A Record $3.8B Streak With 1 Buyer — IBIT ETF Took 85% of the Only Outflow Day Too

US spot Bitcoin ETFs recorded approximately $986.9 million in net inflows during the week ended September 4, extending the positive-flow streak to three consecutive weeks. The weekly total rose from roughly $924.5 million the previous week — an increase of about 6.7%.

Cumulative inflows across that three-week stretch reached approximately $3.8 billion, the strongest consecutive run these products have posted in 2026.

The first four trading days of September alone delivered roughly $769.9 million. Net assets across the complex closed Friday at $101.25 billion after touching $103.3 billion the day prior. Cumulative net inflows since the January 2024 launch stand at approximately $55.6 billion.

On the surface, that is an unambiguously strong set of numbers. Institutional demand for Bitcoin exposure through regulated vehicles is running at its fastest pace of the year, and it is doing so while the underlying asset trades below its recent high.

The structure underneath is the problem, and it is the reason this article exists rather than a one-line summary.

BlackRock's iShares Bitcoin Trust took approximately $691.5 million of that $986.9 million weekly total. That is 70% of every dollar that entered the entire twelve-fund complex, concentrated in a single product.

Bitcoin itself has not responded proportionally. BTC traded around $78,500 to $79,100 on Tuesday, down 0.81% on the session and roughly 2.5% below the September 4 peak of $81,166. Nearly a billion dollars of regulated buying in one week produced no sustained price advance.

That divergence — record flow, flat price — is the second problem, and the two are connected.

The macro backdrop explains part of it. August payrolls came in at 162,000 against a consensus near 56,000, unemployment held at 4.1%, and the probability of a September 16 rate hike moved toward 58% to 60%. The 10-year Treasury sits at 4.80%.

Institutional buying into a tightening cycle is being met by supply from holders who have decided the rate environment is not worth waiting through.

IBIT Took Seventy Percent of It

The concentration figures deserve to be laid out session by session, because the trend within them is more important than any single number.

On August 19, US spot Bitcoin ETFs took in $517.19 million — their biggest daily haul since May 4. IBIT captured $284.7 million of that, roughly 55% of the total. ARK 21Shares' ARKB added $77.7 million, close to 15%, and Fidelity's FBTC brought in $62.4 million, about 12%.

On September 3, the complex recorded $730.8 million. IBIT took $454.0 million — 62%. ARKB followed at $137.7 million, FBTC at $74.4 million, and Grayscale's Bitcoin Mini Trust at $48.8 million.

On September 4, total net inflows were $174.60 million. IBIT took $117.38 million, or 67%. FBTC added $57.22 million. Every other US spot Bitcoin ETF recorded zero.

Across the full week, IBIT's $691.5 million against $986.9 million works out to 70%.

Read those four figures in sequence: 55%, 62%, 67%, 70%. IBIT's share of incoming capital has risen with every measurement window.

That pattern reflects a structural reality about how institutions allocate. When institutional demand returns, it concentrates heavily in the largest and most liquid vehicle first. IBIT has become the default gateway — the deepest secondary market, the tightest spreads, the widest availability across brokerage and advisory platforms.

For an allocator writing a ticket, the choice is not really between twelve products. It is between IBIT and not participating.

The bull reading is that this reflects product-market fit rather than fragility. A single dominant vehicle with genuine liquidity is what allows large tickets to trade without moving the market, and its dominance is self-reinforcing in a healthy way.

The bear reading is arithmetic. When one product accounts for 70% of inflows, the entire regulated demand channel for Bitcoin has one point of failure, and there is no second buyer positioned to offset a reallocation out of it.

September 1 demonstrated exactly what that looks like.

September 3: $730.8 Million and the Biggest Day Since January

The standout session of the streak deserves detailed treatment because it is the closest thing to a genuine institutional surge these products have produced this year.

Net inflows hit $730.8 million on September 3, the largest daily total since mid-January 2026 — a gap of roughly eight months. Multiple funds contributed positively, which is unusual given the concentration pattern.

IBIT's $454.0 million was substantially larger than any other product. ARKB recorded $137.7 million. FBTC took $74.4 million. Grayscale's Bitcoin Mini Trust, which trades under the BTC ticker, added $48.8 million.

