the sample: MARA (the XBRL bitcoin count: the company does not tag this concept for a quarter, half-year or fiscal year); CORZ (the XBRL bitcoin count: the company does not tag this concept for a quarter, half-year or fiscal year); WULF (bitcoin held: BTC count not disclosed; digital assets carried at $133 thousand fair value at 2026-06-30, so holdings are de minimis. 10-Q Note 1: 'Bitcoin earned as consideration is routinely converted to U.S. dollars.'); CIFR (the XBRL bitcoin count: the company does not tag this concept for a quarter, half-year or fiscal year); BTDR (the XBRL bitcoin count: the company does not tag this concept for a quarter, half-year or fiscal year); BTBT (bitcoin held: Not disclosed by the company in the documents read on 2026-09-06.); HIVE (bitcoin held: Not disclosed by the company in the documents read on 2026-09-06.); DMGGF (bitcoin held: Not disclosed by the company in the documents read on 2026-09-06.); ARBK (bitcoin held: Not disclosed by the company in the documents read on 2026-09-06.); VIP (bitcoin held: Not disclosed by the company in the documents read on 2026-09-06.); DGXX (bitcoin held: no research pack carries this company); ANY (bitcoin held: no research pack carries this company); BGDE (bitcoin held: no research pack carries this company).

Distributions

11 figures across the 18 minersFormulas
How to read it
Every figure the pairs draw on, one point a company, largest first, with its sample mean, standard deviation, variance and quartiles, and the companies not on file named with the reason.

Revenue, annualized from the latest reported period, 16 of 18 miners, largest first

USD per year · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1Tier 2Tier 3hollow: computed here

Reading: CLSK is highest, at $766.3M per year, and ANY lowest, at $9.8M per year; the median of the 16 is $239.4M per year.

Revenue, annualized from the latest reported period, 16 of 18 miners, largest first
CompanyCoverage tierRevenue, annualized from the latest reported period (USD per year)Status
CLSKTier 1$766.3Mreported
IRENTier 1$707Mreported
MARATier 1$699.5Mderived
RIOTTier 1$696.9Mderived
CORZTier 1$656.8Mderived
BTDRTier 1$451.4Mderived
HIVETier 2$316.5Mderived
HUTTier 1$299.7Mderived
WULFTier 1$179.1Mderived
BTBTTier 2$128.5Mderived
KEELTier 2$121.7Mderived
CIFRTier 1$99.3Mderived
DGXXTier 3$26.5Mderived
BGDETier 3$24.6Mderived
VIPTier 2$13.5Mderived
ANYTier 3$9.8Mderived
Sample statistics of Revenue, annualized from the latest reported period
nmeansdvariancemedianq1q3minmax
16 of 18325M291M84,900T239M81.1M667M9.81M766M
Each input with period, as-of and source; 2 not on file
Each company's revenue, annualized from the latest reported period with its status, period, as-of and source
CLSK$766.3Mreportedyear ended 2025-09-302025-09-30filing
IREN$707Mreportedyear ended 2026-06-302026-06-30filing
MARA$699.5Mderivedquarter ended 2026-06-30, times 42026-06-30filing
RIOT$696.9Mderivedquarter ended 2026-06-30, times 42026-06-30filing
CORZ$656.8Mderivedquarter ended 2026-06-30, times 42026-06-30filing
BTDR$451.4Mderivedhalf-year ended 2025-06-30, times 22025-06-30filing
HIVE$316.5Mderivedquarter ended 2026-06-30, times 42026-06-30filing
HUT$299.7Mderivedquarter ended 2026-06-30, times 42026-06-30filing
WULF$179.1Mderivedquarter ended 2026-06-30, times 42026-06-30filing
BTBT$128.5Mderivedquarter ended 2026-06-30, times 42026-06-30filing
KEEL$121.7Mderivedquarter ended 2026-06-30, times 42026-06-30filing
CIFR$99.3Mderivedquarter ended 2026-06-30, times 42026-06-30filing
DGXX$26.5Mderivedquarter ended 2026-06-30, times 42026-06-30filing
BGDE$24.6Mderivedquarter ended 2026-06-30, times 42026-06-30filing
VIP$13.5Mderivedquarter ended 2026-06-30, times 42026-06-30filing
ANY$9.8Mderivedquarter ended 2026-06-30, times 42026-06-30filing

Not on file: DMGGF (files on SEDAR+ only; no SEC XBRL facts); ARBK (the company does not tag this concept for a quarter, half-year or fiscal year).

