Tracing the code back to the silence of 2017, I learned that a number is not a truth. It is a claim. When I first read that TD Cowen, an equity research division of TD Securities, had published an estimate that publicly listed companies could eventually hold as much as 2.1 million Bitcoin, I did not ask whether the target was too high or too low. I asked what dependencies were hidden inside the number, what assumptions had been left unlabeled, and what happens when those assumptions break.
The figure carries a clean arithmetic shine. 2.1 million Bitcoin divided by 21 million total supply equals ten percent. A whole number. A psychological threshold. The kind of number that lands easily in a headline and survives the scroll past the chart. But numbers that arrive without a timestamp, without a company list, and without a model are not measurements. They are weather vanes. They tell you which way the institutional wind is blowing, not how hard it will blow.
In the quiet, the protocol reveals its true intent. The protocol here is not Bitcoin consensus. It is the balance sheet layer that Wall Street has been building on top of Bitcoin since MicroStrategy changed its treasury policy in August 2020. TD Cowen's report is not a technical document. It contains no audited smart contract, no new cryptographic primitive, no proposed upgrade. Its technical value is close to zero. But its signal value is not. A mainstream investment bank is telling its clients that Bitcoin belongs on the corporate balance sheet. That changes the conversation, even when it changes no code.
I have watched the corporate treasury story evolve from a founder's quirk into a sector. MicroStrategy proved that a software company with fading fundamentals could borrow money at low interest, convert the proceeds into Bitcoin, and repackage itself as a levered Bitcoin vehicle. The market rewarded the behavior. The reward created an incentive for other companies to copy the behavior. TD Cowen's 2.1 million Bitcoin estimate is a directional claim that the copying will continue, and accelerate, until publicly listed companies hold ten percent of all Bitcoin that will ever exist.
That is a large claim. The largest hidden assumption is the one that makes the number feel plausible: that the borrowing cost of corporate debt will remain lower than the expected annual appreciation of Bitcoin. The entire strategy is a carry trade. It is not a belief that Bitcoin is money. It is a belief that the spread between cheap debt and an appreciating asset will remain positive. If that spread closes, the trade closes. The number closes with it.
Disassembling 2.1 Million
What bothers me most is what the report does not say. We do not know the timeframe. We do not know whether the estimate includes only companies that already hold Bitcoin or also companies that are expected to enter. We do not know whether it counts miners as public companies or excludes them. We do not know whether the model assumes constant buying, lumpy buying, or one-time acquisitions. A number without a model is not a forecast. It is a narrative anchor. The market will remember the round number ten percent long after it forgets the caveats.
The supply microstructure of the estimate deserves a deeper cut. Ten percent of total supply is not the same as ten percent of liquid supply. Bitcoin has a meaningful fraction that will never move again: wallets whose private keys have been lost, forgotten, or locked in cold storage for a decade. Conventional estimates place the lost and dormant share somewhere between twenty and forty percent of the total supply. If the lost share is three to four million Bitcoin, the liquid pool that can actually turn over in markets is closer to seventeen or eighteen million Bitcoin. Two point one million Bitcoin held by public companies would then represent between twelve and fifteen percent of the practically available supply. That is no longer a meaningful institutional allocation. That is a structural position.
The distinction matters for liquidity, because corporate treasuries are not time locks. A private key in a cold wallet does not respond to a margin call. A corporate treasury does. When a company buys Bitcoin, the Bitcoin does not leave the market forever. It leaves the current order book and waits on a balance sheet. It can return under the right conditions: a credit covenant, a share buyback emergency, a change of CEO, a short seller's report, a quarter with weak cash flow. The market treats corporate accumulation as a supply shock. But the accumulation is reversible. The reversibility is the risk that the ten percent narrative cannot contain.
