Bitcoin's Best Week Since 2023 Was a Bet on US Fiscal Policy, Not a Sign of Adoption
Bitcoin gained 23% in seven days after a Treasury decision about the long bond, and the supply of dollars on public blockchains grew 0.26%. The open money read: price has stopped being a scoreboard for open infrastructure, and this is the week you can watch the two come apart.
Summary: Bitcoin spent six weeks stuck between $62,000 and $66,900, broke out on August 20, and reached $79,400 on Friday for its strongest week since March 2023. Ether ran harder, up 28 percent.
The spark for the rally came from the Treasury Department, which on August 19 doubled the maximum size of its long-dated bond buyback operations from $2 billion to at least $4 billion, and the market read it as a signal about how Washington intends to handle its borrowing costs from here.
US spot bitcoin funds took $1.61 billion across four sessions and ether funds another $299 million. Over the same seven days, the supply of dollars sitting on public blockchains grew by $786 million, or 0.26 percent, and it remains below where it stood in May.
Thesis: For fifteen years the argument for this asset class has been that it offers a way out of a financial system run at the discretion of governments and banks. This week bitcoin had its best run in three years because the US Treasury made a decision about its own debt, and the money that did the buying arrived through brokerage accounts and margin desks rather than through anything anyone would call open infrastructure.
Those two facts belong together. The marginal owner of bitcoin in 2026 is someone fully inside the financial system, holding it to express a view about that system, which means the price now moves with Washington rather than away from it. Price stopped being a scoreboard for Open Money, and this is the week you can watch the two come apart in real time.
What actually did the buying
Four things bought bitcoin this week. Only one of them was a person deciding to buy bitcoin.
The largest share came from traders who had bet against it. When the price climbed past the level where their lenders close them out automatically, roughly $4 billion of those positions got bought back over two days, the largest such episode since at least 2021. Nobody in that group wanted to own bitcoin. They were required to.
Next came the exchange-traded funds, which took $1.61 billion across four sessions, with BlackRock's product accounting for $503 million on Thursday alone. Ether funds added $299 million over the same stretch, and cumulative money into US spot bitcoin funds now stands at $53.5 billion. The marginal buyer here opened an app, tapped a ticker, and went back to work. Nothing about the transaction touched a public blockchain until a custodian moved coins between two vaults on their behalf.
Third came borrowed money, as traders who missed the first leg bought the second one on credit.
Fourth, and the only channel that counts as adoption in any sense this newsletter uses the word, the supply of dollars living on public blockchains grew $786 million. Roughly a quarter of one percent, and still about 6 percent below its May peak.
Bitcoin and ether added something like $350 billion in market value over those seven days. Market value is price times supply, so nobody should read that as $350 billion walking in the door. The point is narrower and harder to argue with. Every channel that repriced this market runs through an intermediary that can freeze an account, and the one channel that doesn't barely moved.
The rails were spectators at their own bull market.
The spark was a decision about the long bond
On August 19 the Treasury said it would raise the maximum size of its buyback operations in the 10-to-30 year range from $2 billion to at least $4 billion, running from September 9 through early November. Treasury buys back its own outstanding bonds to keep those markets liquid, and it funds the purchases by issuing short-term bills, so no new money enters the system.
On its own terms this is housekeeping. Lance Roberts called it Operation Twist 2.0, a rerun of a maneuver from the 1960s. ING was blunter: the whole thing is zero-sum and won't change where long-term rates are headed. Treasury has said clearly that this is neither quantitative easing nor an attempt to pin interest rates at a chosen level.
The market traded the implication instead. Mohamed El-Erian put it precisely: the significance sits less in the size of the operation and more in what it suggests might come later, namely a government that has decided its long-term borrowing costs are a problem it intends to manage. Those costs are the highest they've been since 2007. The Federal Reserve is signaling the opposite direction, with three dissents at its July meeting and futures pricing a real chance of a rate rise in September. So the fiscal arm and the monetary arm are pulling against each other, and investors bought the hard-asset hedge against whichever one they think wins.