Revenue per MW, annualized, 11 of 18 miners, largest first

USD per MW per year · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1Tier 2hollow: computed here

Reading: WULF is highest, at $1.76M per MW per year, and VIP lowest, at $130K per MW per year; the median of the 11 is $539K per MW per year.

Revenue per MW, annualized, 11 of 18 miners, largest first
CompanyCoverage tierRevenue per MW, annualized (USD per MW per year)Status
WULFTier 1$1.76Mcomputed
CORZTier 1$1.66Mcomputed
HIVETier 2$719Kcomputed
CLSKTier 1$683Kcomputed
IRENTier 1$678Kcomputed
RIOTTier 1$539Kcomputed
MARATier 1$532Kcomputed
BTDRTier 1$522Kcomputed
CIFRTier 1$480Kcomputed
HUTTier 1$422Kcomputed
VIPTier 2$130Kcomputed
Sample statistics of Revenue per MW, annualized
nmeansdvariancemedianq1q3minmax
11 of 18739K506K256B539K501K701K130K1.76M
Each input with period, as-of and source; 7 not on file
Each company's revenue per mw, annualized with its status, period, as-of and source
WULF$1.76McomputedQ2 20262026-06-30source
CORZ$1.66McomputedQ2 20262026-06-30source
HIVE$719Kcomputed10-Q ending 2026-06-302026-06-30source
CLSK$683KcomputedQ3 FY20262026-06-30source
IREN$678KcomputedQ4 FY26 (quarter ended June 30, 2026)2026-06-30source
RIOT$539KcomputedQ2 20262026-06-30source
MARA$532KcomputedQ2 20262026-06-30source
BTDR$522KcomputedQ2 20262026-06-30source
CIFR$480KcomputedQ2 20262026-06-30source
HUT$422KcomputedQ2 20262026-06-30source
VIP$130Kcomputed10-Q ending 2026-06-302026-06-30source

Not on file: BTBT (Revenue per MW is annualized quarterly revenue over energized megawatts, and this company discloses no energized megawatts: it states 12.55 MW of nameplate or contracted capacity instead (hosted capacity at Merkle Standard, as of 2026-06-30), and the peers' denominator is energized capacity, so computing on nameplate would change what the metric means while keeping its label.); KEEL (Revenue per MW is annualized quarterly revenue over energized megawatts, and this company discloses no energized megawatts.); DMGGF (Revenue per MW is annualized quarterly revenue over energized megawatts, and this company discloses no quarterly revenue, and no energized megawatts: it states 75 MW of nameplate or contracted capacity instead (as of 2026-03-06), and the peers' denominator is energized capacity, so computing on nameplate would change what the metric means while keeping its label.); ARBK (Revenue per MW is annualized quarterly revenue over energized megawatts, and this company discloses no energized megawatts.); DGXX (no research pack carries this company); ANY (no research pack carries this company); BGDE (no research pack carries this company).

Gross margin, 10 of 18 miners, largest first

percent of revenue · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1Tier 2Tier 3hollow: computed here

Reading: WULF is highest, at +72.3%, and KEEL lowest, at -285.1%; the median of the 10 is +41.0%.

Gross margin, 10 of 18 miners, largest first
CompanyCoverage tierGross margin (percent of revenue)Status
WULFTier 1+72.3%derived
IRENTier 1+68.9%derived
HUTTier 1+64.1%derived
CLSKTier 1+55.2%derived
CORZTier 1+42.7%derived
CIFRTier 1+39.4%derived
BGDETier 3+26.1%derived
ANYTier 3+6.7%derived
BTDRTier 1+4.3%derived
KEELTier 2-285.1%derived
Sample statistics of Gross margin
nmeansdvariancemedianq1q3minmax
10 of 189.4710611.3K4111.661.9-28572.3
Each input with period, as-of and source; 8 not on file
Each company's gross margin with its status, period, as-of and source
WULF+72.3%derivedquarter ended 2026-06-302026-06-30filing
IREN+68.9%derivedyear ended 2026-06-302026-06-30filing
HUT+64.1%derivedquarter ended 2026-06-302026-06-30filing
CLSK+55.2%derivedyear ended 2025-09-302025-09-30filing
CORZ+42.7%derivedquarter ended 2026-06-302026-06-30filing
CIFR+39.4%derivedquarter ended 2026-06-302026-06-30filing
BGDE+26.1%derivedquarter ended 2026-06-302026-06-30filing
ANY+6.7%derivedquarter ended 2026-06-302026-06-30filing
BTDR+4.3%derivedhalf-year ended 2025-06-302025-06-30filing
KEEL-285.1%derivedquarter ended 2026-06-302026-06-30filing

