The most reasonable role of centralized exchanges is that of an airport. You can transfer there, but do not settle there.

Written by: Daii

Some people open exchange accounts and see that their balance is still there. The numbers haven't decreased. The page refreshes normally. But the withdrawal button is now queuing. Customer service is left with only automated replies. The platform token continues to drop. The previously most critical question—whether the market will go up or down today—suddenly becomes meaningless.

Because the real question has become: is this money truly yours?

My judgment is very straightforward: all centralized exchanges will eventually exit the historical stage.

It might not necessarily be a failure. They may also be acquired, merged, liquidated, transformed, or proactively shut down like CoinEx described in the title. The difference is whether they return customer assets upon exiting or bury customers alongside the balance sheet.

1. A dignified exit is commendable, but it should not be deified

If an exchange can stop onboarding new users, limit incoming risks, systematically close derivatives and spot trading, and leave users with several months for withdrawal, it is certainly much better than suddenly freezing accounts.

This is passing grade exit management. It is not a great undertaking.

The return of customer assets by the exchange is merely the fulfillment of a debt. Just because there are many cases of embezzlement, misappropriation, and insolvency in the industry does not mean that returning money should be glorified as charity.

The title also states that overdue USDT not withdrawn will enter independent custody and incur a 5% fee each month based on the original balance. If this arrangement is true, it is not a detail to be glossed over. A 5% monthly fee is not a normal custody fee. After twelve consecutive months, even without compounding, the cumulative fee amounts to 60% of the original balance. It constitutes extremely strong withdrawal pressure.

This may help to quickly end operational liabilities. But it is not friendly to missing users, heirs, those with regional restrictions, and those whose account reviews are incomplete. The so-called "orderly clearance" cannot just be based on whether the platform has released a timetable. It must also consider whether users can smoothly complete identity verification, exchange small assets, close positions, and withdraw their coins.

The platform token buyback is the same. A buyback at the issuance price proves that the issuer is willing to provide an exit path. It does not equate to all holders experiencing no losses. The cost of acquiring CET is not the same for everyone. Some acquired it during the issuance phase. Others bought it at high levels in the secondary market. The buyback price may provide a safety net for the former, while still representing a huge loss for the latter.

Therefore, the most accurate evaluation of a shutdown is not "conscience" but four words: are the debts cleared?

2. Centralized exchanges are not on-chain infrastructure but high-leverage financial enterprises

Many people think of exchanges as highways in the crypto world. This analogy is wrong.

Highways do not simultaneously store your cars, determine the valuation of the cars, borrow the cars, use the cars as collateral, and then show you a screen that says "the car is still there."

Centralized exchanges can simultaneously play the roles of broker, custodian, market maker, lending platform, clearing house, and token issuer. Traditional finance typically separates these roles using licenses, capital requirements, customer asset segregation, and conflict of interest rules. Yet crypto exchanges often cram them into the same corporate group, or even into the same set of internal ledgers.

This is precisely the core issue repeatedly addressed by the International Organization of Securities Commissions in its 2023 proposal of 18 policy recommendations for the crypto asset market: conflicts of interest, customer asset protection, market manipulation, cross-border cooperation, and operational and technological risks.

This business has three clocks that count down simultaneously.

The first is liquidity.

In a bull market, transaction volume, listing fees, interest income, and derivatives fees all look great. After the market cools down, income drops quickly, while safety, compliance, customer service, and infrastructure costs do not decrease simultaneously. Smaller exchanges particularly suffer. The fewer users there are, the thinner the order book. The thinner the order book, the less willing professional traders are to enter. When professional traders leave, ordinary users receive even worse prices.

This is a typical reversal of network effects.

The second is regulation.

In 2023, the U.S. Department of Justice reached a settlement with Binance for over $4.3 billion. The Commodity Futures Trading Commission subsequently enforced a court order requiring Binance to return $1.35 billion in illegal earnings and pay a $1.35 billion civil penalty. This case does not prove that all exchanges have the same problems, but it sufficiently illustrates one fact: global exchanges do not face just one licensing fee; they face questions from multiple jurisdictions regarding customer identification, anti-money laundering, sanctions compliance, and derivatives access.

