How to Analyze Bitcoin Mining Stocks: Fleet, Power, Dilution, and HPC Economics

Short answer: Analyze a Bitcoin miner as a power-and-compute business with a volatile commodity output, not as a simple leveraged Bitcoin proxy. Reconcile average operating hashrate, uptime, fleet efficiency, network share, transaction fees, pool method, realized power cost, curtailment, hosting, depreciation, machine capex, BTC sales, debt, and diluted shares. Then stress Bitcoin price, network difficulty, fees, uptime, and power together.

The former guide used stale company statistics, arbitrary EV/EH bands, unverified all-in cost figures, subjective peer scores, and a model portfolio. It also treated high BTC retention as inherently good and every power site as ready for AI. Those shortcuts have been removed.

This replacement provides a filing-based operating model, cost bridge, hashrate reconciliation, miner sensitivity table, capital-allocation framework, HPC conversion gate, and comparison template.

Reviewed July 11, 2026. Company fleets, network conditions, power contracts, BTC holdings, and share counts change frequently.

Mining Revenue Starts With Network Share

Bitcoin miners contribute computing power to find blocks. Revenue comes from the block subsidy and transaction fees. A miner's expected share depends on its effective hashrate relative to the network.

Expected BTC per day = effective miner hashrate / network hashrate x expected blocks per day x expected BTC reward per block x (1 - pool fee)

The protocol targets roughly one block every ten minutes, or about 144 blocks per day over time. Actual daily blocks and company production vary.

After the April 2024 halving, the subsidy is 3.125 BTC per block. Transaction fees are variable and must be added separately:

Expected reward per block = 3.125 BTC subsidy + expected transaction fees

The old shortcut of 450 BTC per day assumes exactly 144 blocks and ignores fees, pool economics, variance, and subsequent changes.

Worked Production Model

Assume:

average effective miner hashrate: 20 EH/s;

average network hashrate: 1,000 EH/s;

blocks per day: 144;

subsidy: 3.125 BTC;

average transaction fees: 0.10 BTC per block;

pool fee: 0.5%.

Gross expected network reward:

144 x (3.125 + 0.10) = 464.4 BTC per day

Miner share:

20 / 1,000 = 2.0%

Expected company production:

464.4 x 2.0% x 99.5% = 9.24 BTC per day

For 30 days:

9.24 x 30 = 277.2 BTC

This is an expectation, not a guarantee. Pool payout method, downtime, network growth, block luck, and curtailment can change realized production.

"Hashrate" Needs a Definition

Public miners disclose several numbers that are not interchangeable.

Metric	Meaning	Typical trap
Owned capacity	Nameplate output of machines owned	Includes units not installed or operating
Installed hashrate	Machines physically deployed	May lack power or commissioning
Energized hashrate	Connected and capable of running	Can differ from actual average output
Operational or active hashrate	Running at the observation time	Point-in-time snapshot
Average operating hashrate	Average delivered over a period	Best link to period production
Managed hashrate	Includes hosted or third-party machines	Economics may differ by contract
Proportionate hashrate	Joint-venture share included	Ownership and cash flow may not map one-for-one
Future target	Management plan	Requires machines, infrastructure, permits, power, and capital

MARA's 2025 filing, for example, reported energized hashrate that included a proportionate share of an equity-method investee and distinguished owned from hosted sites. CleanSpark's filing disclosed owned miners, units in service, average operating hashrate, and point-in-time fleet efficiency. A leaderboard that copies the largest number from each company is not comparable.

Hashrate Reconciliation

For each company and period:

Effective utilization = average operating hashrate / installed or energized hashrate

If 30 EH/s is energized but average operating hashrate is 24 EH/s:

Utilization = 24 / 30 = 80%

The gap can reflect maintenance, curtailment, commissioning, heat, network issues, pool reporting, or power constraints. Investigate rather than assuming poor operations.

Fleet Efficiency: Convert J/TH Into Power Cost

Fleet efficiency, expressed as joules per terahash, is numerically equivalent to watts per terahash under continuous operation.

For one petahash per second:

Power draw in kW per PH/s = J/TH x 1,000 TH/PH / 1,000 W/kW

This simplifies numerically to:

kW per PH/s = J/TH

At 20 J/TH, one PH/s of mining capacity draws about 20 kW at the machine level.