Not on file: MARA (no cost of revenue tagged for the quarter ended 2026-06-30); RIOT (no cost of revenue tagged for the quarter ended 2026-06-30); BTBT (no cost of revenue tagged for the quarter ended 2026-06-30); HIVE (no cost of revenue tagged for the quarter ended 2026-06-30); DMGGF (no SEC XBRL facts on file); ARBK (no revenue fact for a quarter, half-year or fiscal year); VIP (no cost of revenue tagged for the quarter ended 2026-06-30); DGXX (no cost of revenue tagged for the quarter ended 2026-06-30).

Hashrate, realized, 10 of 18 miners, largest first

EH/s · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1Tier 2

Reading: BTDR is highest, at 76.7 EH/s, and BTBT lowest, at 0.4 EH/s; the median of the 10 is 25.7 EH/s.

Hashrate, realized, 10 of 18 miners, largest first
CompanyCoverage tierHashrate, realized (EH/s)Status
BTDRTier 176.7reported
MARATier 170.3reported
RIOTTier 144.4reported
CLSKTier 138.6reported
HUTTier 128.1reported
IRENTier 123.2reported
CORZTier 115.7reported
CIFRTier 111.6reported
WULFTier 19.3reported
BTBTTier 20.4reported
Sample statistics of Hashrate, realized
nmeansdvariancemedianq1q3minmax
10 of 1831.825.766125.712.6430.476.7
Each input with period, as-of and source; 8 not on file
Each company's hashrate, realized with its status, period, as-of and source
BTDR76.7reportedJuly 20262026-07-31source
MARA70.3reportedQ2 20262026-06-30source
RIOT44.4reportedQ2 20262026-06-30source
CLSK38.6reportedJuly 20262026-07-31source
HUT28.1reportedQ2 20262026-06-30source
IREN23.2reportedFY26 year-end2026-06-30source
CORZ15.7reportedFY20252025-12-31source
CIFR11.6reportedQ2 20262026-08-04source
WULF9.3reportedFY2025 year-end2025-12-31source
BTBT0.4reportednot stated2026-06-30source

Not on file: HIVE (status reported-not-comparable: This is installed hashrate, not realized: the filing says the total installed hashrate was approximately 25.2 EH/s, and the same quarter's release gives 24.0 EH/s as the quarter average. The Tier 1 companies report self-mining hashrate actually produced. The company's words: "As of July 31, 2026, the Company's total installed hashrate was approximately 25.2 EH/s with an implied efficiency of 16.5 J/TH". The Q1 FY2027 release gives 24.0 EH/s as the quarter average.); KEEL (Not disclosed by the company in the documents read on 2026-09-06.); DMGGF (Not disclosed by the company in the documents read on 2026-09-06.); ARBK (status reported-not-comparable: Fleet capability, not realized hashrate: the 20-F says the fleet 'can generate up to 2.4 exahash per second'. It is also the oldest hero figure on the site, from the fiscal 2025 balance date. The company's words: "As of December 31, 2025, we had a fleet of approximately 22,600 machines mining Bitcoin that can generate up to 2.4 exahash per second of computing power."); VIP (Not disclosed by the company in the documents read on 2026-09-06.); DGXX (no research pack carries this company); ANY (no research pack carries this company); BGDE (no research pack carries this company).

Bitcoin held, 9 of 18 miners, largest first

BTC · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1Tier 2

Reading: MARA is highest, at 35,577 BTC, and IREN lowest, at 0 BTC; the median of the 9 is 1,861 BTC.