Compliance is not a ticket. It is a fixed cost that continuously eats into profits.

The third is trust.

Bank runs typically occur after negative news. Crypto exchange runs can happen much faster. Assets can be transferred around the clock. Rumors can spread around the clock. Users do not even need to queue up in a storefront. As long as a few large holders withdraw coins first, the platform may be forced to sell illiquid assets at the worst possible time.

If any of these three clocks hits zero, the exchange will begin to contract. If all three hit zero, the founder’s will has very little value.

3. "Reserve ratio over 100%" is not proof of solvency

This is where retail investors can be most easily fooled by a single statement.

Asset reserves answer the question: how much on-chain asset does the platform control?

Solvency must also answer several other questions: how much does the platform owe customers? Are there any undisclosed corporate debts? Are assets pledged or encumbered? Can affiliated companies call upon them? Are there any omitted accounts in the liability statistics? Does the audit cover a snapshot or the ongoing state of operations?

Therefore, a Merkle tree proof showing assets greater than customer balances does not automatically prove the entire company is solvent. It also does not prove that the management did not move funds before and after the snapshot.

The Bank for International Settlements' analysis of the crypto ecosystem notes that centralized intermediaries bear familiar liquidity mismatches, leverage, and run risks from traditional finance, while forming complex connections with on-chain markets. Blockchain can make a portion of asset flows public, but it does not automatically disclose all liabilities, guarantees, and related transactions of the company.

Truly valuable reserve proof must at least be combined with proof of liabilities, customer asset segregation, independent verification, address ownership validation, and continuous updating. Even if all of this is accomplished, it merely reduces information asymmetry. It does not eliminate operational risk.

Translating "on-chain visibility" into "absolute safety" is as ridiculous as claiming that seeing a safe proves the company has no debts.

4. Exchanges are bound to fail, but that does not mean they cannot be used now

The Bitcoin white paper aims to solve the dependency of electronic payments on trusted third parties. The irony is that a large number of users who buy Bitcoin eventually store it back in a trusted third-party database for a long time.

Exchanges are certainly useful.

They provide fiat entry points, price discovery, deep liquidity, risk control tools, and simple interfaces. For the majority of ordinary users, being fully on-chain still means managing private keys, authorization risks, cross-chain risks, smart contract risks, and a higher operating threshold.

But "usable" and "can be entrusted long-term" are two different matters.

The most reasonable role of centralized exchanges is that of an airport. You can transfer there. Do not settle there.

What users really need to do is not complicated: try to move long-term assets into wallets where they can verify control; test small withdrawals with large amounts of capital first; do not just look at the reserve addresses, but also question liabilities and asset segregation; do not turn spot assets into unsecured claims against the platform for a few cents of profit; do not treat platform tokens, platform financial products, and the platform itself as three independent risks, as they often represent three faces of the same risk.

As for the increasing number of exchanges shutting down in the industry, I do not believe this is necessarily a bad thing.

Platforms without liquidity, compliance capabilities, and the ability to continually prove solvency should naturally exit. Struggling only drags operational failure into customer losses. The maturity of an industry is not marked by the continuous increase in the number of exchanges. It is when exit mechanisms become more common than flight mechanisms.

If CoinEx, as described in the title, can ultimately fulfill its commitment to full repayment, it deserves a restrained acknowledgment: it chose to end when it could still return the funds.

However, history will not change the structural risks of centralized custody just because of a dignified exit.

References

Bank for International Settlements, The Crypto Ecosystem: Key Elements and Risks, 2023

IOSCO, Policy Recommendations for Crypto and Digital Asset Markets: Final Report, 2023

U.S. Department of Justice, Binance and CEO Plead Guilty to Federal Charges in $4.3B Resolution, 2023

U.S. Commodity Futures Trading Commission, Federal Court Enters Order Against Binance and Former CEO Changpeng Zhao, 2023