Those four products together accounted for roughly $715 million of the $730.8 million total, leaving approximately $15 million spread across everything else.

The session coincided with Bitcoin breaking above $80,000 on dovish Federal Reserve signalling that briefly cut September hike odds toward a coin flip. The flow arrived with the price move rather than causing it, which is the normal sequence — ETF creations settle after the fact and reflect the prior session's demand.

Comparison with the August 19 print is instructive. That session delivered $517.19 million, the biggest since May 4, with net assets across all US Bitcoin ETFs reaching $84.31 billion — about 6.08% of Bitcoin's total market capitalisation at the time. Cumulative inflows since launch stood at roughly $52.79 billion.

Between August 19 and September 4, net assets climbed from $84.31 billion to $101.25 billion — a $16.94 billion increase against roughly $2.8 billion of net creations over the same window.

That gap is price appreciation, not flow. Bitcoin's rally through late August did more for the funds' assets under management than every dollar of institutional buying combined.

The distinction matters for anyone using assets under management as a demand proxy. A rising AUM number in a rising market says almost nothing about whether new money is arriving.

Daily flow figures can be revised as late fund reports arrive, so single-session totals should be treated as tracker data rather than settled fact.

September 1: The Same Concentration, Running Backwards

The single outflow session inside the streak is the most informative data point available, and it is being largely ignored.

On September 1, the complex recorded net outflows of approximately $236.5 million. IBIT accounted for $201.2 million of that — roughly 85%. FBTC contributed $43.7 million of redemptions. Bitwise's BITB was the only major product printing positive at $8.4 million, and every other Bitcoin product recorded zero net flow.

Eighty-five percent of the redemption came through one fund, on a day when nine of twelve products did nothing at all.

That is the concentration risk made concrete. When IBIT goes negative, the complex goes negative, because there is no offsetting inflow anywhere else in the category. The other eleven funds are not independent demand channels — they are, on most sessions, inert.

The magnitude is worth noting as well. A $201.2 million redemption is a rounding error against $101.25 billion of net assets, roughly 0.2%. That single ticket flipped the entire twelve-fund category from positive to negative for the day.

Extrapolate that. A large allocator rebalancing $2 billion out of IBIT over a week would produce category-wide outflows that no combination of the remaining products could offset, because those products are not receiving flow to redirect.

That is not a hypothetical risk profile. It is the observed behaviour of this market on the one day in the past three weeks when the largest holder sold.

The counterargument is that September 1 was noise — a single session inside a $3.8 billion positive stretch, immediately reversed by $730.8 million on September 3. Two-way flow inside a strongly positive week is what a functioning market looks like, and allocators trimming and adding is healthier than a single directional buyer pushing price.

Both readings are defensible. What is not defensible is describing this as a broad institutional bid when 70% of the inflow and 85% of the outflow move through one ticker.

$101.25 Billion in Net Assets and Roughly 6.5% of Bitcoin

The aggregate scale of these products has become large enough to matter for Bitcoin's market structure, which is a genuine change from a year ago.

Net assets closed at $101.25 billion on September 4 after touching $103.3 billion the previous day. Trading value across the group reached $2.95 billion on Friday.

At $84.31 billion in mid-August, the complex represented approximately 6.08% of Bitcoin's total market capitalisation. At $101.25 billion against a Bitcoin market capitalisation near $1.56 trillion, that share has risen to roughly 6.5%.

Six and a half percent of the entire asset held inside regulated US vehicles is a structural change in Bitcoin's ownership profile. It means a meaningful and growing share of the float sits with holders who transact through a creation-redemption mechanism rather than on exchanges, and whose behaviour is driven by portfolio allocation decisions rather than by crypto-native sentiment.

That has two consequences. It provides a stability layer — allocators do not panic-sell the way leveraged retail does. And it introduces a correlation to traditional risk assets that Bitcoin previously lacked, because those same allocators rebalance across their whole book when rates move.

Tuesday demonstrated the second effect. Bitcoin fell 0.81% on the same session that gold declined 0.38% to $4,395.51 and the Dow Jones Industrial Average shed 570.78 points, while the 10-year Treasury held 4.80%. Every non-yielding asset sold together.