The Treasury spent Wednesday doing maintenance on its debt schedule. The crypto market read it as a smoke signal.
The exit that trades on the thing it exits
Which leaves an awkward fact sitting in the middle of the week. The asset sold for fifteen years as an exit from government monetary discretion had its best week in three because a government exercised some.
Bitcoin traded like a leveraged position on one policy path inside the dollar system, and its price is now a readable function of which arm of Washington is pulling harder. That's a respectable thing for an asset to be. It's also close to the opposite of what a lot of the people buying it believe they own.
The reason is worth sitting with, because it isn't a contradiction so much as a consequence of who owns the thing now.
Prices get set at the margin, by whoever is making the next decision. In 2013 the person making that decision was somebody who wanted out of banks, and bitcoin's price moved on exchange failures, seizures, and capital controls, because those were the events that mattered to the people holding it.
In 2026 the person making that decision runs a portfolio with a mandate and a benchmark and a view on real yields, and reaches for bitcoin in the same motion they reach for gold or inflation-linked bonds. The asset didn't change. The median holder did, and price follows the median holder's worldview.
What that costs is specific. A hedge that moves on expectations about the thing it hedges is a derivative of that thing. If bitcoin rises when Washington signals it will hold borrowing costs down, then owning bitcoin is a position on American fiscal policy, priced off American policy decisions, held in American brokerage accounts, and taxed under American rules. The exit turns out to be a room in the same building, with a view of the lobby.
There's a further consequence that almost nobody in this industry has priced, and it runs in the other direction. An asset whose price reads as a public referendum on a government's borrowing costs is an asset that government will eventually take an interest in.
Gold spent the 1930s and again the 1960s and 70s being managed, restricted, revalued, and pooled precisely because officials understood its price as a verdict on their currency. Bitcoin is now generating that same signal, in public, updating every second, in a market that no longer needs anyone to use a wallet to participate. The more legible that signal becomes, the more it invites a response. That is a strange fate for a system designed so that no response would be possible.
None of this makes the week bearish, and it isn't a complaint about ETFs, which have done exactly what they were built to do. It's an observation about what the price is now measuring. For anyone reading this newsletter to understand where financial infrastructure is going, that number has stopped carrying the information it used to carry.
What this means for open money
The framework this publication uses asks a narrow question about any development: does it make financial infrastructure more open, along five specific lines. Can anyone use it without permission. Can anyone verify it. Can the rules be written as code. Can others build on it freely. Does the user hold their own assets. Price has never been one of the five, and this week is the clearest demonstration in three years of why it shouldn't be.
Run the week through them and four of the five went backwards or sat it out.
Permission got harder. The marginal buyer of bitcoin in August 2026 needed a broker, a jurisdiction that permits the product, and market hours. Custody moved further from the user, and a long way: $53.5 billion of cumulative fund holdings is a growing pile of bitcoin whose economic owners have signed away the one property that separated this asset from a claim on an institution. If the custodian fails, they're creditors in a bankruptcy like anybody else. Programmable rules and open composability never entered the picture at all.
One of the five worked, and it's the reason this piece exists. Verification held completely. The float publishes continuously, fund flows get disclosed daily, and anyone with a browser could establish the composition of this rally in an afternoon without an analyst, a subscription, or a source. In a closed market that argument would still be unresolvable and everyone would be trading anecdotes about who was buying. The open system's contribution to its own best week in three years was letting anybody check the receipts, which is a real thing and a much smaller thing than the price implies.
Put those together and the verdict follows. A rally in the price of open infrastructure, delivered entirely through closed channels, is the old system in new packaging wearing a very good costume. The gate sat where it sat in 2017 and again in 2021: at the brokerage account and the margin desk. What survives underneath is genuine and worth saying precisely. The asset inside those wrappers is still one that anybody can hold directly, still moves on a network nobody can switch off, still settles without asking. The week did nothing to damage that. It also did nothing to spread it.
