Skip to the content
MinerTerminal
BTDR $8.06CIFR $9.13CLSK $8.75CORZ $14.25HUT $71.58IREN $27.84MARA $7.79RIOT $14.02WULF $12.76
Sign in
Reference

Learn

Articles

Checking numbers: How to check a number before you use it

How to check a number an AI tool gives you

Ask for the document and page behind every figure, prefer an answer of not stated to a guess, and do the arithmetic yourself. Worked through MARA's and Bitdeer's hashrate targets and the results they later filed.

6 min read, published

A brass magnifying glass on a stack of blank paper filings, with a green thread running from the magnified line to one pinned page tab. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Hashrate target and filed result (EH/s)

Two targets, each beside the figure a later filing reported.

TargetFiled result
MARA
75
66.4
Bitdeer
50
55.2
Method and sources

MARA's filed result came in below its target and Bitdeer's came in above it.

Covers
targets stated in 2025, results filed in January and February 2026
Unit
EH/s

Source: the ledger rows, each with both documents, in Sources

Hashrate target and filed result (EH/s)
CompanyTarget (EH/s)Filed result (EH/s)
MARA, MARA7566.4
Bitdeer, Bitdeer5055.2

Missing is not zero: what a gap in the data does to an average

A figure a company does not publish is unknown, not zero and not average. See how each way of treating a gap changes an average and a screen, and how a good table labels it.

5 min read, published

A graphite tray of six sockets where three hold blocks of different heights and three stay empty and outlined in green, with a flat slab set apart rather than used to fill them. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

One average, three ways to treat gaps

Six illustrative miners, three of which publish no figure. As of Oct 2, 2026.

Hatched: computed here
Left out
$50K
As zero
$25K
Filled in
$50K
Method and sources

Counting the three gaps as zero halves the average. Filling them with the average leaves it unchanged and hides that half the figures are missing.

Covers
a hypothetical set of six miners, three publishing 40,000, 50,000 and 60,000 dollars and three publishing nothing
Unit
USD per bitcoin

Source: Illustrative arithmetic, not company figures

One average, three ways to treat gaps
Treatment of the gapsAverage cost per bitcoin (USD per bitcoin)
Left out, Gaps left out$50K (computed)
As zero, Gaps counted as zero$25K (computed)
Filled in, Gaps filled with the average$50K (computed)

Every number needs a date, a source and a boundary

An old figure looks just like a new one, two documents can disagree, and two costs can count different things. Worked through IREN's hashrate a year apart and CleanSpark's two costs per bitcoin.

6 min read, published

Two matching graphite blocks tagged with blank paper labels, one crisp and framed by a thin green wire box, the other dulled under a layer of dust. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

IREN installed hashrate at two dates (EH/s)

The same company and the same measure, a year apart.

June 2025
50
June 2026
23.2
Method and sources

IREN reported 50 EH/s installed at the end of June 2025 and 23.2 EH/s installed at the end of June 2026.

Covers
installed self-mining capacity at the end of June 2025 and at the end of June 2026
Unit
EH/s

Source: IREN 8-K Jul 7, 2025 and 10-K Aug 27, 2026

IREN installed hashrate at two dates (EH/s)
Month endInstalled hashrate (EH/s)
June 2025, June 202550
June 2026, June 202623.2

Analysis basics: How to read a statistical result

How to read a least-squares fit

Work through slope, intercept, R squared, p-value and the difference between a fit on filed levels and one on their logarithms.

5 min read, published

A plotting plane with measured points and a green fitted line with residual marks. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Revenue against realized hashrate

Latest comparable revenue and hashrate figures across the filed sample. As of Oct 14, 2026.

Hatched: computed here
BTBT
$128.5M
BTDR
$451.4M
CIFR
$99.3M
CLSK
$766.3M
CORZ
$656.8M
HUT
$299.7M
IREN
$707M
KEEL
$121.7M
MARA
$699.5M
RIOT
$696.9M
WULF
$179.1M
Method and sources

The fitted line rises with hashrate, but the worked example also tests whether company size is driving the relation.

Covers
latest comparable figures on file across 11 miners
Unit
USD per year

Source: MinerTerminal analysis, computed from the cited source data

Revenue against realized hashrate
CategoryMiners (USD per year)
BTBT, BTBT$128.5M (computed)
BTDR, BTDR$451.4M (computed)
CIFR, CIFR$99.3M (computed)
CLSK, CLSK$766.3M
CORZ, CORZ$656.8M (computed)
HUT, HUT$299.7M (computed)
IREN, IREN$707M
KEEL, KEEL$121.7M (computed)
MARA, MARA$699.5M (computed)
RIOT, RIOT$696.9M (computed)
WULF, WULF$179.1M (computed)

How to read the spread of a set of figures

Build the mean, median, quartiles, range, sample variance and standard deviation from the annualized revenue sample.