Bitcoin held, 9 of 18 miners, largest first
CompanyCoverage tierBitcoin held (BTC)Status
MARATier 135,577reported
HUTTier 117,316reported
CLSKTier 113,931reported
RIOTTier 111,380reported
KEELTier 21,861reported
CORZTier 1848reported
CIFRTier 1646reported
BTDRTier 1257reported
IRENTier 10reported
Sample statistics of Bitcoin held
nmeansdvariancemedianq1q3minmax
9 of 189.09K12K144M1.86K64613.9K035.6K
Each input with period, as-of and source; 9 not on file
Each company's bitcoin held with its status, period, as-of and source
MARA35,577reportedQ2 20262026-06-30source
HUT17,316reportedQ2 20262026-06-30source
CLSK13,931reportedJuly 20262026-07-31source
RIOT11,380reportedQ2 20262026-06-30source
KEEL1,861reportednot stated2026-08-07source
CORZ848reportedQ2 20262026-06-30source
CIFR646reportedQ2 20262026-06-30source
BTDR257reportedJuly 20262026-07-31source
IREN0reportedFY26 year-end2026-06-30source

Not on file: WULF (BTC count not disclosed; digital assets carried at $133 thousand fair value at 2026-06-30, so holdings are de minimis. 10-Q Note 1: 'Bitcoin earned as consideration is routinely converted to U.S. dollars.'); BTBT (Not disclosed by the company in the documents read on 2026-09-06.); HIVE (Not disclosed by the company in the documents read on 2026-09-06.); DMGGF (Not disclosed by the company in the documents read on 2026-09-06.); ARBK (Not disclosed by the company in the documents read on 2026-09-06.); VIP (Not disclosed by the company in the documents read on 2026-09-06.); DGXX (no research pack carries this company); ANY (no research pack carries this company); BGDE (no research pack carries this company).

Market capitalization, 9 of 18 miners, largest first

USD · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1hollow: computed here

Reading: IREN is highest, at $17.61B, and CLSK lowest, at $3.26B; the median of the 9 is $7.36B.

Market capitalization, 9 of 18 miners, largest first
CompanyCoverage tierMarket capitalization (USD)Status
IRENTier 1$17.61Bcomputed
HUTTier 1$11.53Bcomputed
WULFTier 1$8.24Bcomputed
RIOTTier 1$8.18Bcomputed
CIFRTier 1$7.36Bcomputed
CORZTier 1$5.75Bcomputed
MARATier 1$4.37Bcomputed
BTDRTier 1$3.36Bcomputed
CLSKTier 1$3.26Bcomputed
Sample statistics of Market capitalization
nmeansdvariancemedianq1q3minmax
9 of 187.74B4.57B20,900,000T7.36B4.37B8.24B3.26B17.6B
Each input with period, as-of and source; 9 not on file
Each company's market capitalization with its status, period, as-of and source
IREN$17.61Bcomputedspot2026-09-04source
HUT$11.53Bcomputedspot2026-09-04source
WULF$8.24Bcomputedspot2026-09-04source
RIOT$8.18Bcomputedspot2026-09-04source
CIFR$7.36Bcomputedspot2026-09-04source
CORZ$5.75Bcomputedspot2026-09-04source
MARA$4.37Bcomputedspot2026-09-04source
BTDR$3.36Bcomputedspot2026-09-04source
CLSK$3.26Bcomputedspot2026-09-04source

Not on file: BTBT (This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $83.6M, debt $336.2M, and 360.6M shares are on file; the price is the term that is missing.); HIVE (This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $208M, debt $234.9M, and 274.3M shares are on file; the price is the term that is missing.); KEEL (This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $768.9M, debt $1.03B, and 617.6M shares are on file; the price is the term that is missing.); DMGGF (This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. None of cash, debt or the share count is on file; the price is the term that is missing. DMG files nothing with the SEC and trades over the counter, so its figures are graded as company releases.); ARBK (This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $2.2M, debt $40.2M, and 28.9B shares are on file; the price is the term that is missing. Argo trades as ADSs worth 2,160 ordinary shares each, so an ADS price multiplied by the ordinary share count would overstate market capitalization by that ratio; the figure is left null rather than computed on a mismatched pair.); VIP (This pass carries no share price for a Tier 2 company, so market capitalization cannot be formed. Cash $3.2M, debt $39.5M, and 18M shares are on file; the price is the term that is missing.); DGXX (no research pack carries this company); ANY (no research pack carries this company); BGDE (no research pack carries this company).

Bitcoin held, as tagged in XBRL, 8 of 18 miners, largest first

units · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1Tier 2Tier 3

Reading: HUT is highest, at 17,316 units, and IREN lowest, at 0 units; the median of the 8 is 1,951 units.