A separate measurement puts total cohort assets under management near $99.56 billion, down from $101.42 billion a week earlier but roughly $20 billion higher than a month ago. Different trackers use different valuation timestamps, so the readings should not be treated as one synchronised dataset.

The trend across all of them is consistent: assets up sharply over a month, flat to slightly down over a week, driven far more by Bitcoin's price than by creations.

$55.6 Billion Cumulative Against a Year-to-Date Hole

The long-run number is impressive and the year-to-date number is not, and both are true simultaneously.

Cumulative net inflows since the January 2024 launch stand at approximately $55.6 billion. A separate measurement puts the figure at $51.8 billion, with the difference reflecting different treatment of the Grayscale conversion and varying reporting windows.

Either way, more than $50 billion of net new capital has entered Bitcoin through regulated US products in under three years. That is one of the most successful ETF launches in the industry's history by any measure.

Year to date in 2026, however, the cohort remains roughly $1 billion net negative.

Read those two facts together. Essentially all of the cumulative inflow was accumulated in 2024 and 2025. Through the first eight months of 2026, these products have net redeemed, and the past three weeks of $3.8 billion have nearly — but not quite — erased that deficit.

That is the honest framing of the current streak. It is not a new wave of institutional adoption. It is the recovery of ground lost during the first half of the year.

The pattern of daily behaviour confirms it. In 2026, net flows have been negative on 54% of trading sessions, up from 40% in 2025 and 31% in 2024.

That progression is the most important structural datapoint in this entire analysis. In the launch year, roughly two of every three sessions saw net creations. Now, more than half of sessions see net redemptions.

These products have matured from a one-way accumulation vehicle into a genuinely two-sided market. That is normal for any ETF category as it ages, and it is arguably healthy — a fund that only ever receives money is a fund with a concentrated, unhedged holder base.

But it changes what the flow data means. A positive week in 2024 signalled adoption. A positive week in 2026 signals that buyers marginally outnumbered sellers across five sessions, which is a much weaker claim.

August at $3.52 Billion — the Best Month Since September 2025

Monthly data provides the cleanest view of the current cycle, and August was genuinely strong.

Spot Bitcoin ETF inflows reached $3.52 billion in August, the strongest monthly result since September 2025. Bitcoin gained roughly 24% over the same month, its best August since 2017.

An eleven-month high in monthly inflows, coinciding with a 24% price move, describes flow following price rather than leading it. That sequencing is consistent across the entire history of these products — creations arrive after rallies and redemptions arrive after declines.

The composition of August's advance is what complicates the bullish read. Roughly $6.55 billion in short liquidations accumulated across two weeks during the rally, while futures open interest fell to a five-month low and crypto-margined open interest hit a record low.

That is the signature of a squeeze, not of accumulation. Price rose because shorts were forced to cover, and ETF flows followed the price higher.

The July comparison sets the baseline. That month delivered a much smaller net figure, and single-day outflows of roughly $6.40 million were recorded as recently as early August with the major funds all closing red on individual sessions.

The current September pace suggests some continuation. Roughly $769.9 million entered during the first four trading days, which annualises to a monthly figure above $3.5 billion if maintained.

Whether it is maintained depends entirely on price. The historical relationship is tight enough that flow forecasts are really price forecasts wearing different clothes.

IBIT specifically has attracted $3.575 billion over the past thirty days — a figure that exceeds the entire complex's August total, reflecting both its dominance of inflows and the fact that other funds have been net negative over portions of that window.

The Thirteen-Session Streak That Cost $4.37 Billion

The bear case for this category is not theoretical. It happened four months ago.

The longest outflow streak of 2026 ran thirteen consecutive sessions from May 15 to June 3, during which the funds shed $4.37 billion.

That is larger in absolute terms than the current three-week inflow streak of $3.8 billion, and it happened over a shorter period.

The mechanics of that episode are what make it relevant now. Thirteen consecutive negative sessions means the redemption pressure was persistent rather than event-driven, and $4.37 billion across thirteen days averages $336 million per session — comfortably above the current inflow pace.