6 min read, published

A caliper spanning varied blocks around a green reference bar. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Annualized revenue spread

CleanSpark, MARA and four sample statistics from the same filed set. As of Oct 14, 2026.

Hatched: computed here
CleanSpark
$766.3M
MARA
$699.5M
Sample mean
$324.8M
Sample median
$239.4M
Sample minimum
$9.8M
Sample maximum
$766.3M
Method and sources

The mean and median describe the sample centre, while the minimum and maximum show how wide the filed values spread.

Covers
latest comparable figures on file across 16 miners
Unit
USD per year

Source: MinerTerminal analysis, computed from the cited source data

Annualized revenue spread
FigureAnnualized revenue (USD per year)
CleanSpark, CleanSpark$766.3M
MARA, MARA$699.5M (computed)
Sample mean, Sample mean$324.8M
Sample median, Sample median$239.4M
Sample minimum, Sample minimum$9.8M
Sample maximum, Sample maximum$766.3M

When a test is flagged for an article

Apply the findings log's p-value ceiling and independent-point floor to two current tests, one comparison at a time.

4 min read, published

A gate with two bars to clear: a p-value ceiling bar and an independent-points floor bar; two test tokens, one clears both bars and gets a green flag, one clears only one and stays unflagged. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Two findings against the p-value ceiling

Two current rows beside the p-value ceiling for article candidates. As of Oct 14, 2026.

Hashrate and revenue, n 11
0.063
CleanSpark leverage, n 365
under 0.0001
Candidate ceiling
0.010
Method and sources

Only the CleanSpark leverage row clears both the p-value ceiling and the independent-point floor in this example.

Covers
current findings log
Unit
p-value

Source: MinerTerminal analysis, computed from the cited source data

Two findings against the p-value ceiling
Findingp-value (p-value)
Hashrate and revenue, n 11, Hashrate and revenue, n 110.063
CleanSpark leverage, n 365, CleanSpark leverage, n 365under 0.0001
Candidate ceiling, Candidate ceiling0.010

The rules behind the statistics

Work through the minimum sample, low-power caveat and the denominators used for regression and sample standard deviation.

5 min read, published

A small brass cylinder on a dark stone plinth under two nested glass bell jars, with a thin green ring at its foot. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Sample size against count rules

The fit's points beside the minimum and low-power count rules. As of Oct 14, 2026.

Fit sample
11
Minimum to run
3
Low-power threshold
8
Degrees of freedom
9
Method and sources

The sample clears the minimum needed to run the test. The low-power rule decides whether a single-company caveat appears.

Covers
current analysis rules and sample
Unit
points

Source: MinerTerminal analysis, computed from the cited source data

Sample size against count rules
CountPoints (points)
Fit sample, Fit sample11
Minimum to run, Minimum to run3
Low-power threshold, Low-power threshold8
Degrees of freedom, Degrees of freedom9

Quarterly figures: How to read quarterly comparisons

How to rebase a quarterly series

Put two companies with very different revenue on one comparable scale by dividing each series by its own first quarter.

4 min read, published

Two paper step-charts of very different height, each pinned at its first step to one shared green start line, then drawn at the same height. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Quarterly revenue, rebased

MARA and Riot, indexed with 2019Q1 set to 100. As of Oct 14, 2026.

MARA
75,806
Riot
12,184
Method and sources

Both values use the company's own starting revenue as 100, so they show change from each starting point rather than dollar size.

Covers
2026Q2 against 2019Q1
Unit
index, 2019Q1 = 100

Source: MinerTerminal analysis, computed from the cited source data

Quarterly revenue, rebased
CompanyRebased revenue (index, 2019Q1 = 100)
MARA, MARA75,806
Riot, Riot12,184

How to read a pooled quarter-to-quarter test

Follow one regression through linked quarter-pairs from several companies, including the independence assumption behind its p-value.

5 min read, published

Many short two-point links (quarter pair) from several differently colored chains gathered onto one single fitted line, with a small note that links of one chain touch each other. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Pooled sample counts

The pooled sample count, split into quarter-pairs and companies. As of Oct 14, 2026.