There is a second illusion hiding in the supply story. Every Bitcoin that a company buys is bought from someone who sells. The total supply does not shrink. What changes is the identity of the holder, and the identity of a holder is a form of market information. When an anonymous whale accumulates, the market can see the balance grow but cannot see the intention. When a public company accumulates, the market sees the filing after the fact and then forms an intention about the intention. The buying is not overnight. It is a sequence of quarterly disclosures. Each disclosure changes the expectation of the next disclosure. That lag creates a market dynamic that no on-chain metric can capture.
The borrow-and-buy loop is the engine beneath the whole estimate, and it deserves forensic scrutiny. The loop is simple enough to be modeled. A company issues convertible notes. The notes carry a coupon, often low, because the buyer receives equity upside if the company's stock rises. The company takes the proceeds and buys Bitcoin. In a bull market, Bitcoin rises, the company's assets rise, the stock rises, and the convertible note becomes cheaper to service. The company issues another note. The loop feeds on itself.
This is not a Ponzi scheme. A Ponzi scheme pays early participants with the capital of new participants. The borrow-and-buy loop pays no one directly; it buys an asset in the open market with debt that carries a contractual obligation. But the loop has the same-shaped fragility as a positive feedback system. It works in both directions. When Bitcoin falls, the company's assets fall, the stock falls, the convertible note becomes difficult to refinance, and the company's ability to buy more Bitcoin disappears. The loop does not pause. It reverses. The reversal is not a margin call in the traditional sense, but it is a forced deleveraging event if the company needs to defend its credit quality.
I have seen this shape before. During my 2017 audit of Bancor's V1 smart contracts, I isolated seven integer overflow vulnerabilities in the liquidity pool logic. The bugs were theoretical until the conditions turned. The conditions always turn. The same is true of the corporate treasury model. The vulnerability is not in the Bitcoin network. It is in the debt structure. An overflow in narrative can produce a financial overflow much faster than an integer overflow in Solidity.
When I looked at OpenSea's off-chain order matching in 2021, I found a signature forgery vulnerability that could have allowed an attacker to approve an order they did not sign. The fix was to verify the signature before accepting the order. The corporate treasury version of that flaw is different. A company can verify that it owns Bitcoin. It can verify that the Bitcoin is in custody. But it cannot verify whether it will have the discipline to hold through a drawdown, or whether its creditors will allow it to hold. The missing validation is not on the chain. It is in the boardroom.
The accounting layer makes the cycle more visible and therefore more dangerous. Since the 2025 fiscal year, companies that hold digital assets are required under US GAAP to measure those assets at fair value, with quarterly changes flowing through net income. This is a dramatic shift from the old cost model, where a falling Bitcoin price could be recorded as an impairment but a rising price was invisible. Under fair value accounting, Bitcoin volatility is no longer a footnote. It is an earnings event. Every quarter, a company that holds Bitcoin reports a profit or a loss that has little to do with its operating business. That creates a new kind of earnings risk. A company that invests two billion dollars in Bitcoin is no longer selling software. It is selling a leveraged opinion on the price of an asset that trades on Saturday and Sunday.
The fair value rule is marketed as transparency. It is more honest, but it is not safer. Under the impairment model, a company could hide the upside and delay the downside. Under fair value, the downside is enforced quarterly. That makes the treasury strategy harder for a conservative board to sustain. The volatility will not be absorbed by a footnote. It will be printed on the income statement, in the same line as revenue, and equity analysts will begin to ask a question that has no comfortable answer: is this company a business or a Bitcoin fund?
The creditor in the room is the most underappreciated actor in TD Cowen's estimate. The buyers of convertible notes are not Bitcoin believers. They are fixed income investors. They have a contractual claim on the company's cash flow. If the company's operations weaken while Bitcoin is rising, the creditor is patient. If Bitcoin is falling while operations weaken, the creditor becomes a forced seller of the company's behavior. The company may sell Bitcoin to defend its credit rating. It may sell Bitcoin to buy back stock. It may sell Bitcoin to fund a covenant. The exit is not a market panic necessarily. It is a sequence of quiet, rational decisions made in a boardroom, each one visible in a 10-K filing, but none of them predictable from an on-chain address.
