Decent Newsroom
Sign In

The Collateral Chain: How $165 Million of Debt Sets Today's Bitcoin Price

By Roger November 13, 2026
Nakamoto owes Kraken $165M against a $104M market cap, with 70% of its coins pledged and bought at $118,171. This drawdown is a collateral event, not an adoption event: the calendar of forced sellers, the 5.1% ten-year, and what the protocol did while the price fell.
The Collateral Chain: How $165 Million of Debt Sets Today's Bitcoin Price

The Collateral Chain: How $165 Million of Debt Sets Today’s Bitcoin Price

On Thursday a Nashville company filed an 8-K with the SEC. It was the second amendment to a loan it took out a year ago, and it contained a sentence that deserves to be read slowly: 3,030 of the company’s 4,283 Bitcoin were pledged to the lender as collateral, bought at an average price of $118,171, while the market pays $68,692.

Nakamoto Inc., ticker NAKA, was built to hold Bitcoin. It now owes its lender, the exchange Kraken, 165 million USDT against a market value of roughly $104 million. The debt is about 59 percent larger than the entire equity of the company. Its shares last changed hands near $5.79.

It would be easy to file this under “another leveraged treasury company got caught”. I think that misses what is happening. Nakamoto is not an accident at the edge of this market. It is the cleanest piece of evidence that the current drawdown is not an adoption event, not a sentiment event, and not a referendum on Bitcoin’s future. It is a collateral event. The price of Bitcoin right now is being set, at the margin, by the calendar of people who have to sell it.

The trade that only works upward

The structure was never complicated. Borrow a dollar-pegged stablecoin, buy Bitcoin, pledge the Bitcoin back to the lender, and service the interest by selling coins when the price falls. When Bitcoin rises faster than the cost of carry, the arithmetic flatters everybody: your collateral grows, your loan shrinks in real terms, and the equity underneath it expands without a single customer.

Nakamoto’s facility opened on 3 December 2025 at $210 million. Since then the company sold roughly 284 Bitcoin in March 2026 and about 600 in the second quarter, applying $45 million of the proceeds against principal to bring the balance to $165 million. That is a lot of selling for a company whose pitch is that it does not sell.

The November 11 amendment bought time rather than solvency. A $60 million tranche due on 4 December 2026 moved to 4 December 2027. A separate $105 million tranche still comes due on 30 June 2027. Kraken also agreed to count the net value of Nakamoto’s put options toward collateral. Chief Investment Officer Tyler Evans called the amendment progress: it “extends our runway and makes our collateral position more resilient”.

Options expire. A put that protects the collateral in November 2026 can be gone by the summer of 2027, which is precisely when the larger tranche is due. Reading the amendment as a rescue means assuming the lender’s forbearance outlasts the assets it was granted against.

Nakamoto collateral coverage against the Bitcoin price

Here is the arithmetic that the filing makes possible, and it is uncomfortable in a way the headline numbers are not. At the 30 September snapshot, 3,030 pledged coins at today’s price of $68,692 are worth $208.1 million against a $165 million loan. That is a coverage ratio of 126 percent. Not distressed. Not yet. But coverage falls one-for-one with the price of the collateral, and parity, the point where the coins are worth exactly the debt, arrives at $54,455 per Bitcoin, roughly 21 percent below today’s price. Below that line, the collateral is worth less than the loan it secures.

Two details make the position weaker than the ratio suggests. The first is the purchase price: at $118,171 per coin, the collateral was bought at a price 72 percent above today’s market, which means the company is not sitting on a cushion it can top up from gains; every additional dollar of collateral has to come from somewhere else. The second is a reconciliation problem in the company’s own rounded figures. Roughly 884 coins sold for $45 million of principal repayment works out to about $51,000 a coin on my own arithmetic — below anything Bitcoin traded for in 2026, where my measurement of daily candles puts the low at $57,800 on 1 July. Either more coins left the treasury than the filing’s approximations suggest, or part of the repayment came from somewhere other than coin sales. Neither reading is reassuring. The filing does not reconcile it, and I am not going to pretend the numbers are cleaner than they are.

It is not one company

The same trade exists across the listed-treasury complex, in weaker and stronger versions. This morning, CoinDesk reported that BTCS, a Nasdaq-listed Ethereum treasury company, sold ether at a loss to keep a $64 million Aave loan healthy. It holds 46,676 ETH against that borrowing. Ether trades at $1,933 as I write this; the loan becomes eligible for liquidation if ether slides to about $1,700.

Do the division yourself, as I did: 46,676 ether at $1,933 is $90.2 million of collateral against $64 million of debt. That is a loan-to-value of 71 percent. At the $1,700 trigger the same coins are worth $79.3 million, which is an LTV near 81 percent. So the market has learned the lender’s threshold, and the distance to it is a 12 percent decline in ether. Twelve percent is not a scenario. It is a Tuesday.

I hold no brief for ether and I will not pretend the collateral matters as much as the structure does. It matters that the pattern repeats with an asset that has no fixed supply and no production cost floor: the trade is not “Bitcoin treasuries went wrong”. The trade is “borrowing short to hold a volatile asset goes wrong when the price of borrowing rises”. Bitcoin is simply the collateral where the balance sheet is public and the numbers are checkable.

