5,900 BTC over three months, compared with 89,000 BTC purchased in July 2025 alone. Public companies that had fueled part of institutional demand for Bitcoin have almost stopped accumulating. The problem is not solely the price. Their average acquisition cost now stands at around $80,500, while BTC is trading below that level. As a result, the sector is collectively showing unrealized losses, new capital raises are becoming harder to convert into accretive purchases, and one of the major sources of demand in 2025 has suddenly weakened.
Bitcoin loses one of its biggest buyers
The comparison immediately shows the scale of the slowdown. According to Glassnode, publicly traded companies tracked as Bitcoin treasuries bought approximately 5,900 net BTC over the past three months. In July 2025 alone, they had added approximately 89,000. In other words, three months of recent accumulation represent less than 7% of the purchases made in a single month at the height of the previous euphoria.
This slowdown deserves more attention than a simple decline in volume. For much of 2025, treasury companies had created a new source of structural demand for Bitcoin. Strategy was raising capital. New companies were adopting the model. Businesses were buying BTC through share offerings, convertible bonds or various preferred instruments.
Bref Crypto noted in early September that Strategy and Strive had resumed accumulating during Bitcoin’s rebound. That mechanism is now moving much more slowly.
Glassnode sums up the situation in a fairly severe formula: a buyer that has stopped buying and is holding an unrealized loss does not provide support for the market.
Glassnode’s full report therefore does not say that companies will necessarily sell their reserves. It says something more immediate: they are no longer providing the marginal demand that previously helped Bitcoin absorb available supply.
Treasury companies’ average cost basis reaches $80,500
The level around $80,500 has become central. Glassnode calls it the Corporate Treasury Cost Basis. This is the estimated average acquisition cost of the bitcoins held by the public companies tracked in its model.
At the time of the report, Bitcoin was trading approximately 6% below this level. Collectively, these companies are therefore “underwater,” meaning they are sitting on unrealized losses. However, this should not be generalized: it does not mean that every company holding Bitcoin is losing money.
Purchase dates and prices vary widely. Strategy is a good example. The company holds 845,050 BTC, with an average cost of approximately $75,412 per unit, according to data cited by Cointelegraph. Its individual threshold is therefore below the group’s $80,500 level.
This distinction matters. Glassnode’s threshold measures a group of companies, not the accounting position of each one. It nevertheless remains relevant to the market because Bitcoin has already attempted twice in 2026 to reclaim this average: first in May, then on September 3.
Both attempts failed. The $80,500 level is therefore beginning to play a dual role. It is the point at which a large part of the sector would theoretically return to its entry price.
It is also an area where some investors could use a return to breakeven to reduce their exposure. A former buying zone can thus become resistance.
Classic market behavior. Much less comfortable when the buyers involved are companies that are supposed to accumulate Bitcoin for several years.
Strategy itself no longer operates as it did in 2025
The Strategy case shows how much the model has evolved. Michael Saylor’s company remains by far the largest publicly traded Bitcoin holder, with 845,050 BTC.
But its recent behavior is far less mechanical than before. After two months without a purchase, Strategy returned to the market at the end of August with 4,603 BTC, acquired for approximately $369.7 million at an average price of $80,318. Purchases then stopped again.
Bref Crypto reported on September 8 that Strategy had chosen to allocate $176.3 million to buying back its STRC securities, while increasing its buyback program for certain preferred securities to $2 billion.
Just a few months ago, a capital raise by Strategy almost automatically prompted the same question: how many bitcoins will Saylor buy? That is no longer automatic.
The company is now weighing Bitcoin, debt, preferred shares, financing costs and share buybacks.
It even sold 6,916 BTC between late June and early August, before returning to buy at a higher price a few weeks later. Bref Crypto detailed this sequence in which Strategy sold around $62,500 before buying back above $80,000.
CEO Phong Le defends this logic: Strategy is not primarily trying to trade the price of BTC. It is managing its cost of capital.
This is precisely what makes the Bitcoin treasury market more interesting. These companies are not passive wallets. They are financial structures.
The mNAV engine works less well when Bitcoin falls
To understand why purchases are slowing, it is necessary to look again at how many treasury companies operate.
The term mNAV broadly refers to the ratio between a company’s valuation and the net value of its assets, particularly its bitcoins.
Consider a company holding $1 billion worth of BTC. If the market values its shares at $2 billion, it benefits from a substantial premium. It can then issue new shares at that high valuation, raise capital and buy more Bitcoin. If the transaction is structured well enough, it can increase the amount of BTC held per share despite the dilution.
This is the famous flywheel—the self-reinforcing wheel.
The problem appears when the premium disappears.
A company holding $1 billion in Bitcoin but valued at only $800 million by the market can no longer easily sell shares to buy BTC without risking the destruction of value for its existing shareholders.
The engine seizes up.
An analysis published in August had already noted that several institutional vehicles had reduced their Bitcoin exposure and that some treasury companies were trading below the value of their assets, calling into question the market-funded accumulation model.
In other words…
When Bitcoin rises, shares rise, companies can raise funds, they buy Bitcoin and potentially reinforce the rally.
When Bitcoin falls far enough, the mechanism can work in the opposite direction.
Not necessarily to the point of forced selling.
But enough for purchases to stop.
That is exactly what Glassnode is observing today.
Unrealized losses change market psychology
An unrealized loss is not a realized loss.
This distinction is fundamental.
A company that bought Bitcoin at $90,000 and sees it fall to $75,000 has not permanently lost $15,000 per BTC as long as it does not sell.