The Credit Ceiling
The credit market has a ceiling that the crypto market often ignores. Every convertible bond issued by a Bitcoin treasury company is absorbed by someone. Those buyers are not unlimited. They underwrite the company's ability to service debt and, after 2025, they follow the fair value earnings swings. A company that has converted its balance sheet into Bitcoin has a narrower margin of safety than a company that uses debt for plant expansions. Credit analysts know this. They will demand higher coupons as volatility rises. Higher coupons shrink the carry trade. The carry trade is the engine of the ten percent estimate. When the engine cools, the estimate loses its fuel.
This gives us a cleaner leading indicator. Do not watch the Bitcoin price alone. Watch the credit spreads of the companies that hold Bitcoin. Watch the size and terms of their new convertible offerings. Watch the percentage of proceeds that goes directly into Bitcoin. If a company issues debt and does not buy Bitcoin, the strategy has reached its credit limit. That will happen long before the on-chain supply data shows a slowdown.
Custody is the layer that everyone assumes is solved. It is not solved; it is delegated. A company that holds Bitcoin on its balance sheet almost always holds it through a qualified custodian. That custodian has its own counterparty risk, its own access controls, and its own regulatory obligations. If public companies accumulate 2.1 million Bitcoin, those concentrated custodial positions become a high-value target. A physical gold vault can be guarded by armed security. A digital vault holding millions of Bitcoin is guarded by software, procedures, and insurance policies. Insurance policies have limits. Software has bugs. Procedures have human fallibility. The crypto market already watched centralized lenders fail in 2022. A corporate treasury is not a lending desk, but it is centralized custody with a different face.
The corporate treasury is becoming the fourth large holder category after miners, exchanges, and ETFs. Each category behaves differently. Miners are forced sellers in a downturn because they must pay electricity bills. Exchanges hold coins for client activity and can see balances leave in a crisis. ETFs have daily creation and redemption flows that are observable. Corporate treasuries have none of those constraints, and none of those observability features. Their sale decisions are discretionary. A miner's sale can be modeled from energy prices. An ETF outflow can be modeled from market price. A corporate treasury sale is a governance event, and governance events are the hardest things to model.
Scenario analysis gives texture to the number. A base case is that public company holdings continue to grow, but only among firms that have already adopted the strategy, and the growth rate decays as debt capacity is exhausted. A bull case requires a company such as Apple, Microsoft, or another mega-cap to put Bitcoin on its balance sheet, because mid-cap companies alone cannot absorb 2.1 million Bitcoin without excessive ownership concentration inside a single credit cycle. A bear case begins with a credit event: a convertible note matures during a Bitcoin drawdown, the company cannot refinance, and a large sale is announced. The bear case is the one that the report does not mention, but it is the most mechanically plausible.
In 2025, I led a team analyzing zero-knowledge proof integration into institutional custody. We found a subtle implementation flaw that compromised the privacy guarantee under a specific call sequence. The public debate was about scalability. The actual vulnerability was in the governance of who could invoke the proof generation. The same pattern appears here. The public debate about 2.1 million Bitcoin will be about adoption. The actual vulnerability is governance: who decides when the company sells, what triggers the sale, and how much warning the market receives before the sale is disclosed.
Contrarian: The Institution Is Not a Safe Harbor
The market narrative assumes that institutional custody is a safer version of self-custody. That is a false comfort. When an individual holds Bitcoin in a hardware wallet, the failure mode is operational: a lost key, a phishing attack, a malicious update. When a public company holds Bitcoin, the failure mode is governance-driven: a risk committee vote, a credit waiver, a shareholder activist demand, a founder's health event, a regulator's request. These are not cryptographic failures. They are human failures. They cannot be audited by running a static analysis tool on the blockchain.