Given the concentration pattern, the overwhelming majority of that would have moved through IBIT, since it is the product large allocators use for both directions.

The lesson for the current streak is calibration. Three weeks of positive flow totalling $3.8 billion sounds durable until it is set against a two-and-a-half-week stretch earlier in the same year that removed more capital faster.

Nothing in the current flow data distinguishes the present streak from the several prior positive stretches that reversed. The pattern through 2026 has been alternating runs in both directions, with the negative sessions marginally outnumbering the positive ones at 54% to 46%.

Flow persistence in these products has been low. The useful analytical stance is to treat any streak as a snapshot of current conditions rather than as evidence of a regime.

Historical extremes provide the outer bounds. On March 4, a single US Bitcoin fund recorded a $2.91 billion net cash outflow in one day — the largest withdrawal in over a decade for any exchange-traded product.

A single session capable of removing $2.91 billion is a single session capable of erasing three-quarters of the current three-week streak.

That is the risk profile of a category where 70% of the flow runs through one door.

The Nine Funds Printing Zero

The most overlooked detail in the daily data is how many products are doing nothing.

On September 4, of the $174.60 million in net inflows across the entire US spot Bitcoin ETF complex, IBIT accounted for $117.38 million and FBTC for $57.22 million. Those two numbers sum to $174.60 million exactly. Every other US spot Bitcoin ETF recorded precisely zero net flow.

On September 1, the same pattern held in reverse. IBIT redeemed $201.2 million, FBTC $43.7 million, BITB created $8.4 million, and all other Bitcoin products recorded zero.

The tracked universe includes twelve products: IBIT, GBTC, Grayscale's Bitcoin Mini Trust, FBTC, ARKB, BITB, Franklin's EZBC, Invesco Galaxy's BTCO, VanEck's HODL, Valkyrie's BRRR, WisdomTree's BTCW and Hashdex's DEFI.

Nine of those twelve routinely print zero on consecutive sessions.

A fund recording zero net creations or redemptions across a full trading day is not receiving allocations. It is holding a static book, with secondary-market trading occurring between existing shareholders rather than new capital entering.

That reframes what "the Bitcoin ETF complex" actually is. It is not a twelve-fund category competing for institutional dollars. It is one dominant product, a credible second, an occasionally active third and fourth, and eight vehicles that exist primarily as legacy positions or platform-specific alternatives.

September 3 was the exception that proves the point. Four funds recorded meaningful positive flow that day — IBIT, ARKB, FBTC and the Grayscale Mini Trust — and it was the biggest session since January.

Genuine breadth in this category is a once-in-eight-months event.

For anyone modelling future demand, that means the relevant question is not whether institutions are allocating to Bitcoin. It is whether one asset manager's distribution network is putting Bitcoin in front of its clients this quarter.

Flow Against Price: $3.8 Billion and No Breakout

The disconnect between the flow data and the price action is the most important analytical problem in this market right now.

Three consecutive positive weeks totalling roughly $3.8 billion. August inflows of $3.52 billion, the best month since September 2025. IBIT taking $3.575 billion over thirty days.

Bitcoin trading at roughly $78,500 to $79,100, down 0.81% on the session and 2.5% below the September 4 peak of $81,166.

Nearly a billion dollars of net regulated buying in the week ended September 4, and price ended the following session lower.

The flow data matters because it suggests the recent advance has had meaningful spot-market support rather than being driven solely by futures leverage. That is genuinely constructive for the structure of any rally.

But ETF buying has provided a liquidity buffer rather than a driver. It has absorbed supply without generating price appreciation, which means an equivalent volume of selling met it.

Where is that supply coming from? Short-term holders sitting on roughly $9.07 billion in unrealized profits after the August advance from the low $60,000s toward $82,300. That cohort has been distributing into every rally, and $986.9 million of weekly institutional demand has been the exit liquidity.

The macro overlay explains the timing. With the 10-year at 4.80%, the 30-year at 5.27%, Brent crude at $99.22 and a 58% to 60% probability of a September 16 rate hike, the opportunity cost of holding a non-yielding asset has risen materially. Holders who bought the August low have a clear reason to take profit and a Treasury note paying 4.80% to put it in.

