AscendEX, BitMEX and BitMart all announced wind-downs within a month. The structural reasons behind the 2026 exchange closure wave, and how to judge whether the venue holding your funds is next.

Last updated: 28 July 2026
In July 2026, three centralised exchanges announced they were shutting down within a single month. AscendEX ceased operations on 1 July. BitMEX, which invented the perpetual swap and ran for eleven years, said on 23 July it would close on 23 September. BitMart followed three days later, ending nine years of operation.
None of these were dramatic collapses. There was no bank run, no frozen withdrawals, no founder disappearing. They were, by and large, orderly wind-downs with published deadlines. That is precisely what makes the pattern worth understanding: this is not a scandal, it is an industry structurally shedding its middle tier.
| Exchange | Announced | Status |
|---|---|---|
| AscendEX | Ceased operations 1 July 2026 | Withdrawals moved to manual review |
| BitMEX | 23 July 2026 | Closing 23 September, after 11 years |
| BitMart | 26 July 2026 | Trading ends 26 August; withdrawals until 31 January 2027 |
Smaller European venues, including Knaken and Zondacrypto, have also wound down operations during 2026. The full picture is one of consolidation rather than crisis.
This is the largest factor and the most under-discussed. A centralised exchange serving European users now needs MiCA authorisation, full Travel Rule infrastructure with a zero-value threshold, sanctions and address screening on every deposit, KYB on corporate customers, and ICT resilience obligations under DORA.
Those costs are broadly fixed. They do not scale down for a venue with a tenth of Binance's volume. A compliance function that a top-five exchange amortises across enormous volume is, for a mid-tier venue, simply an unaffordable line item. Regulation did not intend to eliminate the middle of the market, but that is its arithmetic effect. Our guides to MiCA authorisation and the Travel Rule set out what that overhead actually involves.
Perpetual futures volume on centralised exchanges fell around 10% in the second quarter of 2026, to roughly 12.7 trillion dollars. Meanwhile, perpetual DEXs' share of total open interest rose to about 13.5%. A mid-tier venue is squeezed on both sides: the largest exchanges take the liquidity-sensitive flow above it, and on-chain venues take the self-custody-preferring flow below it.
Mid-tier exchanges historically competed by listing assets the majors would not. That model depended on sustained retail appetite for long-tail tokens, and on listing fees from projects that could afford them. Both have thinned considerably.
Exchange tokens were a financing mechanism: issue a token, tie it to fee discounts and platform activity, and capture value from growth. It works in one direction only. BitMart's BMX fell roughly 58% in a day on the closure announcement, after a decline of around 70% over the preceding year. The token's value was always a derivative of the exchange continuing to exist.
You cannot audit a private company from outside, but the observable signals are more informative than most people assume.
The orderly wind-downs of 2026 have been comparatively benign. Users have had weeks or months of notice and, in BitMart's case, a six-month withdrawal window. The uncomfortable observation is that this outcome was a choice made by the companies involved, not a right the users held.
Assets on a custodial exchange are a claim against a company, not property under your control. In an orderly wind-down the claim is honoured. In a disorderly one it becomes an unsecured claim in an insolvency process, which historically resolves slowly and partially. The difference between those two outcomes is entirely outside the user's influence.
None of this makes exchanges useless. Fiat on-ramps, deep order books and fast execution are real services, and self-custody carries genuine risks of its own, chiefly irreversible user error. The reasonable conclusion is narrower: use exchanges for what they are good at, and do not treat one as a savings account. Our guide to self-custody covers the practical side, and CeFi vs DeFi compares the two models honestly.
Non-custodial protocols remove the counterparty entirely. JewelSwap operates on MultiversX, Sui and Radix, and users interact with smart contracts directly from their own wallets. There is no company holding the assets, therefore no company that can decide to stop returning them, and no wind-down announcement that puts a deadline on your access.
What replaces that risk is smart contract risk, oracle risk and the permanence of user mistakes. These are real. A protocol exploit is as final as an exchange insolvency, and no support desk will reverse a transaction you signed. The honest framing is that you are choosing which category of risk you would rather carry, and non-custodial systems at least put the decisive variables in your hands rather than someone else's.
Compliance costs under regimes such as MiCA and the Travel Rule have become a large fixed expense that mid-tier venues cannot amortise across their volume. At the same time, centralised futures volumes fell and decentralised venues took share. The middle of the market is being squeezed from both directions.
Assets on a custodial exchange are a claim against that company rather than property you control. In an orderly wind-down that claim is honoured; in an insolvency it becomes an unsecured claim. Safety depends on the company's solvency and conduct, which you cannot verify from outside.
AscendEX ceased operations on 1 July 2026. BitMEX announced on 23 July that it would close on 23 September after eleven years. BitMart announced on 26 July, with trading ending 26 August and withdrawals open until 31 January 2027. Smaller venues including Knaken and Zondacrypto also wound down.
Watch for licensing status in your jurisdiction, proof-of-reserves quality including liabilities, unusual weakness in the exchange's own token, quietly increasing withdrawal friction, and retreat from products or markets. Above all, limit how much of your total holdings sits with any single counterparty.
It is not uniformly safer; it carries different risks. Non-custodial protocols remove counterparty and insolvency risk because no company holds your assets. They add smart contract risk, oracle risk and irreversible user error. Which is preferable depends on what you are optimising for.