Bitcoin held, as tagged in XBRL, 8 of 18 miners, largest first
CompanyCoverage tierBitcoin held, as tagged in XBRL (units)Status
HUTTier 117,316reported
RIOTTier 111,380reported
BTBTTier 210,221reported
KEELTier 22,261reported
CLSKTier 11,641reported
DGXXTier 39reported
WULFTier 12reported
IRENTier 10reported
Sample statistics of Bitcoin held, as tagged in XBRL
nmeansdvariancemedianq1q3minmax
8 of 185.35K6.68K44.6M1.95K7.2510.5K017.3K
Each input with period, as-of and source; 10 not on file
Each company's bitcoin held, as tagged in xbrl with its status, period, as-of and source
HUT17,316reportedat 2026-06-302026-06-30filing
RIOT11,380reportedat 2026-06-302026-06-30filing
BTBT10,221reportedat 2025-09-302025-09-30filing
KEEL2,261reportedat 2026-06-302026-06-30filing
CLSK1,641reportedat 2026-03-312026-03-31filing
DGXX9reportedat 2025-12-312025-12-31filing
WULF2reportedat 2026-06-302026-06-30filing
IREN0reportedat 2026-06-302026-06-30filing

Not on file: MARA (the company does not tag this concept for a quarter, half-year or fiscal year); CORZ (the company does not tag this concept for a quarter, half-year or fiscal year); CIFR (the company does not tag this concept for a quarter, half-year or fiscal year); BTDR (the company does not tag this concept for a quarter, half-year or fiscal year); HIVE (the company does not tag this concept for a quarter, half-year or fiscal year); DMGGF (files on SEDAR+ only; no SEC XBRL facts); ARBK (the company does not tag this concept for a quarter, half-year or fiscal year); VIP (the company does not tag this concept for a quarter, half-year or fiscal year); ANY (the company does not tag this concept for a quarter, half-year or fiscal year); BGDE (the company does not tag this concept for a quarter, half-year or fiscal year).

Fleet efficiency, 8 of 18 miners, largest first

J/TH · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1

Reading: RIOT is highest, at 20.2 J/TH, and HUT lowest, at 14.1 J/TH; the median of the 8 is 16.5 J/TH.

Fleet efficiency, 8 of 18 miners, largest first
CompanyCoverage tierFleet efficiency (J/TH)Status
RIOTTier 120.2reported
MARATier 118.6reported
WULFTier 117.5reported
CIFRTier 117.2reported
BTDRTier 115.8reported
CLSKTier 115.8reported
IRENTier 115.0reported
HUTTier 114.1reported
Sample statistics of Fleet efficiency
nmeansdvariancemedianq1q3minmax
8 of 1816.82416.515.617.814.120.2
Each input with period, as-of and source; 10 not on file
Each company's fleet efficiency with its status, period, as-of and source
RIOT20.2reportedQ1 20262026-03-31source
MARA18.6reportedFY20252025-12-31source
WULF17.5reportedFY2025 year-end2025-12-31source
CIFR17.2reportedQ2 20262026-08-04source
BTDR15.8reportedQ2 20262026-06-30source
CLSK15.8reportedQ3 FY20262026-06-30source
IREN15.0reportedJune 20252025-06-30source
HUT14.1reportedQ2 20262026-06-30source

Not on file: CORZ (not disclosed in the FY2025 10-K or 2026 10-Qs); BTBT (Not disclosed by the company in the documents read on 2026-09-06.); HIVE (status reported-not-comparable: Efficiency implied from the nameplate rather than fleet-weighted from production: the filing says 'an implied efficiency of 16.5 J/TH, based on the nameplat[e]'. The company's words: "total installed hashrate was approximately 25.2 EH/s with an implied efficiency of 16.5 J/TH, based on the nameplat[e]"); KEEL (Not disclosed by the company in the documents read on 2026-09-06.); DMGGF (Not disclosed by the company in the documents read on 2026-09-06.); ARBK (Not disclosed by the company in the documents read on 2026-09-06.); VIP (Not disclosed by the company in the documents read on 2026-09-06.); DGXX (no research pack carries this company); ANY (no research pack carries this company); BGDE (no research pack carries this company).

Direct cost per bitcoin, 6 of 18 miners, largest first

USD per BTC · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1

Reading: WULF is highest, at $53,681 per BTC, and IREN lowest, at $26,259 per BTC; the median of the 6 is $41,548 per BTC.