Quarter-pairs
176
Companies
12
Method and sources

Several linked quarter-pairs can come from one company, which is why the worked example treats independence as a caveat.

Covers
quarterly analysis sample
Unit
count

Source: MinerTerminal analysis, computed from the cited source data

Pooled sample counts
Sample measureCount (count)
Quarter-pairs, Quarter-pairs176
Companies, Companies12

Price studies: How to read price studies

How to rebase daily closing prices

Divide every close by the first close so several price series can share one scale without pretending their dollar prices are comparable.

3 min read, published

A fan of daily panes keyed to one central spindle. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

CleanSpark closes, rebased

The first and last closes are shown as a ratio to the first close. As of Oct 21, 2026.

Hatched: computed here
Jul 28, 2026
1
Oct 20, 2026
0.63
Method and sources

The first close becomes one, and the last close becomes its ratio to that starting value.

Covers
Jul 28, 2026 to Oct 20, 2026
Unit
ratio to the first close

Source: MinerTerminal analysis, computed from the cited source data

CleanSpark closes, rebased
CategoryCleanSpark (ratio to the first close)
Jul 28, 2026, Jul 28, 20261 (computed)
Oct 20, 2026, Oct 20, 20260.63 (computed)

What a miner's leverage factor measures

Read the slope of a miner's daily return on bitcoin's return together with its interval, fit, concordance, volatility and market beta.

6 min read, published

An amplification lever with a green pivot and an uncertainty collar. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

CleanSpark bitcoin leverage

The slope and interval measure equity return per unit of bitcoin return across 365 shared sessions. As of Oct 21, 2026.

Lower bound
1.14
Slope
1.3
Upper bound
1.47
Method and sources

The slope summarizes the historical relation, while the interval shows the estimate's uncertainty.

Covers
365 sessions, May 8, 2025 to Oct 20, 2026
Unit
equity return per unit of bitcoin return

Source: MinerTerminal analysis, computed from the cited source data

CleanSpark bitcoin leverage
EstimateLeverage factor (equity return per unit of bitcoin return)
Lower bound, Lower bound1.14
Slope, Slope1.3
Upper bound, Upper bound1.47

Correlation on returns and on price levels

See why daily return correlation and price-level correlation answer different questions even when they use the same pair of stocks.

4 min read, published

Two panels from the same two stocks: left shows jagged day-to-day bars moving in step weakly, right shows two smooth drifting ribbons that run together; one green link between the two panels. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Return and price-level correlation

The same shared price history, measured two different ways. As of Oct 21, 2026.

Daily log returns
0.84
Price levels
0.46
Method and sources

The return and level correlations answer different questions, and either can be larger in a given window.

Covers
365 shared sessions, May 8, 2025 to Oct 20, 2026
Unit
Pearson r

Source: MinerTerminal analysis, computed from the cited source data

Return and price-level correlation
BasisPearson r (Pearson r)
Daily log returns, Daily log returns0.84
Price levels, Price levels0.46

What the indicator study measures

Compare forward returns when a chart signal is on and off, then read the sample, test statistic and overlapping-session caveat.

5 min read, published

A carved channel in a dark slate slab forks in two beside a small brass lamp whose green lens is lit, and a few pebbles rest in each branch. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Low-RSI study sessions

Sessions in the current 21-session study, grouped by low-RSI signal state. As of Oct 21, 2026.

Signal on
1,060
Signal off
23,201
Method and sources

The off group is much larger, and observations overlap because each session can begin its own forward window.

Covers
21 sessions after each signal close
Unit
sessions

Source: MinerTerminal analysis, computed from the cited source data

Low-RSI study sessions
Signal stateSessions (sessions)
Signal on, Signal on1,060
Signal off, Signal off23,201

Cost to mine: What it costs to mine a bitcoin

What pays miners after the last bitcoin subsidy?

The next halving and the final subsidy are different events. Work through fees, network share and the revenue a miner must replace.

6 min read, published

A mining rig whose block-reward stack at the left is half the height it was, shown as a ghost outline of the old tall stack, with fee slips piling up on the right to fill the gap. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Mining revenue at three subsidy levels

An illustrative day, with the same network share and bitcoin price. As of Sep 23, 2026.

Hatched: computed here
3.125 BTC subsidy
$464.4K
1.5625 BTC subsidy
$239.4K
Fees only
$14.4K
Method and sources

With fees held constant, each subsidy reduction cuts revenue even if the miner keeps the same network share.