Bitcoin daily closes, August 2025 to November 2026

The rate is the knife

Everything above is downstream of one number. The US 10-year Treasury yields 5.129 percent as of 09:31 CET this morning, in my own reading of the market. Bitcoin Magazine reported the 10-year rising about 12 basis points to 5.105 percent on Wednesday and touching 5.12, the highest since 21 October, with the two-year at 4.867 after a 14-basis-point move. Fed funds futures are pricing 29 basis points of hikes for December. Hikes. Not cuts, not a pause — the base case is a quarter-point increase.

For an Austrian, this is not a policy detail, it is the whole story. The interest rate is the price of time. When it rises, every claim whose payoff sits in the future is repriced downward at once, and the assets with the longest duration and the least cash flow fall hardest. A non-yielding asset held with borrowed money has the longest duration of anything in the market. There is no dividend to discount, no earnings to grow into the rate; there is only the cost of carry, applied every day.

Put a number on the carry. Nakamoto pays interest on $165 million. At a rate anywhere near the short end of the curve, say 4.9 percent, that is about $8.1 million a year: my own illustrative arithmetic, since the coupon sits in an amendment I have not read. Against a market capitalisation of $104 million, the interest alone consumes roughly 7.8 percent of the company’s equity value every twelve months. The company does not have a Bitcoin problem. It has a financing cost that compounds against an asset that pays nothing.

Meanwhile the fiscal backdrop removed any hope that this is a passing spike. The Treasury reported a $404 billion deficit for October, the first month of fiscal 2027, up 42 percent from $284 billion a year earlier, with receipts of $369 billion against outlays of $773 billion. Brent is quoted at $95.86 this morning, having settled 3.3 percent higher on Thursday after attacks on Saudi energy infrastructure escalated. Gold is $3,897.80 an ounce. This is what the late stage of a credit expansion looks like from the inside: the deficit does not shrink when rates rise, because the interest on the debt rises with them, and the central bank is asked to choose between the currency and the collateral.

The marginal buyer became the marginal seller

For two years the marginal buyer of Bitcoin was not a person with a wallet. It was a pipeline: an ETF share created on demand, a treasury company issuing convertible notes, a basis trade financed in the repo market. Each of those channels had a funding cost. Now the same channels are the sellers, and the sellers do not care about the thesis.

The evidence is stacked in one week. Spot bitcoin ETFs are down more than $1.5 billion in net outflows for the year, per CoinDesk, and BlackRock’s IBIT bled about $2 billion over nine straight sessions through Wednesday before closing 4.5 percent lower at $40.36. The Block counted $798 million of crypto positions liquidated in the four hours before 06:23 CET on Friday, $718.5 million of them longs. James Check of Checkonchain counted over $680 million of realised loss locked in by investors selling in a panic the day before — his read is that these are the buyers who were too indecisive to buy at these levels last time, then bought too much above $80,000.

Strategy, the largest of the treasury companies, closed at $114.41 on Thursday, down 6.5 percent on the day, and leaves MSCI’s USA index at the close on 30 November, which turns every tracking fund into a seller on a date. Index mechanics explain November’s supply. They do not explain Nakamoto, whose problem is not a rule but a loan. And the headline that looked like a rebuttal deserves the footnote it came with: Jane Street raised its Strategy stake 53.8 percent to 4,119,097 shares in the third quarter, while the same 13F listed call options on 13,573,000 shares and put options on 13,991,800 shares. A market maker’s book is not a conviction trade. The options lines are more than three times the size of the stock position, and the filing does not say how much of the stock is a hedge for them.

The whole picture is what a credit cycle looks like when the collateral is liquid and the lenders have lawyers. The coins do not disappear. They change hands at the worst available price, from the entity that had to sell to the entity that chose to buy.

The honest case against me

A good bear case has to survive the bull case, so let me build the strongest version of the other side. Bitcoin has already fallen 46 percent from its October 2025 peak of $126,200, and 21.6 percent this year from an opening print of $87,648; the 2026 low of $57,800 was made in July and held. Realised losses of $680 million in a day is capitulation, and capitulation is historically where the marginal seller runs out. Mean reversion from these levels has been a profitable trade more often than not. If Bitcoin reclaims $118,171 before the June 2027 tranche matures, Nakamoto’s coverage expands, the equity discount narrows, and the restructuring reads as a bridge rather than a trap. The whole structure then looks like a company that borrowed at the top and survived — which is what survival always looks like in hindsight.

That is a real argument, and it is falsifiable, which is more than most of the bullish commentary this week manages. Here is what would settle it against me: a capital raise or refinancing that retires the Kraken facility entirely, removing the counterparty concentration; and a reclaim of the $118,171 basis with the puts still alive. Neither has happened. What has happened is that the maturity schedule was extended by a year and the collateral was padded with instruments that expire. A bridge that is paid for with time is only a bridge if the thing on the other side does not move.

