The floor held, the queue behind it quadrupled

20 September 2026 · Block 967,820 · Dubai

Every number in this brief was generated from our own Bitcoin node — bitcoind, Fulcrum, and mempool running on hardware we control. No third-party data vendors. You can verify all of it against your own node.

The Week in Numbers

Metric	Value
Price	$80,451 · AED 295,456 (+4.8% w/w)
Blocks mined (7d)	1,004 (−48 w/w)
Fee to confirm next block	2 sat/vB (economy: 1) — unchanged
Weekly median block fee	1 sat/vB (fourth week)
Mempool backlog	81,831 tx · 41.3 vMB (+142% by count · +272% by weight w/w)
Hashrate	915 EH/s (−0.6% vs 30-day avg · −2.9% w/w)
Difficulty	6.9% through epoch · next est. −7.27% around Sun 4 Oct
Miner revenue (7d)	3,169.0 BTC — 99.40% subsidy, 0.60% fees

---

Fee & Mempool Read: the floor held, the queue behind it quadrupled

The headline number did not move. Next-block is still 2 sat/vB, economy is still 1, and the median fee across all 1,004 blocks was 1 sat/vB for the fourth week running. Read only that line and nothing happened.

Look behind it and the picture is different. The mempool went from 11.1 vMB to 41.3 vMB — 81,831 transactions, up 142% by count and 272% by weight — and the backlog is now roughly 41 blocks deep against 11 last week. The queue stopped being a formality. At the same time the spikiest block of the week printed a 90th-percentile fee of 444 sat/vB; last week's worst was 56, the week before 12. Three weeks, three orders of magnitude. And the 90th percentile of a typical block edged up from 2 to 3 sat/vB.

Here is the part that matters, and it is the reason the headline fee did not move. A queue that grew 3.7x by weight carries only 2.4x the fees — 0.0912 BTC total, about AED 26,900 to settle the entire backlog. The new arrivals are paying the floor. That does not push the price of the next block up, because the top of the mempool is still cheap. It does mean the bottom of it is now crowded, and 41 blocks of competition sit between a 1 sat/vB transaction and confirmation.

What to do with this: stop paying 1. For three issues the economy tier was free money — the queue was short enough that the floor confirmed in a block or two. It isn't now. One sat/vB this week is a lottery ticket behind 41 blocks of identical tickets; 2 sat/vB is still, in dirham terms, nothing. If you are consolidating UTXOs, sweeping an old wallet, or moving coins off an exchange, the window is still open and it is still cheap — but set the fee yourself and set it above the floor. We said last week that a narrowing window cannot be waited on. This is what narrowing looks like: not a higher price, a longer line.

(How to consolidate safely — including the privacy trade-offs of merging UTXOs — is still on the list for a future issue.)

Mining & Security: the first projected cut

Last Saturday's adjustment came in at +4.16%. We had projected +4.75% and told you to expect something between +4% and +5% — close enough, and the third consecutive increase.

The fourth will not be. Blocks have averaged 12m 21s in the new epoch against 9m 33s at our last reading, and the projection is −7.27% at block 969,696, around 17:50 Dubai time on Sunday 4 October. That would be the first difficulty cut since we started counting. The honest caveat is louder than usual this week: we are only 6.9% through the epoch — 139 blocks in, 1,876 to go — so this figure carries roughly ten times the noise of last week's 57%-through reading. Treat −7.27% as a direction, not a decimal. The direction itself is well supported: hashrate printed 915.0 EH/s, down 2.9% on the week and 0.6% below its 30-day average of 920.1 EH/s. That is the first weekly fall after three consecutive rises, and 1,004 blocks mined against last week's 1,052 says the same thing.

Last week we asked what hashrate would do if price kept sliding while difficulty kept climbing. The answer arrived from the other side: price went up 4.8% and hashrate came off anyway. Machines left while revenue per coin improved, which points at the +4.16% adjustment and the cost side — power, hosting, efficiency — rather than at the BTC price. Fees remain no help: 18.97 BTC of the 3,169.0 BTC earned this week, 0.60% of miner revenue, statistically the same 0.6% we have reported every week. The security budget is still the 3.125 BTC subsidy, and a −7.27% difficulty cut is the network doing exactly what it is designed to do — making the next epoch cheaper for whoever stayed.

Pool concentration, the uncomfortable chart nobody looks at: Foundry USA (25.6%), AntPool (18.9%) and F2Pool (16.1%) — 60.6% of blocks between three coordinators. Our readings have run 58.5%, then 58.4%, now 60.6%: first time above 60. The mover is F2Pool, up 2.3 points in a week while AntPool gave back 0.4 and Foundry added 0.3. Below them ViaBTC (8.7%) and SpiderPool (8.4%) both slipped. For three weeks we noted that the top three rotate while the total holds; this week the total moved, in the wrong direction. Same caveat as always: pool ≠ owner of the hashrate, and miners can switch — but block-template control sits with three parties today, and slightly more of it than last week.

The UAE Angle

At AED 295,456 per coin, an AED 1,000 monthly DCA buys ~338k sats this week, down from ~355k. Three issues, three different answers — 341k, 355k, 338k — off a 3.7% rise, a 4.0% fall and now a 4.8% rise. You still do not get to pick the week, and that is still the point.

The operational change is on the withdrawal side, and it is worth being specific about because it costs UAE holders real money in a way the fee chart hides. Many exchanges withdraw at an "economy" or batched rate to save on their own costs — which is fine when the queue is 11 blocks deep and worthless when it is 41. If your exchange lets you choose the fee, choose it this week. At 2 sat/vB a typical single-input self-custody withdrawal (~150 vB) costs about 300 sats — under a dirham, fourth week running. The difference between the fast lane and the slow one is a few fils and a few days.

And a few fils is the whole comparison. The line that costs you on a UAE exchange statement still is not the network fee. Across 2,270 fills we measured, buying directly in BTC/AED cost +63 bps over reference against +19 bps via USDT (100 bps = 1%). On an AED 10,000 buy that 44 bps gap is AED 44 — about fifty times the cost of the withdrawal you were worrying about. Pay above the floor to get your coins out; fix the rail you buy through, because that is where the money goes.

One date for the diary: the difficulty adjustment lands around Sunday evening Dubai time, 4 October, and is currently projected as a cut. It changes nothing for a holder. If you run or host machines in the region, −7.27% after +1.31% and +4.16% would be the first relief in a month — but check it again nearer the date, because at 6.9% through the epoch that number will move.

Go deeper

The network fee is still under a dirham and the queue behind it just quadrupled — but even a bad fee week costs you fifty times less than the spread on the rail you buy through, and we measured that. The Real Cost of Buying Bitcoin in the UAE — 2,270 measured fills, direct BTC/AED +63 bps vs +19 bps via USDT. Free excerpt: https://bitcoiners.ae/intel/uae-rails-report · Full PDF, AED 149: https://buy.polar.sh/polar_cl_fXKk8gcM8YVJQ2tG3HXcAnp6cQnLGNJ5Vhhyc0djHk3

---

Bitcoiners.ae Intel is educational and informational content only. Nothing here is investment, financial, tax, or legal advice. Do your own research.

Generated from our own node at block 967,820 · 20 Sep 2026, 16:00 GST.