Profitability Is a Ratio, Not a Number
Ask ten mining operators whether their fleet is profitable and you will get ten different answers. Profitability is not a property of a machine. It is the relationship between what a unit of computation earns and what it costs to keep that computation running, and that relationship shifts every single day. A rig that prints money under one power contract is a liability under another. The useful exercise is not asking whether mining is profitable in general, but understanding which variables move that ratio and by how much.
The Revenue Side: Hashprice Sets the Ceiling
Revenue per terahash per day, usually called hashprice, bundles three inputs: the spot price of the coin being mined, the block subsidy, and the share of block rewards coming from transaction fees. Of these, only the first is visible in headlines. The second changes on a schedule everybody knows in advance, which is why halving dates matter more to fleet planning than any short-term price move.
The quiet variable is network difficulty. Every additional unit of computing power pointed at a SHA-256 chain spreads the same block rewards across a larger denominator. Difficulty growth is the tax that hardware buyers pay for their own optimism: the machines purchased today compete against the machines purchased yesterday and the machines that will arrive next quarter. An operator modelling revenue using today's difficulty is modelling a number that is almost certainly too generous.
Fee revenue deserves a caution of its own. During quiet periods fees contribute a thin slice of block rewards, and during congestion they can temporarily dominate. Building a capital plan around fee spikes is a good way to overpay for hardware.
The Cost Side: Efficiency, Power, and Uptime
On the cost side, efficiency measured in W/TH sets the floor. It is not simply a matter of buying the most efficient units available. The relevant figure is the efficiency actually realised in the rack, after delivery losses, transformer losses, firmware tuning, and thermal reality. A coherent cost model includes:
- All-in power cost per kWh, including transmission, demand charges, and standby fees.
- Realised efficiency of the deployed unit, not the datasheet figure.
- Uptime, including scheduled curtailment, cleaning, and repair windows.
- Cooling overhead, whether air, hydro, or immersion, plus the cost of managing heat rejection.
- Hosting margin, where the operator does not self-host.
- Capital amortisation and the expected residual value of the machine at the end of its useful life.
Breakeven and the Shutdown Decision
Every machine has a shutdown price: the power cost per kWh at which revenue per day equals running cost. Operators rarely shut down cleanly, because stopping a miner forfeits the option value of restarting when conditions improve, and because contractual power commitments do not pause. This is why curtailment programs matter so much. A fleet that can flex down during peak pricing converts a fixed cost into a variable one, and that flexibility is often worth more than a marginal efficiency gain.
Fleet Composition Decides Who Survives
Low efficiency units do not disappear when margins compress; they migrate to the cheapest power in the world, where they can still run profitably. That migration is a reliable signal of where the market believes power is cheapest. Meanwhile, high-efficiency immersion units tend to stay with operators who pay more per kWh but demand maximum uptime and density. Neither strategy is universally correct. The mistake is running high-consumption hardware on mid-tier power while assuming a bull market will cover the gap.
What to Watch Instead of Headlines
Ignore profitability forecasts built on price predictions, because they inherit the forecast's error without adding information. Track difficulty growth rates, hashprice trends, and the balance of second-hand supply against new arrival volume. Buy hardware when the efficiency of the unit materially changes your all-in breakeven, not when the price of the coin is exciting. The operators who survive multiple cycles are the ones who treat mining as a power-cost arbitrage business with a hardware component, rather than the reverse.