There is also a structural point the bulls cannot answer with a price target. Seventy percent of a treasury pledged to a single exchange is not custody risk in the abstract. Kraken holds the keys. Shareholders own a claim on a claim. The coins cannot be verified on-chain without trusting the lender’s attestation, cannot be voted with, and cannot be moved if the exchange is frozen, hacked or seized. The same week brought a small reminder of how that ends: the Indian exchange Unocoin disclosed that one of its BitGo addresses had sent out transactions it cannot trace, cannot log in to its enterprise accounts, and cannot fully see on-chain because of change addresses. Ninety percent of its customer crypto was in cold storage. The lesson there is not about India. It is that “held by a custodian” and “held” are different sentences, and treasuries learn it in the order the calendar allows.

What is not broken

I spent this morning measuring the network instead of the price, because the two move on different clocks.

Block 975,648. Difficulty 148.36 trillion, up 11.8 percent from where it stood thirty days ago and still 4.9 percent below its level a year ago, which I checked by comparing block headers at those heights. Miners commit capital on a 2,016-block horizon, not a four-hour one, and they are currently adding hashrate into a falling price. A mempool holding 37,321 unconfirmed transactions across 11.22 megabytes. The last block carried 4,188 transactions, paid the 3.125-coin subsidy plus 0.0195 coins in fees — 1.95 million satoshis, about 466 satoshis per transaction, roughly 32 cents at today’s price. Fees are 0.62 percent of the subsidy in that block. That is my own arithmetic from the node, and it is the number I care most about: the security budget is still overwhelmingly a subsidy, and the subsidy is denominated in a price that just fell 46 percent from its peak. Miners are the other forced seller in this story, and they sell to cover electricity, not to make a point.

The protocol absorbed all of it without noticing. The same day the market lost 4 percent, someone published a draft BIP for post-quantum addresses on Delving Bitcoin: a new bc1z format using ML-DSA-65, committing to a 32-byte SHA-256 hash of the public key and revealing it only at spend time, with 3,309-byte signatures and 1,952-byte public keys. Old nodes would see those outputs as anyone-can-spend, which is the standard soft-fork trick and also the reason the debate will take years. The draft cites roughly 6.9 million BTC sitting in addresses with exposed public keys. Its dollar figure is about $586 billion, which implies a Bitcoin price near $84,900 — my own division, and a reminder of how fast these documents age. At today’s price the same 6.9 million coins are worth $474 billion.

That gap between what the protocol is working on and what the market is arguing about is the whole point. One conversation is about signatures that will matter in a decade. The other is about who has to post collateral on Monday.

The other deflation

There is a second market being repriced by the same rate, and it is worth one paragraph because the connection is not obvious.

Nvidia lost 4.3 percent to $166.19 at Thursday’s close after Jensen Huang resigned as president, CEO and board member, effective immediately, with the CFO as interim. The Nasdaq fell 0.39 percent. That is a valuation event, not an earnings event — a company whose entire multiple rests on a founder-shaped story losing the founder.

At the same time, the price of a decision is collapsing. Google shipped Gemini Decision 1, a model that returns a calibrated choice instead of generating text, priced at $0.10 per million input tokens with nothing billed for output, 1,000 questions per request, 150 to 250 milliseconds end to end. TypeSafe’s Jev, the model that started this category in September, costs $0.042 per million input tokens. Artificial Analysis puts the cost per million tasks at $345 for Decision 1 against $103 for Drex Max and $156 for Jev. Cloudflare described a hybrid design at the same conference that cut six model calls down to one and 40,000 tokens down to 1,800.

That is genuine deflation, the productive kind, where the same output costs less because the method improved. It requires no credit and no central bank. And it is exactly what a 5.1 percent discount rate does to an industry that spent three years promising returns in the future: it separates the businesses that got cheaper from the businesses that only got more expensive. Bitcoin and open-source AI are the same argument here. Both take a function that used to be rented from an institution — money issuance, intelligence — and make it a commodity whose cost falls over time. Both are being repriced this week, for opposite reasons: one as collateral in a carry trade, the other as a capital-intensive promise. The difference is that one of them has no CEO to resign.

Three clocks

I do not know where Bitcoin trades in three weeks. I know what is scheduled, and schedules are more reliable than forecasts.

On 30 November, index funds tracking MSCI’s USA index must sell Strategy, because the index says so. On 30 June 2027, $105 million comes due to Kraken from a company whose collateral is worth 126 percent of that loan today and about 100 percent at $54,455. On 4 December 2027, another $60 million comes due, by which time the put options counted as collateral this week may have expired. Somewhere between those dates sits BTCS’s $1,700 ether trigger, which is 12 percent below the current price and does not care about anyone’s thesis.

The falsifiable version of my argument is simple: if Bitcoin is above $118,171 before June 2027, the collateral expands faster than the debt and this entire episode becomes a story about surviving a drawdown. If it is below $54,455 when a maintenance call lands, the coins go to the lender and the market takes them at whatever price the lender chooses to accept. Everything between those two numbers is noise with a narrative attached.

Hard money does not protect anyone from their own leverage. It only makes the bill arrive in the currency that cannot be printed, on a date that was agreed in advance, and it does not negotiate.

Timeline of dated forced-selling and repayment events

Sources

Write a comment