Bitcoin miners face revenue squeeze after April 2024 halving

Every four years or so, Bitcoin cuts the pay of the people who secure it, on a schedule nobody can negotiate with. There is no committee that reviews conditions first, no mechanism to delay the cut if margins are thin. The subsidy halves at a fixed block height and miners either absorb it or leave. Two years on from the April 2024 halving, with hash price near multi-year lows and a fee market that has effectively vanished, the industry is living through what that arithmetic looks like when the price cycle does not arrive to rescue it.

The subsidy is a schedule, not a policy

Bitcoin’s issuance is written into consensus rules as a step function. Every 210,000 blocks, roughly four years at a ten-minute average block time, the reward paid to whoever mines a block is cut in half. The fourth such cut landed at block height 840,000 on April 19, 2024, taking the subsidy from 6.25 BTC to 3.125 BTC per block. The next is expected at block 1,050,000, currently projected for around April 17, 2028, which will take it to 1.5625 BTC.

What makes this economically distinctive is that it is a shock to revenue with no corresponding change on the cost side. A miner’s electricity bill, hosting contract, and hardware depreciation do not halve alongside the subsidy. Historically, rising bitcoin prices have papered over that gap within months. That is a pattern, not a rule, and the last two years have demonstrated the difference.

Hash price is the number that actually matters

Miners rarely think in terms of the subsidy directly. The working metric is hash price: total network revenue, subsidy plus fees, divided by total hashrate, expressed as dollars per petahash per second per day. It collapses bitcoin’s price, the subsidy level, network difficulty, and fee income into a single figure that tells an operator what one unit of computing power earns.

That number has been moving in one direction. When difficulty adjusted downward in mid-June 2026, hash price stood at $32.31 per PH/s per day, and that was an improvement on the high-$20s levels seen earlier that month. CoinShares’ Q1 2026 mining report recorded hash price falling below $30 per PH/s per day, a five-year low, and estimated that 15 to 20 percent of legacy rigs were running at an outright loss at those levels. The report’s conclusion was blunt about the sorting mechanism at work: mid-generation hardware needs access to power below five cents per kilowatt-hour to stay cash-profitable, while newer sub-15 J/TH fleets retain meaningful margin at ordinary industrial rates.

Difficulty adjustment is the release valve

The protocol does have one self-correcting mechanism, and it is the reason Bitcoin mining does not simply spiral into collapse when revenue falls. Every 2016 blocks, approximately two weeks, the network measures how long those blocks actually took and retargets difficulty so the next 2016 arrive at the ten-minute average. Consensus rules cap any single retarget at a fourfold increase or a 75 percent decrease, a guard against a single corrupted data point breaking the network in one step.

In practice this means unprofitable miners switching off makes the remaining miners more profitable. On the weekend of June 14, 2026, at block height 953,568, difficulty fell 10.09 percent, from 138.96 trillion to 124.93 trillion. It was the eleventh-largest downward adjustment in Bitcoin’s history and, at the time, the second-largest of 2026. The mechanical effect was that surviving hashrate earned roughly 11 percent more bitcoin per unit of computing power overnight. The scale of the capitulation behind that adjustment is visible in the hashrate series itself: from a peak above 1.3 zettahashes per second in October 2025, the network fell about a third to the 860 to 900 EH/s range that has prevailed through mid-2026, with difficulty hovering near 126 trillion in August, close to its low for the year.

The fee market did not arrive

The long-standing theory of Bitcoin’s security budget is that as the subsidy shrinks toward zero, transaction fees take over as miner compensation. That transition is not happening on the timeline the theory needs. Fees accounted for just 0.69 percent of miner revenue in August 2026, after touching a decade low of 0.52 percent in April, and had stayed under one percent for nearly a year. Glassnode co-founder Rafael Schultze-Kraft observed that the last time fee share was this depressed, bitcoin traded below $400.

The practical consequence is that miners are more dependent on the fixed subsidy than at any point in the past decade, precisely when that subsidy has been cut and cut again. Whether the fee market develops meaningfully before the 2028 and 2032 halvings is arguably the most consequential open question in Bitcoin’s long-term security model, and there is currently very little evidence that it is developing.

The squeeze, measured

Put the pieces together and the arithmetic becomes unforgiving. CoinShares put the weighted-average cash cost of producing one bitcoin among listed miners at roughly US$79,995 in the fourth quarter of 2025. By August 2026, the estimated average production cost stood at $78,254, roughly 23 percent above the spot price that prevailed through the first three weeks of the month. Checkonchain’s model put the figure at $84,300 in June, when bitcoin traded near $63,780 after a roughly 15 percent monthly decline. For a large share of the network, mining spent most of 2026 producing an asset worth less than it costs to make. The rally that took bitcoin above $77,000 on August 21 closes that gap rather than reversing it: at a spot price hovering around the $78,254 cash cost, the average listed miner is at break-even before depreciation, interest and hosting escalators are counted, and one difficulty retarget in the wrong direction puts it back underwater.

That gap explains the industry’s strategic reorientation better than any narrative about disillusionment with Bitcoin. Miners hold the two things AI infrastructure buyers cannot quickly acquire, namely secured grid capacity and permitted sites. Miners have signed GPU co-location and hyperscaler deals with an aggregate value reported above US$70 billion, at operating margins far above what hashing currently delivers.

What it means for a Canadian reader

Canada is unusually exposed to this shift, because a disproportionate share of the listed mining sector was built here on cheap hydroelectricity and cold climates. The clearest case is Bitfarms, long one of Quebec’s largest hashers. It announced in late 2025 that it would wind down bitcoin mining, told investors in February 2026 that it was “no longer a Bitcoin company,” and completed its rebrand to Keel Infrastructure alongside a redomiciliation to the United States, trading on both the Nasdaq and the TSX under KEEL. Hut 8 has been building out AI infrastructure alongside its ASIC compute business, and Vancouver-based HIVE Digital runs its BUZZ HPC platform next to its mining fleet.

For a Canadian investor, the implication is that a position in a “bitcoin mining stock” listed in Toronto increasingly is not one. The revenue mix and risk profile of these companies are migrating toward contracted data-centre leasing, which behaves nothing like leveraged exposure to bitcoin’s price. It is worth reading the segment breakdowns in the quarterly filings rather than assuming the ticker still describes the business.