Difficulty Update: Tonight’s Retarget Is Projected at +1.20% After a Projection That Went +1.42 → +0.80 → +0.23 → +1.47 → +1.20 in Five Mornings — Marker D2 Fails and F1 Passes Unless the Last 90 Blocks Average Over Twelve Minutes; Hashrate 987 EH/s on Thursday and 913 on Friday, Fees at 1 sat/vB

Bitcoin’s 2,016-block difficulty period ends at height 965,664, and at 06:10 UTC on Saturday the chain tip was 965,574: 90 blocks to go, 95.54% of the period complete, and mempool.space projecting the adjustment at +1.200%, arriving at about 21:00 UTC tonight. The period has averaged 593.19 seconds per block against the 600-second target, which is the whole reason the number is positive: blocks have come 1.1% fast. If the projection holds, difficulty goes from 125.807 trillion, where the −1.312% adjustment of 23 August (00:48 UTC, block 963,648) left it, to about 127.32 trillion.

This update exists because two markers were set on the result. D2, from 30 August, asked for the retarget to land between −1.50% and 0.00%; F1 asked for zero or better. On the morning of 1 September the projection was +1.415%; it then fell to +0.796% and +0.234% over two mornings, at which point D2 was failing by less than a quarter of a point, before Thursday’s fast blocks pushed it back to +1.468% and Saturday’s to +1.200%. A projection that moves by more than a point in a day when 200 blocks remain is normal, which is the lesson of the difficulty guide; the question now is whether 90 blocks can move it by 1.2 points, and the arithmetic says: only with a very slow finish.

What the last 90 blocks would have to do

The adjustment is set by how long the 2,016 blocks took against the 1,209,600 seconds they were supposed to take. With 1,926 blocks done at an average of 593.19 seconds, the period has used roughly 1,142,500 seconds. If the remaining 90 blocks come at the same pace the period totals about 1,195,900 seconds and the adjustment is about +1.15% — close to mempool’s +1.200%, which uses a slightly different estimator, and we print both. For the adjustment to be zero or below, which is what D2 needs, the remaining 90 blocks would have to consume the full 67,100 seconds left in the target: about 12.4 minutes per block, roughly a quarter slower than the period’s 9.9-minute pace, sustained across all 90. Block times are random enough that 90-block stretches of that kind do occur; with hashrate at its current level it would take either a sharp drop in hashrate this afternoon or an unusual run of variance. We do not say it cannot happen. We say D2 is failing by 1.2 points with 90 blocks left, and that tomorrow’s markers piece will print the realised number beside this paragraph.

mempool.space difficulty-adjustment endpoint, pulled each morning at 06:10 UTC by Bitcoin Mastery; blocks-remaining for 1–3 September are approximate (from the progress percentage logged those mornings), 4 and 5 September exact.

Hashrate: 987 EH/s on Thursday, the highest daily reading since 30 August; 913 on Friday

mempool.space’s daily hashrate estimates, one row per UTC day: Monday 31 August 921.44 EH/s, Tuesday 901.73, Wednesday 853.78, Thursday 3 September 987.32, Friday 913.44. (A note on the rows: mempool stamps each daily estimate with the timestamp at which the day closes, so the row dated 4 September is Thursday’s; we established that because the value was already 987.32 at our 06:10 UTC pull on Friday morning and did not change through Saturday, and we have relabelled Thursday’s log accordingly.) Thursday’s 987 is the highest daily reading since Sunday 30 August’s 1,053.07 and the reason the projection recovered from +0.234% on Thursday morning to +1.468% on Friday morning: a fast day near the end of a period moves the average more than a fast day at the start. Daily hashrate estimates are noisy by construction — they are inferred from block arrival times, and a day with 155 blocks instead of 144 reads as a hashrate surge whether or not any machine was switched on — so the number to carry is the trailing average: the mean of the last thirty daily rows (6 August to 4 September) is 905.77 EH/s, the three-month mean is 912.73 EH/s, and mempool’s own current estimate from recent blocks is 936.32 EH/s, all against a three-month daily high of 1,100.45 EH/s on 28 June. The trailing month sits about 18% below that single-day June high, and the daily series has been flat since the August cut.