It simply holds an asset whose market value has fallen below its purchase price.
For a long-term investor, this may seem secondary.
For a publicly traded company, the effects are much broader.
A decline in BTC affects the balance sheet’s value.
It can affect the share price. It changes the premium or discount relative to the value of the bitcoins held and influences the terms on which the company can raise capital.
And in some highly financialized models, it directly changes the ability to buy more Bitcoin.
The price therefore becomes both the market’s outcome and a variable in future financing.
That is why the $80,500 level tracked by Glassnode matters.
A move back above it would collectively return the tracked treasuries to unrealized profits.
It would also remove what Glassnode describes as a layer of “overhead supply”—an area where holders returning to breakeven might be tempted to sell.
Below this threshold, the dynamic is more uncomfortable.
Companies are accumulating little.
The market knows that some of them would probably like to recover their entry price.
And every rebound toward $80,500 must contend with this psychology.
Bitcoin ETFs slow at the same time
The problem would be less significant if other institutional buyers immediately took over.
That is not really the case.
U.S. spot Bitcoin ETFs recorded $462.7 million in net outflows over the five sessions through September 11, after three consecutive weeks of inflows.
Bref Crypto had already noted this shift in a week dominated by the CLARITY Act and the Federal Reserve.
Treasuries are slowing.
ETFs are recording outflows.
Stablecoin growth remains weak.
And capital realized on Bitcoin is also beginning to show signs of weakness.
Glassnode says Bitcoin’s Realized Cap rose for 27 consecutive days through September 14, before recording its first decline on September 15. This metric assigns each BTC the price at which it was last moved and provides an indication of the capital that has actually entered the network.
Its value stands at around $1,069 billion.
A single negative session obviously does not signal a collapse.
What interests Glassnode is the combination.
Corporate demand is stagnating.
ETFs have temporarily stopped absorbing capital.
Stablecoins are not growing sufficiently.
On-chain capital is losing momentum.
Each of these indicators taken separately would be manageable.
Together, they describe what the firm calls a “market in waiting.”
A market that is waiting.
The Fed makes treasury financing even more difficult
The macroeconomic timing is not helping.
The U.S. Federal Reserve has just raised rates by 25 basis points, bringing the Fed funds target range to between 3.75% and 4%. This is the first hike since July 2023.
For a Bitcoin treasury company, this matters on several levels.
Debt becomes more expensive.
Investors have access to higher yields on bonds.
The opportunity cost of exposure to a volatile asset increases.
And financing transactions become more demanding.
A company could raise capital relatively easily when its shares traded at a substantial premium to its BTC holdings and global liquidity favored risk assets.
The same structure becomes much less obvious with a depressed share price, an unrealized loss on its treasury and U.S. rates close to 4%.
Bitcoin does not need credit to function.
Companies that want to accumulate it on a large scale often do.
That is the key difference.
Financial conditions can therefore reduce corporate demand without changing a single line of the Bitcoin protocol.
Supply remains limited.
The halving remains in place.
Blocks continue to be produced.
But the marginal buyer has less cheap capital available.
And over a period of several months, that is enough to fundamentally alter the relationship between supply and demand.
Not every company is abandoning Bitcoin
However, the slowdown should not be turned into a generalized exodus.
Some companies continue to buy.
Capital B increased its reserves again in September.
Strive remains exposed.
Miners collectively continue to hold tens of thousands of BTC.
And Strategy still holds a position without equal.
BitcoinTreasuries currently lists more than 1.16 million BTC held by U.S. public companies, with Strategy alone accounting for 845,050 units.
The stock has therefore not disappeared.
What has temporarily disappeared is mainly the flow.
This distinction between stock and flow is essential.
A company that has held 40,000 BTC for a year indirectly supports market scarcity if it does not sell.
But it creates no new buying pressure today.
A company adding 5,000 BTC this week does.
In 2025, the market benefited from both: large holdings were removed from circulation while new purchases continued to absorb supply.
In September 2026, the first part remains largely intact.
The second has weakened.
That is why Glassnode’s 5,900 BTC matter more than the spectacular size of the balance sheets already built.
Price is formed at the margin.
Not by bitcoins that have been sitting in an institutional wallet for six months.
$80,500 becomes the level to reclaim
The treasury sector is therefore not necessarily dying. Rather, it is entering its first major test after the euphoria. As long as Bitcoin was rising, the case was relatively easy to make.
A company raised capital. It bought BTC. BTC rose. The balance sheet’s value increased. The stock could benefit from a premium. That premium made it possible to raise more. Then the cycle began again.
The current decline forces a less comfortable question: does the model still work when Bitcoin no longer rises enough to finance its own accumulation?
Some companies have operating revenue. Others have solid financing structures. Still others may simply be able to wait for several years.
They are not all therefore doomed to sell. But Glassnode’s figures clearly show that the group’s behavior has changed.
5,900 BTC over three months. 89,000 in July 2025 alone. An average cost of $80,500. Two failed attempts to move sustainably back above this threshold in 2026.
And institutional demand is also slowing on the ETF side. The next battle is therefore likely to be found less in another spectacular treasury announcement than in the price itself.
Above $80,500, part of the sector returns to unrealized profits. Stock-market premiums can rebuild. Capital issuance potentially becomes more attractive. Purchases could resume.
Below this level, the situation remains much less comfortable. The bitcoins already accumulated do not disappear. But their owners no longer necessarily want—or have the same means—to buy more. That is the central nuance of the current market. The Bitcoin treasury boom has not yet been erased. Its engine has simply stalled.
