The Liquidity Crunch: Why Crypto Exchanges Are Secretly Fearing a Bank Run
Customers withdrew about $6 billion in three days prior to FTX’s collapse in November 2022. The exchange, which was the second-biggest in the world at the time, halted redemptions, declared bankruptcy, and left over 100,000 creditors with losses of between $10 billion and $50 billion that might never be fully recovered. Sam Bankman-Fried, the founder of FTX, was ultimately found guilty and given a 25-year prison sentence. The crypto era’s defining cautionary tale was the collapse of the exchange. However, the structural factors that initially made FTX susceptible to a run—opaque reserves, user funds placed in illiquid positions, and a platform that depends solely on confidence to operate—remain mostly unchanged throughout the majority of the industry today.
A cryptocurrency bank run’s mechanics are straightforward, but they happen quickly. On Twitter or Telegram, there are rumors that a major exchange is insolvent or that its reserves don’t match what users think they are. These rumors are frequently unverified and occasionally purposefully planted. A few big holders start to pull out. The withdrawal activity appears on the chain.
As more people see it, they take it to mean that the rumor is true. The withdrawals pick up speed. The exchange starts processing requests more slowly because it might not have all user funds in instantly liquid form. The delay turns into a signal in and of itself. The run is well under way by the time the exchange’s communications team drafts a statement refuting any issues. This is not a fictitious order. That’s what FTX experienced. Celsius Network experienced something similar. It is similar to what Diamond and Dybvig detailed in their 2022 Nobel Prize in Economic Sciences paper, but unlike traditional banks, cryptocurrency exchanges lack deposit insurance, a lender of last resort, and a regulatory system that consistently intervenes before users lose everything.
| Topic | Crypto Exchange Liquidity Risk and Bank Run Vulnerability |
|---|---|
| Key Historical Case | FTX — collapsed November 2022 after $6 billion withdrawn in 3 days; $8.9 billion in customer funds missing |
| FTX Creditors Affected | 100,000+ creditors; $10–$50 billion in unpaid debts |
| Comparable Traditional Bank Failure | Silicon Valley Bank (SVB) — $212 billion in assets; second-largest U.S. bank failure in history |
| USDC Stablecoin De-Peg | Dropped to $0.88 after $3.3B in reserves held at SVB were revealed (March 2023) |
| Traditional Bank Protection | FDIC insures up to $250,000 per depositor at U.S. banks |
| Crypto Exchange Protection | No federal deposit insurance; no lender of last resort |
| Fractional Reserve Risk in Crypto | Many exchanges lend or invest user deposits; don’t hold 100% liquid reserves |
| Social Media Amplification | Twitter, Reddit, Telegram can trigger mass panic withdrawals before any investigation completes |
| Research Supporting Framework | Diamond and Dybvig (Nobel Prize 2022) — showed how illiquidity + rumor can trigger runs on solvent institutions |
| Blue Owl Parallel (2026) | $5.4 billion redemption requests in Q1; redemptions capped at 5% — showing the same “gate” mechanism |
| Key Regulatory Gap | No standard proof-of-reserves requirements; limited regulatory oversight for most exchanges |
| Reference | Oliver Wyman Forum — Bank Runs Shake Crypto Markets and Conventional Finance |
The infrastructure to stop this kind of cascade has been built over the course of a century by the conventional banking system. Deposits at US banks are insured by the FDIC up to $250,000 per account. As a lender of last resort, the Federal Reserve can provide institutions with short-term cash shortages with emergency liquidity. Despite these safeguards, Silicon Valley Bank ran out of money in 48 hours in March 2023 after clients tried to withdraw $42 billion in a single day. The bank was unable to quickly liquidate the money from its securities portfolio. SVB failed despite having $212 billion in assets. Without any of those safeguards, cryptocurrency exchanges are structurally more vulnerable and frequently much less open about what they truly hold.
At the core of this vulnerability is the proof-of-reserves issue, which is surprisingly still mostly unresolved. Several significant exchanges released what they referred to as proof-of-reserves audits following the failure of FTX, which included screenshots or snapshots of assets held. The issue is that a snapshot can be taken when borrowed money has momentarily raised the balance and then promptly returned once the audit is finished. Some of the world’s most well-known exchanges might be functioning with true transparency and sound reserves. It’s also possible that the opaque sectors of the industry are home to the same practices that brought down FTX, such as reserve buffers that exist on paper but aren’t in reality, leveraged positions that appear fine until they don’t, and user deposits used for illiquid investments.
Even though it has to do with private credit rather than cryptocurrency, the comparison to what occurred with Blue Owl Capital in early 2026 is instructive. Blue Owl responded by capping redemptions at 5% per quarter, as allowed by fund terms, after revealing that investors had asked to withdraw $5.4 billion from two funds in a single quarter—21.9% of one fund and 40.7% of the other. The gates remained open. However, the fact that so many investors attempted to sell at the same time showed something crucial: confidence can decline more quickly than any liquidity structure is designed to withstand. Such gates are not included in the terms of service of cryptocurrency exchanges. They are predicated on the idea that users are free to leave at any time. The run starts as soon as that promise is questioned.
Observing this space gives the impression that the industry has viewed FTX as an anomaly rather than a sign of systemic risk. The post-FTX reforms, such as the publication of reserve snapshots by certain exchanges and the implementation of basic crypto licensing requirements by certain jurisdictions, only address the surface of the issue without addressing its underlying causes. Although there are bad actors, they are not the true problem. The true problem is that the architecture of how cryptocurrency exchanges function—holding user funds, allocating them in different ways, and processing withdrawals from a combination of current holdings and liquid assets—is intrinsically vulnerable to the same dynamics that have brought down banks for centuries. These dynamics are accelerated, not slowed, by social media. Additionally, when a cryptocurrency exchange bank run occurs, no institution is in a position to stop it before the damage is done because there is no significant backstop in place.
Whether significant regulatory reform will occur prior to the next significant exchange failure is still up in the air. The conditions for one seem to be unchanged, and the users who are most vulnerable are those who haven’t given their assets much thought.