Daily machine energy:

20 kW x 24 hours = 480 kWh per PH/day

At $0.05 per kWh:

Machine electricity cost = 480 x $0.05 = $24 per PH/day

This excludes cooling, transformers, fans, pumps, networking, and other facility load.

Include Power Usage Effectiveness

Power usage effectiveness:

PUE = total facility energy / IT equipment energy

If machine load is 20 kW per PH and PUE is 1.08:

Facility load = 20 x 1.08 = 21.6 kW per PH

Daily facility energy cost at $0.05:

21.6 x 24 x $0.05 = $25.92 per PH/day

Comparing miners on nameplate J/TH without PUE, climate, overclocking, downtime, and actual power price is incomplete.

Hashprice Connects Revenue to Compute

Hashprice expresses expected miner revenue per unit of hashrate over time, commonly dollars per PH/s per day. It reflects BTC price, network difficulty or hashrate, subsidy, transaction fees, and sometimes a vendor-specific convention.

Gross compute margin per PH/day = realized hashprice - facility energy cost per PH/day - pool fee impact - variable site cost

Suppose:

hashprice: $48/PH/day;

facility energy cost: $25.92/PH/day;

pool and variable operations: $3/PH/day.

Gross compute margin = $48 - $25.92 - $3 = $19.08/PH/day

If hashprice falls to $28:

Margin = $28 - $25.92 - $3 = -$0.92/PH/day

A miner may curtail rather than operate at a negative variable margin, depending on contracts and credits.

Power Break-Even Hashprice

Power break-even hashprice = J/TH x PUE x 24 x electricity price per kWh

Because J/TH equals kW per PH numerically, the units resolve to dollars per PH/day.

At 25 J/TH, 1.10 PUE, and $0.06/kWh:

25 x 1.10 x 24 x $0.06 = $39.60/PH/day

This is a power-only threshold. Corporate survival requires more revenue for labor, repairs, leases, SG&A, capex, debt, and taxes.

"Cost per BTC" Has No Standard Definition

Company-reported costs often use different numerators and denominators.

Cost layer	Possible inclusions
Electricity-only	Metered or purchased energy
Net electricity	Power less credits or demand-response benefits
Direct mining cost	Energy, labor, repairs, insurance, site costs
Cost of revenue	Accounting definition; may exclude depreciation
Cash cost	Direct costs plus selected cash overhead
All-in operating cost	Direct cost, SG&A, stock compensation choices
Full economic cost	Cash costs, depreciation, interest, sustaining capex, dilution

Riot's 2025 filing says power curtailment credits are presented outside cost of revenue but significantly reduce overall mining cost. MARA separately disclosed purchased energy at owned sites and third-party hosting. CleanSpark disclosed average electricity cost per kWh and cost per BTC under its own method. Copying those values into one table without adjustment is misleading.

Build a Cost Bridge

Start with GAAP statements:

1.electricity and hosting;

2.site labor and repairs;

3.mining pool and network costs;

4.power credits;

5.depreciation and impairment;

6.corporate SG&A;

7.stock compensation;

8.interest;

9.BTC produced.

Publish several layers:

Net direct cash cost per BTC = (energy + hosting + direct site cost - power credits) / BTC produced

Operating cash cost per BTC = (net direct cash cost + cash SG&A) / BTC produced

Accounting cost per BTC = (cost of revenue + depreciation + allocated operating expense) / BTC produced

Economic replacement cost per BTC = operating cash cost + sustaining machine and infrastructure capex per BTC + financing cost

Do not subtract power credits twice if the company already nets them against expense.

Worked Miner Income Model

Assume one year:

BTC produced: 3,300;

realized BTC revenue price: $85,000;

mining revenue: $280.5 million;

energy and hosting: $110 million;

direct labor and repairs: $25 million;

power credits: $12 million;

cash SG&A: $35 million;

depreciation: $75 million;

interest: $15 million;

sustaining capex estimate: $45 million.

Net direct cash cost:

$110M + $25M - $12M = $123M

Net direct cash cost per BTC:

$123M / 3,300 = $37,273

Operating cash cost per BTC:

($123M + $35M) / 3,300 = $47,879

Accounting cost before taxes:

($123M + $35M + $75M + $15M) / 3,300 = $75,152

Economic cash requirement including sustaining capex:

($123M + $35M + $15M + $45M) / 3,300 = $66,061

The miner can claim a sub-$40,000 direct cost while needing substantially more value per BTC to cover the business.

