$83,479

Bitcoin spot price, 28 Sep 2026

$39.87

Spot hashprice per PH/s per day, 28 Sep 2026

$30.37

June 2026 average hashprice, the lowest month on record

~18 mo

Until the next halving (≈ April 2028) cuts the block subsidy in half

Mining · S19j Pro class (29.5 J/TH)

$56

Mining · S21 Pro class (15 J/TH)

$111

Mining · S23 Hyd class (9.5 J/TH)

$175

HPC colocation (power shell)

$86–300

AI cloud (GPUs owned by operator)

$807–1,213

$114bn+

Long-term AI/HPC contracts announced by public bitcoin miners (Aug 2026)

≈ 1/4

Of that contracted capacity actually built so far

30% → 70%

AI share of public-miner revenue: Q1 2026 vs analysts' end-2026 forecast

≈ 15×

Capex vs period revenue at six miners already earning recurring HPC revenue

Years 0–2

Frontline

Industrial farms on the cheapest firm power. Efficiency is the moat; capex must pay back before the halving.

Years 2–4

Secondary

Resold to interruptible or stranded power: flare gas, curtailment, solar surplus. Low capex makes low uptime acceptable.

Years 4+

Heat-first

Water heaters, space heaters, district heating. The electricity is bought as heat; hashing becomes a rebate.

End of life

Harvest

Hashboards are desoldered; good chips go into Bitaxe-style micro-miners. The rest is e-waste.

Hydro $0.035

Industrial $0.08

EU retail $0.20

Hot water

Superheat H1 (CES 2026): a ≈$2,000 water heater with an ASIC as its heat core. The maker claims ≈$1,000/yr of mining income, ≈2-year payback and a 10-year life, at the same energy use as an electric water heater.

Vendor claims

District and space heat

Canaan supplies hydro-cooled units to an 8 MW Nordic district-heating project for ≈2,800 homes. MARA runs a 2 MW pilot in Finland; Hashlabs hosts six Finnish heat-reuse sites.

Best with year-round heat demand

Micro-miners

Bitaxe started with chips desoldered from used Antminer hashboards. ≈1 TH/s at 15–20 W earns ≈$17.5/yr in a pool, against ≈$31/yr of power at $0.20/kWh.

Value is the device, not the hash

New ASICs are a bet on BTC price

On hydro power they need ≈$54/PH/day, 34% above today, to pay back before the April 2028 halving.

Cheap capex suits cheap, intermittent energy

Idle hours waste capital, not power. Paid-off machines turn solar surplus or heat demand into fast paybacks.

Mine with new ASICs

≈ $3.5M

≈$2.6M of ASICs at $25/TH plus $0.7–1M site

Net ≈ $1.14M/yr now, ≈ $0.42M/yr after the halving (hydro $0.035, 95% uptime)

Revenue from day one; hardware refresh every cycle

Convert the site to AI

$3–4M

Conversion capex per MW of an existing mining site

Lease ≈ $1.5–2M/yr per MW; TeraWulf's contract implies ≈$1.85M

12–20 year terms; needs a tenant, firm power and fibre

Build new AI capacity

$10–12M

Per MW from scratch; $8–15M if liquid-cooled

Returns ≈ 12–15% on new builds

Most certainty, most capital, longest time to revenue

01

Firm power, a grid connection and capital? Lease it to AI: more revenue per MW, contracted for 12–20 years.

02

Cheap but interruptible power? Mining is its natural buyer; match ASIC age to uptime.

03

Need heat anyway? Old ASICs pay back in months against resistive heating; against a heat pump only the newest generation wins, and only until the halving.

04

Micro-miners are a product and a hobby, not an income. Monetise the device, not the hashrate.

05

Model every ASIC purchase across April 2028. Assume revenue per kWh halves unless BTC price rises to offset it.