Covers
a hypothetical miner with 1% of the network, 144 blocks a day, 0.1 BTC in fees a block and bitcoin at 100,000 US dollars
Unit
USD per illustrative day

Source: Block reward mechanics

Mining revenue at three subsidy levels
Subsidy caseMining revenue (USD per illustrative day)
3.125 BTC subsidy, 3.125 BTC subsidy$464.4K (computed)
1.5625 BTC subsidy, 1.5625 BTC subsidy$239.4K (computed)
Fees only, Fees only$14.4K (computed)

What it costs to mine a bitcoin, and what a cost figure leaves out

Compare direct cost and cost with depreciation. See what CleanSpark, Riot and TeraWulf include, how each cost compares with bitcoin earned, and why officer pay is excluded. Includes an interactive power-to-bitcoin example.

8 min read, published

A cost-boundary cutaway around one energy conduit. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Cost per bitcoin by company

Five miners, direct and with depreciation, as each company states them.

Direct, quarter ended Jun 30, 2026With depreciation, quarter ended Jun 30, 2026
CleanSpark
$44,406
$96,277
Riot
$49,912
$90,631
TeraWulf
$44,547
$97,486
MARA
$38,690
not stated
American Bitcoin
$36,500
not stated
Method and sources

For each of the three companies that state both, the figure with depreciation is the larger.

Covers
quarter ended Jun 30, 2026 for every bar, each on its company's own definition
Unit
USD per bitcoin

Source: the filings in Sources 1 to 6

Cost per bitcoin by company
CompanyDirect, quarter ended Jun 30, 2026 (USD per bitcoin)With depreciation, quarter ended Jun 30, 2026 (USD per bitcoin)
CleanSpark, CleanSpark$44,406$96,277
Riot, Riot$49,912$90,631
TeraWulf, TeraWulf$44,547$97,486
MARA, MARA$38,690not stated
American Bitcoin, American Bitcoin, a Hut 8 subsidiary$36,500not stated

Leverage: Why a miner's stock moves more than bitcoin

Why a miner's stock moves more than bitcoin, worked from its own cost figures

A miner's margin is the gap between a slow-moving cost and a bitcoin price it does not set. A small price move can create a larger margin move. The example uses CleanSpark's and Riot's filed costs and revenue per bitcoin.

9 min read, published

Two stacked slabs, a thin green wedge of margin between a high revenue slab and a fixed cost slab; a hand-sized nudge to the top slab visibly thickens the wedge many times over. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Cash margin per bitcoin

Filed direct costs stay fixed as the bitcoin price moves a tenth each way.

CleanSparkRiotHatched: computed here
A tenth lower
$20,117
$14,588
As filed
$27,286
$21,755
A tenth higher
$34,455
$28,922
Method and sources

A tenth off the price takes about a quarter to a third off the cash margin. A tenth on adds the same.

Covers
the quarter ended Jun 30, 2026, as filed, and the two outer cases worked out here
Unit
USD per bitcoin, cash margin on the direct-cost basis

Source: CleanSpark 10-Q Aug 6, 2026 and Riot 8-K Aug 10, 2026

Cash margin per bitcoin
Price caseCleanSpark (USD per bitcoin, cash margin on the direct-cost basis)Riot (USD per bitcoin, cash margin on the direct-cost basis)
A tenth lower, A tenth lower$20,117 (computed)$14,588 (computed)
As filed, As filed$27,286$21,755
A tenth higher, A tenth higher$34,455 (computed)$28,922 (computed)

Raising money: How miners raise money

Do rate cuts help bitcoin miners? Follow the debt before the headline.

A lower policy rate can change borrowing costs, cash income and valuation assumptions. A worked example keeps those effects separate from a bitcoin prediction.

6 min read, published

A brass screw clamp eased back so the green pad under its upper jaw lifts just clear of a stack of slate plates. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Interest cost before and after a cut

An illustrative year, with a one-point cut in the benchmark rate. As of Sep 23, 2026.

Hatched: computed here
Before cut
$7.8M
Cut, same spread
$7M
Cut, wider spread
$10M
Method and sources

A wider credit spread can outweigh the benchmark cut, while cash earns less interest in either cut scenario.