The practical loss here is a scoreboard. For a decade this industry has used price as a proxy for progress, and the proxy mostly worked, because for most of that decade the only way to buy the asset was to use the system. Every business plan, hiring cycle, and grant budget in crypto has some version of that assumption buried in it. This week the link came apart in public. A company reading a 23 percent week as a demand signal is reading a message about US fiscal policy and mistaking it for one about its own customers.
The gold ETF ran this experiment already
State Street and the World Gold Council listed GLD in November 2004, and over the next seven years gold ran from roughly $450 an ounce to about $1,900. The investor base widened enormously. Pension funds and advisors who would never have arranged vaulting, insurance, and assay took positions through a ticker.
Physical gold ownership stayed exactly as niche as it had always been. GLD didn't teach anyone to own gold. It taught them to own a ticker that tracks gold, which is a different product for a different customer, and the two have coexisted comfortably for twenty years. Once a wrapper exists, the marginal buyer always chooses it, because it's easier and it fits the mandates and custody arrangements institutions already run.
Where the comparison breaks should bother anyone building here.
Gold never claimed to be plumbing. Nobody argues that gold's importance depends on people settling trade in it, so GLD holders never taking delivery costs gold nothing at all.
Bitcoin can survive the same fate intact, because a bearer asset that appreciates quietly in a vault is doing its job. The rest of this industry cannot. Stablecoins, tokenized funds, onchain lending, settlement rails: every one of them makes a claim about being used, and a payments network nobody routes payments through is not a payments network at any valuation.
GLD describes a future where the asset wins and the system it was supposed to prove never gets built.
The honest case against this read
The objection I take most seriously is that the float is the wrong instrument. It counts dollars parked on public chains, and a bitcoin bought through a fund was always going to be invisible to it, so pointing out that the float didn't move comes uncomfortably close to observing that fund buyers used a fund.
If institutional demand keeps arriving, some of it eventually needs onchain dollars for settlement and collateral, and the float would follow with a lag. Anyone calling a structural break off seven days is overreaching, mine included. Treat this as a measurement rather than a verdict.
The second objection is that all of this is unfair to bitcoin specifically. Bitcoin's argument never required anyone to transact in it. A scarce bearer asset held for decades is the entire design, and judging it by settlement volume is judging a savings account by its turnover.
Correct, which is why this piece separates bitcoin from the rest of the stack. The charge lands on the assumption that a rising price validates the open infrastructure case. It doesn't, it never did, and the fund wrapper is the machine that finally makes that legible.
Third, and most uncomfortable for me: this is a market still 37.6 percent below where it was a year ago. Reading deep meaning into a violent recovery inside a drawdown could lead to embarrassment. A 23 percent week off that base is consistent with a bottom, a bear market rally, or noise, and we don't know which it is yet. The narrower claim is the one worth defending: the composition of the buying is knowable, and it says something the direction doesn't.
What to watch
The float is a real test rather than a rhetorical one, which is the nice thing about picking a number that publishes every hour. If it's still flat in three weeks with bitcoin holding above $75,000, the separation is structural and this argument holds. If it climbs back through the May peak of $321 billion, the onchain dollars were simply lagging.
The bigger one is Jackson Hole, August 27 through 29. The Kansas City Fed made this year's theme financial innovation and its implications for payments and policy, and it's Kevin Warsh's first symposium as chair. "
A Fed that picked payments as its flagship subject in the same month Treasury proposed a rule requiring stablecoin issuers to be able to freeze and burn their own tokens has something on its mind about who runs the plumbing. That speech matters more here than another ten percent on bitcoin, and so does the GENIUS Act final rule the Comptroller has promised for November. If the next leg up arrives with the float growing, regulated onchain dollars have started doing real work.
