Bitcoin Supply Shock: ETFs Absorbed 4.2x New Supply
How much bitcoin did ETFs buy against new supply?
Over the twenty reporting sessions from July 28 through August 24, US spot bitcoin ETFs took in 38,016 BTC of net demand against 9,000 BTC of new issuance across the same sessions — an absorption ratio of 4.22x. That is the sharpest bitcoin supply shock reading of the last three and a half months, and it arrives after a year in which the same funds were, on balance, handing coins back to the market rather than taking them off it.
The arithmetic is deliberately simple. Each session's net ETF flow is converted to BTC at that day's close, then divided by the 450 BTC the network mints per day. Above 1.0, the funds consumed more bitcoin than miners created. Below zero, redemptions returned it. bryptoflows computes the series from Farside Investors flow data and Hyperliquid closes; the chart lives on the institutional signal dashboard and the underlying flows on the ETF flow dashboard. Every figure here is drawn from that dataset, current through the August 24 session.
The bitcoin supply shock reading is back to its May highs
Context is what makes a ratio mean anything. The site's rolling 20-session series has 653 readings going back to January 2024, and the current 4.22x sits above 97.5% of every reading recorded in 2026 and at the 73rd percentile of the full history. The last time the rolling window printed this high was May 7, at 4.32x.
The 2026 baseline it is measured against is genuinely weak. The median rolling 20-session ratio this year is -0.44x — through a typical month of 2026 the funds handed back roughly 0.44 coins for every one the network mined. The year's low came on June 10, when the trailing window read -8.62x: over the twenty sessions from May 13 to June 10, the category shed 77,543 BTC while the network mined 9,000. That is the reference point for how quickly this metric moves in both directions.
The climb has been fast. The rolling reading was 0.74x on August 14. Six sessions later it is 4.22x. The path runs 0.87, 0.86, 1.57, 2.88, 3.73, 4.22 — no single outlier session doing the work, but six consecutive positive sessions compounding. On the last session alone, August 24, +$337.6M of net flow converted to 4,273 BTC at the $78,996 close, or 9.50x that day's issuance. Across the last five sessions the ratio is 11.95x: 26,889 BTC absorbed against 2,250 BTC mined.
Month to date, August stands at 5.40x — 38,914 BTC absorbed against 7,200 issued over 16 sessions. Set against the rest of the year, that is the anomaly, not the trend:
- 2026 year to date: -41,949 BTC absorbed against 72,450 issued, a ratio of -0.58x
- 2025 full year: +177,472 BTC against 115,650 issued, 1.53x
- 2024 from launch: +507,926 BTC against 146,700 issued, 3.46x
Read down that list and the story is a demand channel that has been decaying since inception, interrupted by a month that looks like the first two years again. Thirteen of the last twenty sessions cleared 1.0x; six were negative. This is a strong stretch inside a weak year, not a weak stretch inside a strong one.
Bitcoin daily issuance is 450 BTC, and nothing about that number negotiates
The reason this ratio is worth computing at all is the asymmetry between its two halves. ETF demand is a market variable — it responds to price, to macro, to allocator mandates, to whatever moved this week. Bitcoin daily issuance is a protocol constant: 3.125 BTC of block subsidy times roughly 144 blocks a day equals 450 BTC, and it will stay 450 until the next halving cuts it to 225.
That denominator does not care what miners are going through, which matters right now because miners are going through a lot. Mining difficulty fell 10.09% on June 14, from 138.9 trillion to 124.9 trillion — the second-largest downward adjustment of 2026 — after a roughly 15% June price slide squeezed margins and pushed hashprice below $30 per petahash per second, per Yahoo Finance's coverage of the retarget. By August 1, CoinDesk reported difficulty running about 14% below the year's high as plunging revenues forced operators to pivot, much of that capacity redirected toward AI data-center contracts.
None of that changes the supply side of the equation. Difficulty adjustments exist precisely to hold issuance at target regardless of how much hashrate is attached; capacity leaving the network makes the remaining miners more profitable per unit of work, not the network less productive. Whether hashrate is 900 exahash or 1,100, the network still mints 450 BTC a day. That is the whole point of the halving supply schedule, and it is why the numerator is the only side of the ratio that can produce a surprise.
ETF holdings now equal 3.9 years of issuance
The cumulative figure puts the flow numbers in proportion. Since spot ETFs launched in January 2024, net creations converted at each day's close total 643,449 BTC. At the current issuance rate that is 1,430 days of production — a shade under 3.9 years — and 3.06% of bitcoin's 21 million hard cap, accumulated in thirty-one months.