Network Difficulty Can Offset Fleet Growth

A miner can add hashrate and produce fewer BTC if the network expands faster.

Expected production growth is approximately:

New production / old production = (new miner hashrate / old miner hashrate) / (new network hashrate / old network hashrate) x reward change x uptime change

Suppose:

company average hashrate rises 25%;

network hashrate rises 40%;

reward and uptime are unchanged.

Production ratio = 1.25 / 1.40 = 0.893

Expected BTC production falls about 10.7% despite fleet growth.

Bitfarms' 2025 filing provides a real disclosure pattern: production reflected higher company hashrate and improved efficiency but was offset by the halving and a materially higher average network difficulty. Use this causal bridge rather than celebrating EH/s alone.

Uptime Is an Economic Metric

Installed machines earn nothing while offline. Track:

machine availability;

power availability;

network connectivity;

scheduled maintenance;

unscheduled outage;

curtailment;

commissioning;

pool rejection rate.

If nameplate capacity is 40 EH/s but average delivered hashrate is 32 EH/s:

Portfolio uptime/utilization proxy = 32 / 40 = 80%

At $48/PH/day, the missing 8 EH/s represents gross foregone revenue:

8,000 PH/s x $48 x 365 = $140.16 million

That gross figure overstates lost profit because downtime saves power and may earn credits. Calculate net opportunity cost:

Foregone net mining margin - curtailment or alternative revenue

Power Contracts Can Be Assets or Traps

Review:

fixed, indexed, or market price;

demand charges;

transmission and congestion;

take-or-pay obligations;

pass-through fees;

curtailment rights;

demand-response eligibility;

term and renewal;

collateral or deposits;

counterparty credit;

interconnection status.

A low advertised energy rate may exclude transmission, taxes, demand charges, or hosting markup. An interruptible contract may be cheaper but reduce production.

Curtailment Decision

Mine when incremental mining contribution exceeds the value of selling power or earning credits.

Mine contribution per MWh = mining revenue per MWh - variable operating cost per MWh

Curtail when power credit or avoided market cost per MWh > mine contribution per MWh

Riot describes manually curtailing when power-market economics exceed forgone mining revenue and separately participates in grid programs. Credits should be analyzed as part of the power strategy, not automatically treated as recurring Bitcoin revenue.

Owned Versus Hosted Infrastructure

Owned and Operated

Benefits:

control over power and site design;

potential curtailment economics;

infrastructure residual value;

lower long-term hosting markup.

Risks:

land, substation, building, and cooling capex;

construction delays;

permits and interconnection;

maintenance and staffing;

stranded site risk.

Third-Party Hosting

Benefits:

faster deployment;

lower initial infrastructure capex;

geographic flexibility.

Risks:

host credit and performance;

contract renewal and termination;

pass-through power;

less control over uptime;

deposits and disputes;

concentration.

Report hashrate and economics by ownership model. Two miners with 30 EH/s may own very different assets and liabilities.

ASIC Depreciation and Replacement Capex

ASICs lose economic value through wear, new generations, higher network difficulty, and changing power economics. Accounting useful life is an estimate, not physical truth.

CleanSpark's 2025 filing disclosed a three-year estimated useful life for miners and a wide range of fleet efficiencies. That information helps estimate replacement needs.

Sustaining Capex Model

Annual sustaining ASIC capex = hashrate requiring replacement x replacement cost per TH

If 10 EH/s, or 10 million TH/s, must be replaced at $18/TH:

10,000,000 TH x $18 = $180 million

Add shipping, tariffs, containers, installation, deposits, and downtime.

Efficiency Savings

Replacing 28 J/TH machines with 16 J/TH machines saves 12 kW per PH.

At 10 EH/s:

10,000 PH x 12 kW = 120 MW

Annual energy saved at full operation:

120 MW x 8,760 hours = 1,051,200 MWh

At $45/MWh:

Annual gross energy savings = $47.3 million

Compare savings and production benefits with purchase, installation, downtime, and financing costs.

HODL Rate Is Capital Allocation, Not Quality

BTC retention:

Retention rate = (BTC produced + BTC purchased - BTC sold - other outflows) / BTC produced

This can exceed 100% if a miner purchases BTC.

Retaining BTC may be sensible when:

operating cash is strong;

debt maturities are manageable;

capex is funded;

shares are not issued destructively;

management intentionally wants BTC exposure.

Selling production may be sensible when:

it avoids expensive debt or dilution;

power and payroll are due;

machines offer attractive returns;

concentration limits are exceeded;

treasury risk is too high.