The adjustment history frames tonight’s number. The last six retargets: −10.09% (14 June), +7.15% (27 June), −5.00% (11 July), −0.74% (25 July, the smallest cut of the year to that date, per our 27 July update), +0.99% (8 August), −1.31% (23 August). A +1.2% tonight would be the second increase in three retargets and would leave difficulty 0.13% below where it stood after the 8 August adjustment (127.48 trillion). In other words, the network has spent two months oscillating within about 2% of 127 trillion after the June swing, and tonight’s adjustment keeps it there.

mempool.space mining API, difficulty series, pulled 5 September 2026 06:10 UTC. Adjustment percentages are the ratio of consecutive difficulties, computed here.

Who mined the week, and what the blocks paid

Over the past seven days the network produced 1,038 blocks — 30 more than the 1,008 a 600-second pace would give, which is the same fact as the positive projection stated a different way. Pool shares per mempool.space: Foundry USA 284 blocks (27.4%), AntPool 178 (17.1%), F2Pool 148 (14.3%), SpiderPool 84 (8.1%), ViaBTC 73 (7.0%), SECPOOL 60 (5.8%); the top two hold 44.5% and the top three 58.8% of the week. The fee market has nothing in it: at 06:10 UTC the recommended fee for next-block inclusion was 2 sat/vB and the half-hour, hour and economy tiers were all 1 sat/vB, the minimum. Miner revenue is therefore almost entirely the 3.125 BTC subsidy, which at Friday’s $79,676.60 close is about $249,000 per block and, at 1,038 blocks a week, about $258 million a week across the network before fees. A 1.2% difficulty increase at a flat price and an empty fee market is a 1.2% cut in revenue per unit of hashrate, which is how the retarget transmits to the miners who are already marginal; the hashprice piece has the framework.

Price context, and the marker

Bitcoin closed Friday at $79,676.60 on Bitstamp, down 1.955% from Thursday’s $81,265.00, after the August payrolls report came in at three times consensus and pushed September hike odds back above 50%; today’s lead has the day. For miners the relevant figure is the price at which the subsidy is sold, and a 1.2% harder network at a price 2% lower than Thursday is the combination that ends periods of hashrate growth; the trailing series above says growth has already been flat for a month. The next retarget after tonight’s is due around 19 September, at block 967,680, and its projection will begin printing tomorrow morning from a handful of blocks; we will not quote it until at least a quarter of the period is in.

As standing practice we mark one falsifiable claim on this. D2 and F1 grade tonight and we do not touch them. P1: the retarget at block 967,680, due around 19 September 2026, is a positive adjustment (> 0.00%) per mempool.space’s difficulty series. The reasoning is that the past week’s 1,038 blocks and Friday’s 987 EH/s daily reading are consistent with hashrate at or slightly above the level tonight’s adjustment will target, so a second small increase is the base case; a fall in price through the mid-$70,000s, or a retreat in the daily series toward the 850s, would make it a cut. If it grades as a cut we will have overweighted one fast Thursday. We set it before the first block of the period.

Method: prices, funding, open interest, basis and on-chain figures in this article are pulled directly by Bitcoin Mastery at the timestamp stated — Bitstamp BTC/USD daily candles for closes, Binance BTCUSDT spot and USDT-margined perpetual for intraday, open interest, funding and account ratios, Binance COIN-M quarterly contracts for basis, mempool.space for difficulty, hashrate, pool shares and fees, alternative.me for the Fear & Greed series, Farside Investors’ table for ETF flows (every named fund column is read from a print of that column with its header) and US Treasury CMT par yields, nominal and real, for rates. Where a third-party figure is cited we name the source and its date; where two sources disagree we print both. Every streak or extreme figure is published with the first date of its series in the same sentence.

Disclaimer: This article is for informational purposes only and does not constitute investment advice. Cryptocurrencies are volatile and you can lose money. Nothing here is a recommendation to buy or sell any security, digital asset or exchange-traded fund, including MSTR. Do your own research and consult a licensed financial advisor before making investment decisions.