Covers
a hypothetical company with 100 million US dollars of floating-rate debt and 25 million in cash that earns interest
Unit
USD per illustrative year

Source: Floating-rate and credit-risk mechanics

Interest cost before and after a cut
Financing caseNet interest cost (USD per illustrative year)
Before cut, Before cut$7.8M (computed)
Cut, same spread, Cut, same spread$7M (computed)
Cut, wider spread, Cut, wider spread$10M (computed)

How bitcoin miners raise money, and what each way costs a shareholder

Compare at-the-market programs, convertible notes, equipment financing, prepaid hosting and bitcoin-backed loans. Each has a worked example from a real filing.

10 min read, published

Five distinct instruments around a treasury cube. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Money raised in five worked examples

One route each, as each filing stated it.

Hatched: computed here
IREN loan
$2.4B
MARA ATM
$2B
IREN prepay
$1.9B
WULF note
$877.6M
CLSK loans
$200M
Method and sources

The five worked examples put between 200 million and 2.4 billion dollars in a company's hands, each by a different route.

Covers
as each filing stated it, Mar 2025 to Aug 2026
Unit
USD

Source: the filings in Sources 1 to 8

Money raised in five worked examples
ExampleCash or funding (USD)
IREN loan, IREN, GPU financing at 9.0 percent$2.4B
MARA ATM, MARA, at-the-market ceiling$2B
IREN prepay, IREN, Microsoft prepayment$1.9B (computed)
WULF note, TeraWulf, convertible note, net$877.6M
CLSK loans, CleanSpark, two bitcoin-backed facilities$200M

Dilution: What dilution does to a shareholder

Dilution, explained with two miners' own share counts

What a rising share count does to what you own, worked through CleanSpark's and MARA's filed share counts, an at-the-market program and a convertible note.

9 min read, published

An ownership mosaic subdividing around one green cell. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Shares outstanding, CleanSpark and MARA

Each year end and a quarterly cover page, as filed.

CleanSparkMARA
2020
24.1M
82M
2021
41.5M
102.7M
2022
71.7M
145.6M
2023
185.6M
242.8M
2024
280.8M
340.3M
2025
255.7M
379.5M
2026 cover
256.8M
386.3M
Method and sources

CleanSpark's share count rose eleven-fold in four years and then fell. MARA's rose every year to its 10-Q cover page of Jul 30, 2026.

Covers
year-end counts 2020 to 2025 and the 10-Q cover-page counts of Aug 4, 2026 and Jul 30, 2026
Unit
shares

Source: SEC XBRL company concept, each filing

Shares outstanding, CleanSpark and MARA
CategoryCleanSpark (shares)MARA (shares)
2020, 202024.1M82M
2021, 202141.5M102.7M
2022, 202271.7M145.6M
2023, 2023185.6M242.8M
2024, 2024280.8M340.3M
2025, 2025255.7M379.5M
2026 cover, 2026 cover256.8M386.3M

Guidance: What a miner promised, and what it delivered

How to check what a miner promised against what it delivered

Find a forward statement, pin its number and date, then match it to a later filing. Six hashrate promises from five companies are scored the same way. A result already in hand is not a promise.

9 min read, published

A forecast flag planted at a target mark on a ruler, and a later outcome marker landing short of it with a green measured gap between them. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Hashrate promised and delivered (EH/s)

Six promises from five companies, each settled in a later filing.

PromisedDelivered
MARA 75
75
66.4
CLSK 50
50
50
BTDR 50
50
55.2
BTDR 40
40
41.2
HIVE 25
25
25
ARBK 1.7
1.7
2.4
Method and sources

Two of the six landed on the number, three landed above it, and one fell short, by 11.5 percent.

Covers
promises stated Dec 2024 to Dec 2025, outcomes filed Aug 2025 to Apr 2026
Unit
EH/s

Source: the ledger rows, each with both filings, in Sources

Hashrate promised and delivered (EH/s)
PromisePromised (EH/s)Delivered (EH/s)
MARA 75, MARA, 75 EH/s energized by Dec 31, 20257566.4
CLSK 50, CleanSpark, 50 EH/s from 211.5 MW5050
BTDR 50, Bitdeer, 50 EH/s by year-end 20255055.2
BTDR 40, Bitdeer, 40 EH/s by end of October 20254041.2
HIVE 25, HIVE, 25 EH/s by US Thanksgiving 20252525
ARBK 1.7, Argo, 1.7 EH/s go-forward fleet1.72.4

Insider trades: What a Form 4 tells you

What a Form 4 tells you about an insider's trade, and what it does not

Read the transaction codes, separate open-market sales from shares withheld for tax, and understand 10b5-1 timing. The article covers 12 months of filed sales and purchases across the covered companies.