That dynamic is not resolved by more ETF flow. It is resolved by the rate outlook changing.

Which puts Friday's inflation print at the centre of this entire analysis.

What Friday's CPI and September 16 Do to the Flow

The flow data over the next fortnight is a function of two scheduled events rather than of anything happening inside the ETF complex.

US Consumer Price Index data for August arrives Friday, September 11 at 08:30 Eastern. The Federal Open Market Committee decides on September 16. Payrolls detail is published at bls.gov and the meeting calendar at federalreserve.gov.

A cool inflation print cuts hike odds from roughly 58% toward the 40s. Yields fall, the opportunity cost of holding Bitcoin drops, and the short-term holder cohort has less reason to sell into institutional bids. Under that scenario the existing $986.9 million weekly pace produces actual price appreciation rather than absorption, and flow accelerates on the follow-through.

A hot print locks the September 16 hike, pushes the 10-year toward 5.00%, and turns the flow arithmetic negative. The May 15 to June 3 precedent — thirteen sessions, $4.37 billion out — is the template for what a sustained rate shock does to this category.

An in-line print extends the current stalemate: positive flow, flat price, and the distribution of profitable August buyers continuing to absorb institutional demand.

The August CPI reading carries one structural limitation. It will not capture this week's energy move, with Brent at $99.22 after strikes on Saudi infrastructure. That lands in the September data, published after the FOMC has already decided.

So a benign Friday buys room next week and guarantees nothing about December.

Cross-asset flow provides a useful control. On the same Friday that Bitcoin ETFs took $174.60 million, Solana products lost $5.21 million and XRP products recorded no net flow at all. Ethereum ETF weekly inflows fell 74% to $218.41 million and XRP's fell 83% to $18.96 million.

Regulated crypto demand is narrowing toward Bitcoin across the board. Within Bitcoin, it is narrowing toward IBIT. Both narrowings are symptoms of the same condition — capital consolidating into the largest and most liquid instrument when the macro picture is uncertain.

Verdict: A Record Streak Running Through a Single Door

The headline numbers are strong and they should be read carefully rather than dismissed. US spot Bitcoin ETFs took approximately $986.9 million in the week ended September 4 against $924.5 million the prior week, extending a three-week streak to roughly $3.8 billion — the strongest consecutive run of 2026 — with $769.9 million arriving in September's first four sessions, net assets at $101.25 billion after touching $103.3 billion, cumulative inflows since launch at $55.6 billion, and August's $3.52 billion the best month since September 2025. The structure underneath is the story. BlackRock's IBIT took $691.5 million of that $986.9 million weekly total, or 70%; $454.0 million of the $730.8 million on September 3, the biggest session since mid-January, or 62%; and $117.38 million of $174.60 million on September 4, or 67% — with every other US spot Bitcoin ETF except FBTC recording exactly zero that day. The concentration works identically in reverse: on September 1, when the complex went negative $236.5 million, IBIT accounted for $201.2 million of it, or 85%, and nine of twelve products again printed zero. That is not a twelve-fund institutional bid. It is one distribution network, and there is no second buyer positioned to offset a reallocation out of it — as the thirteen-session run from May 15 to June 3 demonstrated when $4.37 billion left in under three weeks, more than the entire current streak has brought in. The flow is also not producing price. Bitcoin sits near $78,500, down 0.81% and 2.5% below the $81,166 September 4 peak, because roughly $9.07 billion of short-term-holder unrealized profit has been distributing into every institutional bid with a 4.80% ten-year as the alternative. Year to date the cohort remains roughly $1 billion net negative and has printed negative flow on 54% of 2026 sessions, up from 40% in 2025 and 31% in 2024. What to watch: weekly inflows holding above the $900 million pace confirms the streak; IBIT's share falling back toward the 55% seen on August 19 would signal genuine breadth returning; and any session where IBIT redeems while the other eleven stay at zero is the category's structural weakness showing itself again. Friday's CPI at 08:30 Eastern and the September 16 FOMC decide which of those happens, and nothing inside the ETF data itself will tell you first.

That's TradingNEWS