The Hidden Half: Why “Unknown Ownership” Bitcoin Is Not What You Think

On October 14, Alex Thorn, Head of Research at Galaxy Digital, made a simple but explosive claim: more than half of all Bitcoin sits in addresses whose owners cannot be identified. Ledgers don’t lie. But interpretation does. Anomaly detected. Look closer. That single sentence sent ripples through the institutional narrative that has been building for two years—the story that Wall Street has quietly taken over Bitcoin. Suddenly, the data says otherwise. And yet, the way most people are reading this finding is exactly wrong. Let me step back and give you the methodology first. On-chain analysis is not a crystal ball. It’s a forensic tool. When I was auditing the EOS pre-sale back in 2017, I spent months manually verifying transaction hashes against witness lists. That taught me a lesson: you can only know what the chain tells you, and the chain is often silent. The standard technique for assigning an address to a real-world entity is called clustering. The most common heuristic is simple: if multiple addresses spend coins together in a single transaction, they likely belong to the same owner. That’s the Common Input Ownership Heuristic. It works well for exchanges, custodians, and whale wallets that consolidate funds. But it has a fundamental blind spot. It fails on CoinJoin transactions. It fails on mixer outputs. It fails on exchange batch withdrawals where thousands of users share the same input. And critically, it fails on addresses that have never transacted with anyone else—because there is no co-spend signal to cluster on. So when Alex Thorn says “unknown ownership,” he is not telling you that there are shadowy whales hiding half the supply. He is telling you that the current analytical tools have a ceiling. That ceiling is not a bug. It is a structural feature of Bitcoin’s self-custody design. In a system where the private key is the only proof of ownership, an address that has been used once and then left untouched is indistinguishable from a lost coin. There is no KYC layer, no registry, no centralized authority to say “this belongs to X.” The protocol was built that way on purpose. And that means the “unknown” category is not a transient data gap—it is permanent. Follow the gas, not the hype. The gas here is the pattern of historical transactions, and it tells us a story that no single research note can fully capture. Let me break down what actually sits inside that “unknown” bucket. Based on my own forensic work and the published estimates I trust, there are three distinct populations. First, permanently lost coins. Private keys that were destroyed, forgotten, or written on paper that got thrown out. Industry estimates range from 1.5 million to 3.7 million BTC. That’s roughly 8% to 18% of the entire supply. Second, dormant addresses. The most famous example is Satoshi Nakamoto’s original 1.1 million BTC, which has never moved. Those coins are not “unknown” in the sense of being hidden—they are simply frozen. Third, and most interestingly, active self-custody addresses that simply haven’t been clustered by any analytics firm. These are real human beings who hold their own keys, use hardware wallets, and transact in ways that resist automated tagging. When you lump all three together under “unknown,” you blur the most important distinction in Bitcoin: the difference between dead coins and living coins. Why does this matter? Because the market is treating Thorn’s finding as a bearish signal for institutional adoption. Let me show you why that’s a misreading. During the 2024 ETF inflow analysis I ran, I tracked the movement of funds from custodians to Coinbase Prime and correlated those flows with exchange reserves. I found something counterintuitive: even as ETF inflows were massive, the majority of coins in circulation were still not held by institutions. The flow data showed that institutional buying was being offset by long-term holders taking profit. The supply was not being absorbed—it was being recycled. That is exactly what Thorn’s “unknown ownership” statistic reinforces. But here’s the contrarian angle: “unknown” does not mean “retail.” It does not mean “decentralized.” It means “unclassified.” If you actually look at the on-chain evidence, you will see that a large share of those unclassified addresses are likely what analysts call “satoshi-era coins”—coins mined in the first two years of the network’s existence. These are not dispersed retail holders. They are early miners and whales who have held for over a decade. That is not a decentralization vote. That is a concentration of unmoved wealth. The fact that they have not moved does not mean they will never move. What I call “the sleeping whale problem” is one of the most underappreciated risks in Bitcoin. If a fraction of those dormant coins suddenly became liquid, the market would face a supply shock that no ETF absorption could match. So the real insight here is not “institutions haven’t taken over.” The real insight is that nobody—not Galaxy, not Glassnode, not Chainalysis—can tell you with certainty what the true ownership structure is. The data is opaque, and pretending otherwise is dangerous. Now, let me address the elephant in the room. Alex Thorn works for Galaxy Digital. Galaxy is not a neutral observer. It runs a spot Bitcoin ETF, it operates mining infrastructure, and it has a vested interest in promoting Bitcoin as a decentralized store of value. That is not a knock on Thorn’s integrity—he is one of the most rigorous on-chain researchers in the business. But it is a structural incentive. When your business depends on the narrative that Bitcoin is not controlled by a small group of institutions, you are more likely to highlight data points that support that narrative. The “unknown ownership” statistic conveniently undermines the fear that ETFs are swallowing the float. That does not make the statistic false. But it does mean you should treat the conclusion—that Bitcoin is dominated by self-custody—with a healthy dose of skepticism. History repeats, if you read the chain. And the chain tells you that data can be shaped by the person holding the microphone. There is a deeper point here that most commentary misses. The high proportion of unclassified addresses is not just an analytical curiosity. It has direct implications for the regulatory battlefront. If regulators decide that “unknown ownership” is a red flag, they will use it as justification to crack down on self-custody tools. The EU’s MiCA framework already has provisions that effectively restrict non-custodial wallets. The FATF Travel Rule requires exchanges to transmit sender and recipient information, which is impossible for a self-custody transfer. So the more the data shows that Bitcoin is dominated by self-custody, the more those regulators will argue that Bitcoin is a compliance nightmare. That is the opposite of bullish. That is a long-term drag on the protocol’s ability to function as a global monetary layer. Let me give you a concrete example from my own experience. In 2020, during the DeFi Summer, I built a Python script to track whale flows through Compound. I discovered that a single entity was rotating assets across 50 wallets to exploit interest rate discrepancies. That clustering methodology worked because those wallets were transacting with each other. But when I tried to apply the same technique to Bitcoin addresses that had never interacted, I hit a wall. The data was simply not there. That experience taught me that the very tools we use to bring order to the chain are limited by the chain’s privacy design. The “unknown” bucket is a mathematical consequence of that design. It is not a temporary failure that will be fixed by better algorithms. It is a permanent feature of a system that values sovereignty over surveillance. So what should we take away from this? First, do not interpret “over half unknown” as a bull or bear signal. It is neither. It is a measure of our own ignorance. Second, do not assume that self-custody equals decentralization. A concentrated group of early adopters holding keys in the 2010s is not the same as a broad base of retail participants in 2025. Third, watch the regulatory response. Self-custody is not just a personal choice; it is a political statement that regulators will answer. The next signal to monitor is whether any major jurisdiction proposes a registry for self-custody addresses. If that happens, the “unknown” half of Bitcoin will become the center of the most consequential crypto policy fight in a decade. Until then, treat this statistic as a calibration tool, not a trade signal. The code remembers what people forget—but the code also hides what people want to hide.