Direct cost per bitcoin, 6 of 18 miners, largest first
CompanyCoverage tierDirect cost per bitcoin (USD per BTC)Status
WULFTier 1$53,681reported
RIOTTier 1$49,912reported
CLSKTier 1$44,406reported
MARATier 1$38,690reported
HUTTier 1$36,500reported
IRENTier 1$26,259reported
Sample statistics of Direct cost per bitcoin
nmeansdvariancemedianq1q3minmax
6 of 1841.6K9.93K98.5M41.5K37K48.5K26.3K53.7K
Each input with period, as-of and source; 12 not on file
Each company's direct cost per bitcoin with its status, period, as-of and source
WULF$53,681reportedFY20252025-12-31source
RIOT$49,912reportedQ2 20262026-06-30source
CLSK$44,406reportedQ3 FY20262026-06-30source
MARA$38,690reportedQ2 20262026-06-30source
HUT$36,500reportedQ2 20262026-06-30source
IREN$26,259reportedJune 20252025-06-30source

Not on file: CORZ (Not disclosed as a per-coin figure. Derived here for reference only, not a company figure: H1 2026 self-mining power fees 45.1M USD / about 695 BTC = about 64,900 USD per BTC (power only).); CIFR (Not disclosed. Derived only: Q2 2026 cost of revenue $15.0M (primarily Odessa power) / ~346 BTC = ~$43,500 per BTC before depreciation and SG&A (derived here, not a company figure).); BTDR (Not disclosed. Derived electricity-only cost = ($84.7M + $13.3M) / 2,694 BTC = ~$36,400 per BTC (derived here, excludes co-mining hosting fees in 'other costs').); BTBT (Not disclosed by the company in the documents read on 2026-09-06.); HIVE (Not disclosed by the company in the documents read on 2026-09-06.); KEEL (Not disclosed by the company in the documents read on 2026-09-06.); DMGGF (Not disclosed by the company in the documents read on 2026-09-06.); ARBK (Not disclosed by the company in the documents read on 2026-09-06.); VIP (Not disclosed by the company in the documents read on 2026-09-06.); DGXX (no research pack carries this company); ANY (no research pack carries this company); BGDE (no research pack carries this company).

EV per EH/s, 6 of 18 miners, largest first

USD per EH/s · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1hollow: computed here

Reading: CIFR is highest, at $1.1B per EH/s, and BTDR lowest, at $61.9M per EH/s; the median of the 6 is $158.4M per EH/s.

EV per EH/s, 6 of 18 miners, largest first
CompanyCoverage tierEV per EH/s (USD per EH/s)Status
CIFRTier 1$1.1Bcomputed
HUTTier 1$677.4Mcomputed
RIOTTier 1$191.5Mcomputed
CLSKTier 1$125.4Mcomputed
MARATier 1$91Mcomputed
BTDRTier 1$61.9Mcomputed
Sample statistics of EV per EH/s
nmeansdvariancemedianq1q3minmax
6 of 18371M416M173,000T158M99.6M556M61.9M1.08B
Each input with period, as-of and source; 12 not on file
Each company's ev per eh/s with its status, period, as-of and source
CIFR$1.1Bcomputedspot2026-09-04source
HUT$677.4Mcomputedspot2026-09-04source
RIOT$191.5Mcomputedspot2026-09-04source
CLSK$125.4Mcomputedspot2026-09-04source
MARA$91Mcomputedspot2026-09-04source
BTDR$61.9Mcomputedspot2026-09-04source