Daniel McGlynn
Last week's issue
Strategic implications
For builders. A move like this generates a quarter of inbound interest that has nothing to do with your product, and every instinct will tell you it's validation. Check the float before you rewrite the roadmap.
An enormous number of people bought exposure this week and roughly none of them acquired the ability to use an onchain system, so the funnel you care about didn't widen an inch. The opportunity is sitting in that gap. A fund holder has dollars, a brokerage account, and a fresh interest in the asset class, and no path from there to using anything. Whoever builds the on-ramp that makes step two feel like step one gets a cohort that showed up this week with nowhere to go.
For capital allocators. Be honest about the exposure you're now carrying. A position whose best week in three years was caused by a Treasury decision about the long bond is a bet on US fiscal policy, correlated with everything else you own that responds to real yields.
It may be exactly the trade you want, and it has stopped being a hedge against the thing it was sold to hedge. Size it accordingly. For anything held on an adoption thesis rather than a macro one, track the float instead of the price, because it's the one number here that borrowed money can't inflate and enthusiasm doesn't move.
For policymakers. Two facts from this week belong in the same sentence. The volume of dollar settlement money on public infrastructure didn't grow, and the price of the assets on that infrastructure rose 23 percent through vehicles the SEC approved and regulated custodians secure.
The regime built so far funnels the public into wrappers while leaving the open rails to the people already standing on them, which delivers exposure to the asset without any of the resilience the underlying system was supposed to provide. If the goal includes payment infrastructure that keeps working when an intermediary fails, the current settings aren't producing it, and November's stablecoin rules are the near-term chance to change that.
Rules that make regulated onchain dollars usable by an ordinary business would do more for financial stability than another approved wrapper. The float would tell you whether it worked.
Sources
[1] CoinDesk. "Bitcoin tops $77,000 as best week since 2023 pulls altcoins along for the ride." August 21, 2026. https://www.coindesk.com/markets/2026/08/21/bitcoin-tops-usd77-000-as-best-week-since-2023-pulls-altcoins-along-for-the-ride
[2] CoinDesk. "Bitcoin breaks out of six-week range, tops $71,000 as $3 billion in shorts get wiped out." August 20, 2026. https://www.coindesk.com/markets/2026/08/20/bitcoin-breaks-out-of-six-week-range-tops-usd71-000-as-usd3-billion-in-shorts-get-wiped-out
[3] CoinDesk. "Treasury's latest measure isn't QE or YCC. Still, bitcoin is skyrocketing. Here's why." August 21, 2026. https://www.coindesk.com/markets/2026/08/21/treasury-s-latest-measure-isn-t-qe-or-ycc-still-bitcoin-is-skyrocketing-here-s-why
[4] US Department of the Treasury. "Treasury Announces Increase in Buyback Operation Sizes." August 19, 2026. https://home.treasury.gov/news/press-releases/sb0607
[5] DefiLlama. "Stablecoins." Accessed August 21, 2026. https://defillama.com/stablecoins
[6] Farside Investors. "Bitcoin ETF Flow." Accessed August 21, 2026. https://farside.co.uk/btc/
[7] Farside Investors. "Ethereum ETF Flow." Accessed August 21, 2026. https://farside.co.uk/eth/
[8] CoinGecko. "Bitcoin." Accessed August 21, 2026. https://www.coingecko.com/en/coins/bitcoin
[9] CoinGecko. "Global Charts." Accessed August 21, 2026. https://www.coingecko.com/en/global-charts
[10] Federal Reserve. "Minutes of the Federal Open Market Committee, July 28-29, 2026." August 19, 2026. https://www.federalreserve.gov/newsevents/pressreleases/monetary20260819a.htm
[11] Office of the Comptroller of the Currency. "Comptroller Gould Discusses Digital Asset Innovation, GENIUS Next Steps." News Release 2026-69, August 19, 2026. https://www.occ.gov/news-issuances/news-releases/2026/nr-occ-2026-69.html