10 min read, published

An insider's trade form with two stamped boxes: one box a sale at open-market, a second box shares withheld to pay tax, a small clock-lock marking a pre-set plan; the sheet is signed with a seal. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

A CleanSpark director's Form 4 sale

Shares held before, sold and kept afterward. As of Dec 5, 2025.

Hatched: computed here
Held before
140.3K
Sold
33K
Owned afterward
107.3K
Method and sources

The sale took 33,000 of the 140,289 shares the director held, 23.5 percent, and left 107,289.

Covers
the trade of Dec 4, 2025, the holding before worked out from the other two
Unit
shares

Source: CleanSpark Form 4, filed Dec 5, 2025

A CleanSpark director's Form 4 sale
HoldingShares (shares)
Held before, Held before140.3K (computed)
Sold, Sold33K
Owned afterward, Owned afterward107.3K

Datacenter deals: What a datacenter deal is worth

What a datacenter deal headline is worth, and the four numbers that say so

Compare contract value, term and megawatts. Separate a letter of intent from a signed lease and trace the build funding. Twelve leases from nine companies share one scale.

10 min read, published

A contract slab resting on four independent supports. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Contract value of datacenter leases

Twelve leases from nine companies, the total over each term, as filed.

WULF 20 yr
$19B
CORZ AMD
$14B
CORZ CRWV
$10B
HUT Beacon
$9.8B
IREN MSFT
$9.7B
RIOT lab
$9.1B
HUT River
$7B
CLSK
$6.6B
CIFR AWS
$5.5B
BTDR Tydal
$4.7B
CIFR FLS
$3B
DGXX
$1.1B
Method and sources

Headline values run from 1.1 billion to 19 billion dollars, and a 20-year lease carries twice the total of a 10-year lease at the same rent.

Covers
leases announced 2024 to Aug 2026, as filed
Unit
USD over the stated term

Source: the filings in Sources, one per lease

Contract value of datacenter leases
LeaseTotal contract value (USD over the stated term)
WULF 20 yr, TeraWulf and Anthropic, 401 MW, 20 years$19B
CORZ AMD, Core Scientific and AMD, 530 MW, 15 years$14B
CORZ CRWV, Core Scientific and CoreWeave, 590 MW, 12 years$10B
HUT Beacon, Hut 8 Beacon Point Phase 1, 352 MW, 15 years$9.8B
IREN MSFT, IREN and Microsoft, 200 MW, 5 years$9.7B
RIOT lab, Riot and a frontier AI lab, 191 MW, 20 years$9.1B
HUT River, Hut 8 River Bend, 245 MW, 15 years$7B
CLSK, CleanSpark Sandersville, 175 MW, 20 years$6.6B
CIFR AWS, Cipher and AWS at Black Pearl, 216 MW, 15 years$5.5B
BTDR Tydal, Bitdeer at Tydal, 121 MW, 16 years$4.7B
CIFR FLS, Cipher and Fluidstack at Barber Lake, 168 MW, 10 years$3B
DGXX, Digi Power X and Cerebras, 40 MW, 10 years$1.1B

Taxes: Taxes on miner stock and bitcoin

Taxes for someone who holds miner stock or bitcoin, in plain terms

Holding periods, capital gains, wash sales, bitcoin differences and 1099s. It also explains why a miner's tax position appears in filings. United States, education only.

10 min read, published

A calendar clock cut at the one-year mark, a short-term stack and a long-term stack of plain tokens beside it, with a wash-sale mirror duplicate tile and a filed form slip. A small MinerTerminal mark sits in the top-left corner.
Illustration: AI-generated artwork created for MinerTerminal in ChatGPT from original article concepts.

Where the capital gain rates end

Taxable income where the 0 and 15 percent rates stop, by filing status. As of Oct 2, 2026.

0 percent up to15 percent up to
Single
$48,350
$533,400
Married filing jointly
$96,700
$600,050
Head of household
$64,750
$566,700
Method and sources

The 0 percent band ends between 48,000 and 97,000 dollars of taxable income. The 15 percent band runs to more than half a million.

Covers
taxable years beginning in 2025
Unit
USD of taxable income

Source: IRS Topic no. 409

Where the capital gain rates end
Filing status0 percent up to (USD of taxable income)15 percent up to (USD of taxable income)
Single, Single$48,350$533,400
Married filing jointly, Married filing jointly$96,700$600,050
Head of household, Head of household$64,750$566,700