Not on file: CORZ (status computed-not-comparable: all inputs present; BTC holdings not netted (shown separately); CAVEAT: restricted cash is not netted in EV (spec formula uses cash and equivalents); see the dossier cash note for the restricted balance; Not comparable: HPC/AI revenue is 83% of quarterly revenue, so a hashrate denominator or a mining-cost run-rate does not describe this company); IREN (status computed-not-comparable: all inputs present; BTC holdings not netted (shown separately); CAVEAT: restricted cash is not netted in EV (spec formula uses cash and equivalents); see the dossier cash note for the restricted balance; Not comparable: HPC/AI revenue is 51% of quarterly revenue, so a hashrate denominator or a mining-cost run-rate does not describe this company); WULF (status computed-not-comparable: all inputs present; BTC holdings not netted (shown separately); CAVEAT: restricted cash is not netted in EV (spec formula uses cash and equivalents); see the dossier cash note for the restricted balance; Not comparable: HPC/AI revenue is 71% of quarterly revenue, so a hashrate denominator or a mining-cost run-rate does not describe this company); BTBT (This pass carries no share price for a Tier 2 company, so enterprise value cannot be formed: it is market capitalization plus debt less cash, and market capitalization is the missing term. Cash $83.6M, debt $336.2M, and 360.6M shares are on file.); HIVE (This pass carries no share price for a Tier 2 company, so enterprise value cannot be formed: it is market capitalization plus debt less cash, and market capitalization is the missing term. Cash $208M, debt $234.9M, and 274.3M shares are on file.); KEEL (This pass carries no share price for a Tier 2 company, so enterprise value cannot be formed: it is market capitalization plus debt less cash, and market capitalization is the missing term. Cash $768.9M, debt $1.03B, and 617.6M shares are on file.); DMGGF (This pass carries no share price for a Tier 2 company, so enterprise value cannot be formed: it is market capitalization plus debt less cash, and market capitalization is the missing term. None of cash, debt or the share count is on file. DMG files nothing with the SEC and trades over the counter, so its figures are graded as company releases.); ARBK (This pass carries no share price for a Tier 2 company, so enterprise value cannot be formed: it is market capitalization plus debt less cash, and market capitalization is the missing term. Cash $2.2M, debt $40.2M, and 28.9B shares are on file. Argo trades as ADSs worth 2,160 ordinary shares each, so an ADS price multiplied by the ordinary share count would overstate market capitalization by that ratio; the figure is left null rather than computed on a mismatched pair.); VIP (This pass carries no share price for a Tier 2 company, so enterprise value cannot be formed: it is market capitalization plus debt less cash, and market capitalization is the missing term. Cash $3.2M, debt $39.5M, and 18M shares are on file.); DGXX (no research pack carries this company); ANY (no research pack carries this company); BGDE (no research pack carries this company).

All-in cost per bitcoin, 3 of 18 miners, largest first

USD per BTC · latest figure on file per company · as of 2026-09-06 · data/analysis/cross_section.json

Coverage tier:Tier 1

Reading: CLSK is highest, at $96,277 per BTC, and RIOT lowest, at $90,631 per BTC; the median of the 3 is $95,611 per BTC.

All-in cost per bitcoin, 3 of 18 miners, largest first
CompanyCoverage tierAll-in cost per bitcoin (USD per BTC)Status
CLSKTier 1$96,277reported
WULFTier 1$95,611reported
RIOTTier 1$90,631reported
Sample statistics of All-in cost per bitcoin
nmeansdvariancemedianq1q3minmax
3 of 1894.2K3.09K9.52M95.6K93.1K95.9K90.6K96.3K
Each input with period, as-of and source; 15 not on file
Each company's all-in cost per bitcoin with its status, period, as-of and source
CLSK$96,277reportedQ3 FY20262026-06-30source
WULF$95,611reportedFY20252025-12-31source
RIOT$90,631reportedQ2 20262026-06-30source

Not on file: MARA (not disclosed as a per-BTC all-in figure. The company reports cost per petahash per day ($27.7 in Q2 2026, down 4% from $28.7) instead. Derivable inputs: purchased energy $48.8 million, O&M $26.9 million, hosting and other energy $69.2 million, D&A $174.7 million against 2,422 BTC.); CORZ (Not disclosed. Derived here for reference only, not a company figure: 80.9M USD / about 695 BTC = about 116,400 USD per BTC (segment cost incl. depreciation, excl. SG&A).); IREN (Not disclosed. Derived FY26 cost of revenue per BTC is about $33,400 (see direct cost note); no depreciation-inclusive figure is published.); HUT (Not disclosed by Hut 8 or ABTC (checked Q2 2026 releases and 10-Q).); CIFR (Not disclosed; Company reports no all-in cost metric.); BTDR (Not disclosed. Derived all-in cost of revenue = ($171.0M + $28.8M) / 2,694 BTC = ~$74,200 per BTC including depreciation and SBC in cost of revenue (derived here, not a company figure).); BTBT (Not disclosed by the company in the documents read on 2026-09-06.); HIVE (Not disclosed by the company in the documents read on 2026-09-06.); KEEL (Not disclosed by the company in the documents read on 2026-09-06.); DMGGF (Not disclosed by the company in the documents read on 2026-09-06.); ARBK (Not disclosed by the company in the documents read on 2026-09-06.); VIP (Not disclosed by the company in the documents read on 2026-09-06.); DGXX (no research pack carries this company); ANY (no research pack carries this company); BGDE (no research pack carries this company).