IREN's filing describes a model that typically sold mined BTC daily while investing in infrastructure and AI services. That is not inherently inferior to a high retention rate. Compare BTC per diluted share, net debt, and operating value.

BTC per Diluted Share for Miners

Miner shareholders can be diluted while company BTC holdings rise.

BTC per diluted share = corporate-owned BTC / conservative diluted shares

Suppose:

beginning BTC: 8,000;

ending BTC: 10,000, up 25%;

beginning diluted shares: 200 million;

ending diluted shares: 280 million, up 40%.

Beginning BTC per share:

8,000 / 200M = 0.000040 BTC

Ending:

10,000 / 280M = 0.00003571 BTC

BTC per share falls 10.7%.

Track at-the-market issuance, converts, warrants, stock compensation, acquisition shares, and earnouts. Growth funded by repeated equity issuance is not free.

Balance-Sheet Survival Test

Build a 12- to 24-month liquidity schedule.

Sources

unrestricted cash;

expected operating cash under stress;

BTC available for sale;

committed credit;

contracted power credits;

asset sales.

Uses

power and hosting;

payroll and SG&A;

debt interest and principal;

machine purchase commitments;

construction capex;

leases;

taxes;

collateral calls.

Worked Stress

Assume:

unrestricted cash: $200M;

unrestricted BTC: $300M at current price;

annual cash operating need: $260M;

committed capex: $180M;

debt service: $40M;

stress mining cash contribution: $80M.

12-month funding gap before BTC sales:

$260M + $180M + $40M - $80M - $200M = $200M

The company must sell BTC, cut capex, refinance, issue equity, or renegotiate obligations. If BTC is pledged, it may not be available.

Valuing a Miner: Use Sum of Parts

Enterprise value:

EV = common market cap + debt + preferred claims - cash

Then value:

1.unrestricted BTC and other investments;

2.mining operation;

3.owned land, power, and infrastructure;

4.hosting or AI/HPC contracts;

5.development pipeline;

6.corporate liabilities and commitments.

Mining Operation DCF

Forecast by quarter:

average effective EH/s;

network EH/s or difficulty;

subsidy and fees;

realized BTC price;

power and direct cost;

pool fee;

capex;

taxes.

Use a finite forecast because machines and contracts change quickly. Apply a conservative terminal or salvage value rather than assuming a permanent hashrate premium.

EV per EH Is a Screening Metric

EV per effective EH = adjusted enterprise value / average effective hashrate

Adjust EV for BTC and non-mining assets if comparing mining capacity:

Mining EV = EV - unrestricted BTC value - non-mining asset value

Then:

Mining EV per effective EH = mining EV / average effective EH/s

Even adjusted EV/EH ignores efficiency, power cost, ownership, uptime, capex, network growth, and site life. There are no universal "cheap" bands.

Worked Miner Comparison

Consider two hypothetical miners:

Metric	Miner A	Miner B
Energized hashrate	30 EH/s	30 EH/s
Average effective hashrate	27 EH/s	21 EH/s
Fleet efficiency	18 J/TH	24 J/TH
PUE	1.06	1.12
All-in power	$0.05/kWh	$0.04/kWh
Owned infrastructure	80%	20%
Net debt	$300M	$50M
Annual sustaining capex	$140M	$90M

Power cost per PH/day:

A = 18 x 1.06 x 24 x $0.05 = $22.90

B = 24 x 1.12 x 24 x $0.04 = $25.80

Miner B's cheaper electricity does not overcome its less efficient fleet and higher PUE. Miner A has higher uptime and site ownership but also more debt and capex.

No one score answers which stock is better. Value each cash-flow and capital-structure path.

AI and HPC Are Separate Businesses

Access to power and land is useful, but a Bitcoin mine is not automatically an AI data center.

HPC and AI customers may require:

high availability and redundancy;

firm, non-curtailable power;

fiber diversity and low latency;

liquid cooling;

high-density racks;

physical and cyber security;

certifications and service-level agreements;

long construction and commissioning;

customer credit and contracts;

GPUs or customer-provided equipment;

substantial capital.

Bitcoin mining often values interruptible power, simple sheds or containers, lower network requirements, and rapid relocation. Those designs can conflict.