The deeper blind spot is that public company disclosure is a lagging indicator. Companies report their positions quarterly, or in some cases only when the position becomes material. The market sees the accumulation after it has happened. This creates an information asymmetry that is different from the asymmetry created by a private whale. A private whale has no obligation to report. A public company has no obligation to report in real time. The company can accumulate Bitcoin for weeks before a filing reveals the position. In a market already sensitive to narrative, that filing will trigger a response, and the company may wait for the response before buying more. The result is not a smooth demand curve. It is a staircase of quarterly surprises.
Regulators have not caught up with this staircase. If public companies accumulate 2.1 million Bitcoin, the SEC will face a set of uncomfortable questions. Does a company that borrows to buy Bitcoin need to disclose its plan in advance? Does a group of companies that follow the same treasury strategy constitute a coordinated action? Should a company's Bitcoin position be treated like a commodity hedge position, or like an inventory position, or like a speculative investment? Every answer changes the risk profile of the strategy.
The anti-concentration assumption is worth naming. Bitcoin's security model does not depend on who holds the coins. But its censorship resistance narrative does. If ten percent of the supply sits on the balance sheets of a handful of companies, each one subject to a government regulator in its own jurisdiction, Bitcoin has acquired a new set of trusted parties. These trusted parties are not Byzantine fault tolerant. They are ordinary institutions with quarterly incentives. They are fallible. They can be compelled. A subpoena to a custodian is not the only pressure point; a subpoena to a company's board is equally effective.
The TD Cowen report, or at least the public version of it, does not model this. It does not tell us which companies, within what timeframe, under what interest rate assumptions, paid by what funding sources. It gives us a destination without a route. A destination without a route is not a forecast. It is a slogan. And in the crypto market, a slogan is a tradable object.
I want to be fair. An investment research report is not required to be a stress test. Its job is to provoke thinking, and this report has provoked a useful kind of thinking. It forces Bitcoin investors to ask whether the next holder of last resort is going to be a corporate treasurer rather than a retail trader. That is a meaningful shift in the market architecture. But the same shift brings with it a danger that is not captured in the headlines: corporate treasuries are not dead hands. They are living, breathing sellers.
Authenticity is not minted, it is verified. The crypto market has a habit of treating audited code as the only form of verification. But a balance sheet is also a form of code, and it has not been audited with the same rigor. When a smart contract has a privileged admin, security researchers call it a centralization risk. When a company has a founder who can move billions of dollars into a volatile asset, the market calls it a treasury strategy. The language is different because the venue is different, but the governance risk is the same. The audit community should apply the same lens to corporate treasuries that it applies to protocols.
What would make me believe the 2.1 million Bitcoin number? A report that separates existing holders from new entrants. A report that shows a distribution of company sizes and funding costs. A report that stress-tests the debt buy-and-buy loop at different interest rates, different Bitcoin drawdowns, and different credit market closures. A report that names the forced-sale circuit breakers and estimates their probability. That report would not be a price prediction. It would be a risk map. I would trade every bull case headline for a single risk map.
We audit not to judge, but to understand. In the quiet of the bear market, all the polished numbers go silent, and the underlying assumptions become visible. Solitude clarifies the signal amidst the noise. The signal here is that Bitcoin is no longer being evaluated as a network. It is being evaluated as an asset class, a treasury reserve, and a liability machine. Each of those framings carries a different set of risks. Each of them will produce a different kind of audit failure.
Layer two is a promise, not just a layer. The corporate treasury is a promise as well. The promise is that a company will hold Bitcoin in a disciplined, transparent, and durable way. That promise is not minted by buying Bitcoin. It is verified by the company's accounting controls, its debt covenants, its board oversight, and its ability to survive a Bitcoin winter without selling. The next big crypto audit will not be a smart contract. It will be a 10-K.
If public companies do reach 2.1 million Bitcoin, Bitcoin will survive. But it will survive as a different asset from the one I audited in 2017, and from the one that promised to remove trust from the system. The trust will not disappear. It will migrate to corporate management. That migration is not progress. It is a new form of concentration wearing a suit.
