Bitcoin's Production-Cost Floor Broke — And It Was Never a Floor to Begin With

For most of Bitcoin's history, one number was treated as a soft floor under the price: the cost of producing a coin. The logic was intuitive. Miners must sell to pay for electricity and machines; if price falls below cost, the weakest miners switch off, hashrate drops, difficulty adjusts down, and the survivors' costs fall until the market clears. The floor was supposed to hold.

It has not. As of July 2026, bitcoin trades around $64,200 while credible estimates of the all-in production cost for listed miners sit near $78,000 — and some estimates run higher. JPMorgan has told clients that mining economics have 'worsened,' with bitcoin trading roughly 19% below production cost. The floor did not hold. It broke, and it has stayed broken for months. That deserves an explanation rather than a slogan.

What the numbers actually say

Start with the spread between the estimates, because it is the first thing that should make you cautious about the whole heuristic. 'Production cost' is not one number. Industrial miners spend roughly $32,000 to $55,000 per bitcoin in hosting and power alone, depending on their electricity rate, machine efficiency and uptime — the most efficient hydro-powered fleets at around $0.07/kWh sit at the bottom of that range. Add hardware depreciation, corporate overhead, debt service and the picture changes: all-in cost across public miners lands near $78,000, and some models push past $100,000.

So which number is 'the floor'? At $64,200, bitcoin is comfortably above the marginal cash cost of the most efficient operators and far below the fully-loaded accounting cost of the average listed miner. Both statements are true simultaneously. That is not a paradox; it is the entire point.

A miner shuts down when price falls below the cash cost of the next kilowatt-hour, not when price falls below the accounting cost that includes machines already bought and debt already borrowed. Sunk costs do not turn off ASICs.

The network's behaviour confirms it. Hashrate has come down hard — from above 1,000 EH/s in late April and early May to roughly 862–963 EH/s depending on the measurement window — and difficulty fell 10.09% at block 953,568, from 138.9 trillion to 124.9 trillion, the eleventh-largest downward adjustment in Bitcoin's history and the second-largest of 2026. Machines have genuinely left. But they have not left in the numbers the 'price floor' theory would predict, because most miners can still cover their power bill at $64,000.

Hashprice, not production cost, is the variable that binds

The metric that actually governs miner behaviour is hashprice: the revenue a miner earns per unit of hashrate per day, quoted in dollars per petahash per second per day. It collapses price, difficulty and transaction fees into one number, and it is the number a mining CFO looks at every morning. In July 2026 hashprice is hovering around $29/PH/s/day — a level the industry has never seen before.

This is why the production-cost floor failed. Production cost is an output of hashprice, not an input to it. When hashprice falls, the cost of producing a coin for any given miner rises mechanically, because the same electricity now buys fewer coins. Quoting 'bitcoin is 19% below production cost' therefore describes the symptom, not a cause, and it certainly does not describe a mechanism that pulls price back up. The causal arrow runs from price to cost, not from cost to price.

Difficulty adjustment does help — but it helps the survivors, not the price. The 10.09% cut handed the miners who stayed online roughly 11% more bitcoin per unit of active hashrate. That improves the economics of the remaining fleet. It does nothing whatsoever to create a marginal buyer of bitcoin.

The split: survivors and sellers

The more useful way to read this regime is that it is sorting the industry rather than supporting the price. Roughly 15–20% of the global fleet is estimated to be underwater at current economics. Those operators face a narrow menu: sell treasury coins to fund operations, raise equity or debt, sell machines, pivot capacity to AI and high-performance computing, or shut down.

Each of those choices has a different market consequence, and only one of them is bullish. Miners who sell treasury BTC add supply into an already thin bid. Miners who pivot to AI stop being bitcoin-levered businesses at all. Miners who shut down remove hashrate, which lowers difficulty and helps the survivors' margins — the only genuinely self-correcting channel, and it operates on the network, not on the price.

Steel-manning the other side

There is a serious argument that this is a bottoming signal, and it should not be dismissed. Historically, extended miner capitulation — hashrate falling, difficulty cutting, weak hands exiting — has clustered near cyclical lows. The reasoning is that forced miner selling is a finite, exhaustible source of supply. Once the underwater 15–20% has sold what it must sell, that overhang is gone, and the remaining float is held by stronger hands at a lower cost basis. On this view, the broken production-cost floor is not evidence that the model is wrong; it is evidence that the flush is in progress.

The honest response is that both things can be true. Miner capitulation can be a decent contrarian indicator for timing while 'production cost is a price floor' remains a bad model of causation. The first is an empirical regularity about supply exhaustion. The second is a claim about a mechanism that does not exist. This desk holds the first loosely and rejects the second.

What would falsify this

A clear test: if the production-cost floor were real, bitcoin should not be able to sustain a price 19% below it for months. It has. If the miner-capitulation-as-bottom thesis is right, we should see hashrate stabilise and then recover while price holds — a hash-ribbon-style crossover — followed by a durable bid. Watch three things over the next month: whether hashrate stops falling; whether miner treasury balances stop declining; and whether hashprice recovers off the $29 level. If hashrate keeps sliding and treasuries keep draining while price stalls at $64,000, the flush is not finished.

In the meantime, treat 'bitcoin is below production cost' as a description of miner pain, not as a reason it must go up. The market does not owe miners their cost basis.

Why this cycle broke the model when previous ones did not

The production-cost heuristic did appear to work in 2018 and 2022, and it is worth asking why it has stopped working now rather than simply declaring it never did. Three structural changes explain most of it.

First, the April 2024 halving cut issuance in half without cutting costs. Post-halving, a miner earns half the bitcoin for the same electricity. That doubled the cost per coin overnight and permanently widened the gap between the cash cost of hashing and the accounting cost of a produced coin. The two numbers used to be closer together; now they are not, and the 'floor' was always the cash cost, not the accounting cost.

Second, miners are no longer forced sellers in the way they once were. Public miners have access to equity markets, convertible debt and bitcoin-backed credit lines. Several hold large treasuries deliberately. A miner that can finance operations without selling production does not transmit its distress into spot supply on the old timetable — which means the 'miners sell, price bottoms, miners stop selling' rhythm that produced the historical pattern is muted and delayed.

Third, and most importantly, the marginal buyer changed. In previous cycles the marginal bitcoin buyer was a crypto-native participant whose behaviour was correlated with miner behaviour. Today the marginal buyer is an ETF allocator, a basis-trade desk, or a treasury company, and none of them look at hashprice. When roughly $8.26 billion left the ETF complex over eight weeks, no amount of miner shutdown was going to offset it. Supply-side models were always going to lose to demand-side flows in a market where flows became this large.

What this means for how you read miner headlines

Expect a steady stream of coverage over the coming weeks framing miner stress as bullish — hash ribbons, capitulation, 'the bottom is in.' Some of it will be reasonable. The specific claim to reject is the causal one: that bitcoin cannot stay below production cost, or that it must revert to it. It can, it has, and there is no mechanism that forces it back. The market clears where buyers and sellers meet, and a miner's spreadsheet is not a bid.

The second claim to treat carefully is the pivot narrative. When a miner announces it is redirecting capacity to AI or high-performance computing, that is frequently reported as a positive — a smart diversification. For the bitcoin network it is the opposite of a bullish signal: it means an operator has concluded that selling compute to a data-centre tenant beats selling hashrate to Bitcoin at current hashprice. Enough of those decisions and network security economics change materially. It may well be the right call for the company. It is not a vote for the asset.