Core Scientific's 2025 filing illustrates the capital intensity: it disclosed major annual capex and substantial contractual commitments for high-density colocation conversion. IREN reports Bitcoin mining and AI Cloud Services as separate segments and warns about customer acquisition, competition, capital, and execution.

HPC Conversion Gate

Award value only when evidence advances:

Stage	Evidence	Valuation treatment
Narrative	Management says site could support AI	No separate value
Feasibility	Engineering, fiber, cooling, permits, firm power	Option value with probability haircut
Contracted	Named creditworthy customer, term, MW, pricing, capex allocation	Contract-level DCF
Under construction	Funded capex and milestones	Execution-adjusted value
Energized	Customer acceptance and billing	Operating cash-flow value

Avoid valuing every megawatt at an AI multiple.

Environmental and Regulatory Analysis

Review by site:

grid and utility;

energy source and contract;

water and cooling;

emissions exposure;

noise;

zoning;

tax incentives;

community agreements;

demand-response rules;

import tariffs on machines;

data-center moratoria or load restrictions.

EIA notes that mining power use can change rapidly as miners respond to prices and machine availability and that estimates carry wide ranges. Company-specific metered data is stronger than broad industry estimates.

A Filing-Normalized Miner Worksheet

For each quarterly period, collect:

Production

BTC produced;

BTC sold and purchased;

average realized price;

pool method and fee;

transaction-fee contribution if disclosed.

Compute

owned capacity;

installed and energized EH/s;

average effective EH/s;

fleet J/TH;

uptime or utilization;

network hashrate/difficulty source.

Power

MWh consumed;

total and net power cost;

$/kWh or $/MWh;

PUE;

credits and curtailment;

owned versus hosted split.

Costs

energy and hosting;

direct labor and repair;

depreciation;

SG&A and stock compensation;

interest;

sustaining and growth capex.

Capital

cash;

BTC by restriction;

debt and maturity;

shares issued;

diluted shares;

machine and construction commitments.

Diversification

contracted HPC MW;

customer and term;

capex responsibility;

segment revenue and margin;

milestones.

Attach filing page references. Do not fill missing values with peer assumptions unless clearly labeled.

Frequently Asked Questions

What is the most important miner metric?

No single metric. Average effective hashrate connects to production; fleet efficiency and realized power cost connect to variable margin; capital structure and dilution determine what common shareholders own.

Is lower J/TH always better?

It lowers machine power per hash, but purchase cost, reliability, PUE, power price, uptime, and financing matter. A slightly less efficient paid-off fleet on cheap power can outperform expensive new machines.

Is cost per BTC comparable across companies?

Not until definitions are normalized. Check whether the numerator includes power credits, hosting, labor, depreciation, SG&A, interest, stock compensation, and capex.

Does hashrate growth guarantee production growth?

No. Network hashrate or difficulty can grow faster, while uptime, curtailment, pool luck, and the block reward can change.

Is retaining all mined BTC a sign of strength?

Not necessarily. Retention funded by debt or dilution can weaken shareholders. Selling BTC to fund high-return operations or avoid expensive financing can be rational.

Are power credits recurring revenue?

They depend on contracts, grid programs, market prices, weather, and curtailment decisions. Analyze their source and presentation over several periods.

Is a miner with large power capacity an AI company?

No. HPC requires firm power, cooling, fiber, reliability, capital, technical capability, and customers. Value contracted and funded projects separately from potential.

Why can miner stocks fall faster than Bitcoin?

Bitcoin price can reduce revenue while difficulty, power, payroll, debt, and capex remain. Equity dilution and valuation compression add another layer of downside.

Research Method and Limits

This framework uses Bitcoin protocol documentation, SEC filings from MARA, CleanSpark, Riot, Bitfarms, Core Scientific, and IREN, plus EIA power-demand analysis. Company disclosures are primary but can use different non-GAAP definitions and management assumptions.

The guide omits a live peer ranking because comparable analysis requires synchronized quarterly filings, current diluted shares, power contracts, machine status, and network conditions. Worked examples are illustrative and are not company forecasts.

What to Read Next

Continue with <a href="/insights/bitcoin-mining-economics-2026">Bitcoin Mining Economics 2026</a> for the network-level difficulty, hashprice, power-cost, fee, fleet-efficiency, and post-halving model that underpins this stock framework.

About the Editorial Team

CryptosEyes Research normalizes miner disclosures before comparison. We distinguish energized from effective hashrate, direct from full cost, sustaining from growth capex, and contracted HPC revenue from power-site optionality.