Across the whole life of the category, absorbed against issued, the ratio is 1.92x. The ETF complex has consumed a little under two coins for every one mined since it opened. That is the durable version of the bitcoin scarcity argument, and it is more useful than any single week's reading because it does not depend on which twenty sessions you happen to be standing in.
It also frames how much the 2026 slippage actually cost. This year's -41,949 BTC of net redemptions gave back about 6% of the position built by the end of 2025 — real, but a dent rather than an unwind. The structural holding built in 2024 and 2025 has not been meaningfully liquidated; it has stopped growing and then resumed.
A rising price quietly weakens the absorption number
One honest correction belongs on this week's reading. The absorption ratio is measured in coins, but the flow arrives in dollars, and the price moved hard inside the window. Bitcoin closed at $63,906 on July 28 and $78,996 on August 24 — up 23.6% across the twenty sessions.
The $2,685.6M of net flow in that window bought 38,016 BTC, an effective average execution price of $70,645. Had the same dollars arrived at the window's opening price, they would have taken 42,024 BTC off the market — a ratio of 4.67x rather than 4.22x. The rally cost roughly 4,000 BTC of absorption.
That cuts both ways as an interpretation. It means the current reading understates how much money showed up relative to a flat-price week. It also means absorption in coin terms is self-limiting: the more successfully ETF demand bids the price up, the fewer coins each subsequent dollar removes. Anyone extrapolating a rising absorption ratio into a mechanical price path is arguing against the metric's own arithmetic. This is a demand proxy. It is not a forecast, and the relationship between it and price runs in both directions.
What the ratio does not count
The metric is narrow on purpose, and the narrowness is worth stating plainly.
- It counts US spot ETF flow only. OTC desks, exchange balance changes, sovereign holders, offshore funds and private accumulation are all outside it. Real demand for bitcoin is larger than this number, in an amount nobody can measure precisely.
- A negative reading is redemptions, not overproduction. When the ratio goes below zero it means ETF investors sold and baskets were destroyed. It does not mean miners out-produced the world's buyers.
- The denominator uses reporting sessions, not calendar days. The network mined every day of the window, weekends included. Those twenty sessions span 28 calendar days and 12,600 BTC of actual issuance; measured that way, the same demand covers 3.02x rather than 4.22x. The session-based convention is the right one for comparing like with like across the series, but the calendar figure is the one to quote if the question is literally how much of the month's new supply was consumed.
- Issuance is not the same as available supply. Roughly 20 million bitcoin already exist and any of them can be sold. Miners under the revenue pressure described above can and do liquidate treasury holdings, adding sell-side supply the denominator never sees.
Strategy is the other absorber, and its dataset has gone quiet
Corporate treasuries are the second channel that removes coins faster than miners create them, and Strategy is still the whole of it at scale. Across its 2026 filings through August 3, holdings moved from 672,497 BTC at the end of December to 842,138 BTC — roughly 169,600 BTC added, against about 97,650 BTC of issuance over the same span. One company absorbed roughly 1.74x new supply in a year when the entire ETF complex was net negative.
The direction has turned, though. Of its 29 filings captured this year, 21 were purchases, five reported no change, and three were sales — including both of the most recent moves, -3,588 BTC on July 6 and -1,638 BTC on August 3. bryptoflows' SEC filing dataset carries no newer filing than August 3 and is flagged delayed, so nothing here describes the last three weeks. Thursday's treasury piece is where that thread gets picked up properly.
What would confirm or break this reading
The rolling window is the thing to watch, not the daily prints. Because it is a twenty-session average, the strong August sessions roll out of it over the next month; the reading decays toward zero on its own unless new positive sessions keep arriving. A flat fortnight from here drags the ratio back under 2x without a single redemption day.
Two specific tests. First, whether the six-session positive streak survives contact with a down week — 2026 has produced several inflow bursts that reversed inside ten sessions, as the weekly recap of the best flow week of the year and the outflow week before it both show. Second, whether breadth holds: Monday's flow analysis found 69% of last week's net came from a single fund, which makes the absorption figure a bet on one allocator base rather than a broad one. A supply-shock reading built on one issuer is a thinner thing than the same number spread across twelve.
The metric is a clean measurement of one channel. It is currently saying that channel is taking coins off the market at four times the rate they are being created, for the first time since spring. That is worth knowing precisely, and worth not over-reading.