Bitcoin ETFs reverse midweek withdrawals with $433 million Friday inflow

U.S. spot bitcoin exchange-traded products recorded $433.0 million of net inflows on Friday, September 18, reversing the direction of the prior two sessions and leaving the five-session week barely positive. The result mattered at the September 20 cutoff because it showed that the regulated U.S. demand channel had reopened quickly after heavy midweek withdrawals—but the weekly total was too small to establish a durable allocation trend.

This is retrospective coverage for Coinburn’s September 20 open slot and is being published later. The analysis uses only information available by September 20; subsequent flows and market moves are excluded.

Fidelity supplied most of Friday’s reversal

Farside Investors’ U.S. bitcoin ETF table reports $433.0 million of aggregate net inflows for the September 18 trading session. Fidelity’s FBTC accounted for $310.7 million, or 71.8% of that total when calculated from Farside’s rounded figures. BlackRock’s IBIT supplied another $108.4 million, or 25.0%. Bitwise’s BITB, Ark and 21Shares’ ARKB, and VanEck’s HODL together contributed the remaining $13.9 million; the other products in the table were reported at zero for the session.

The one-day concentration is important. Friday’s headline was a category-wide inflow, but nearly three quarters came through a single product. It therefore says more about a sharp change in that day’s primary-market activity than about uniformly stronger demand across every issuer.

The weekly path was much less dramatic than Friday alone. Farside reports $159.9 million of inflows on Monday, $450.4 million of outflows on Tuesday, $295.9 million of outflows on Wednesday, $159.5 million of inflows on Thursday and $433.0 million of inflows on Friday. Summing those rounded daily observations produces a net weekly inflow of $6.1 million for September 14–18. The Block, using SoSoValue data, reported $6.2 million for the same week and $433 million for Friday. The $0.1 million weekly difference is consistent with provider or rounding variation and is not economically meaningful.

What the flow number does—and does not—measure

ETF flow estimates track net creation and redemption activity, not exchange trading volume. They should not be read as a timestamped record showing that retail or institutional investors bought exactly $433.0 million of bitcoin on spot exchanges during Friday’s session.

Fidelity’s latest quarterly filing available at the cutoff explains the underlying mechanism. FBTC issues and redeems shares only through authorized participants, in baskets of 25,000 shares, in exchange for bitcoin or cash. The filing establishes how creations and redemptions change the trust’s share count and bitcoin holdings; it does not independently verify Farside’s September 18 daily estimate.

Fidelity’s product disclosure adds another boundary: FBTC passively seeks to track bitcoin, holds bitcoin, and strikes net asset value at 4 p.m. Eastern on weekdays. Its shares trade during traditional market hours while bitcoin trades continuously, and fees or operational delays can create tracking differences. That makes an ETF-flow print a useful measure of activity in one regulated access channel, not a complete measure of global bitcoin demand.

A reversal, not yet a trend

Friday’s inflow more than offset the combined $746.3 million withdrawn on Tuesday and Wednesday only after Thursday had already brought in $159.5 million. The week’s $6.1 million net result was effectively flat relative to the much larger daily swings.

The defensible conclusion at the September 20 cutoff was narrow: U.S. spot bitcoin products recovered from two heavy outflow sessions, and Fidelity dominated the final-day rebound. The data did not prove why allocations changed, whether buyers would return in the next session, or whether the shift caused any move in bitcoin’s price. Those questions required evidence after the assignment cutoff and are outside this recovery report.