PRIM3 Brief #23: The Listing Ladder Lost Its Middle Rungs

CoinEx stops spot trading on September 29, nine days from now. The exchange published its cessation notice on September 14, 2026, gave users until December 22 to withdraw, and said it will buy back CET without a quantity cap at 0.005 USDT — the price the token was issued at nine years ago (CoinEx wind-down summary, The Crypto Times). Reserves are above 100%. Nobody lost money. The coverage has been almost entirely about withdrawal deadlines.

We think the withdrawal story is the small one.

CoinEx is the fourth mid-tier venue to publish a wind-down calendar this quarter. AscendEX stopped operating on July 1, BitMEX confirmed its exit on July 23, BitMart followed on July 26 (MarketScreener). Four venues in ninety days, all in roughly the same weight class, all citing the same arithmetic. And the entire standard TGE sequence that Web3 founders have run since 2021 — launch, list on a tier-2 or tier-3 book, accumulate volume, use that volume to argue for a tier-1 listing — assumes that weight class exists.

Key takeaways

• Four mid-tier centralized venues wound down between July 1 and September 29, 2026. CoinEx was carrying about $81.0 million in daily spot volume and roughly $145.6 million in reported assets when it called it, which is a solvent business that stopped covering its own compliance and security overhead.
• CET is being retired at its issue price. An exchange token's terminal value turned out to equal its starting value, which is the cleanest available lesson about tokens whose distribution depends on a venue surviving.
• The founder consequence isn't fewer places to list. It's that a centralized listing in 2027 is a recurring liquidity cost rather than a growth event, and almost every TGE budget we review still books it as the latter.

What Actually Closed, and Why It Wasn't Insolvency

None of the four venues failed the way exchanges failed in 2022. CoinEx published a staged calendar, kept withdrawals open for three months past the end of trading, and reported reserves over 100%. Changpeng Zhao noted on September 15 that this winter's wind-downs have let users withdraw, unlike the frozen-account failures of prior cycles. That's a real improvement and it's worth saying plainly.

But solvency was never the binding constraint. CoinEx was clearing roughly $81.0 million of spot volume a day against a cost base built for a global user footprint — KYC and travel-rule tooling per jurisdiction, sanctions screening, listing review, custody, proof-of-reserves infrastructure, 24/7 security monitoring. Those costs scale with the number of countries served, not with volume. When retail spot activity thins out, fee income falls and the compliance floor doesn't move. The company can be fully backed and still uneconomic. CoinEx's own notice said compliance requirements had "exceeded reasonable boundaries," which is a polite way of describing a fixed-cost problem.

Meanwhile the flow kept concentrating upward. Binance alone accounted for more than 40% of tracked centralized volume in August 2026. On the other side, DEX spot volume hit a record 24% of centralized exchange volume in July 2026, up from 6.0% in January 2021 (CoinGecko), and Hyperliquid now clears roughly 44% of all on-chain perpetuals. Liquidity left the middle in both directions at once.

So the thing that died wasn't a set of companies. It was a tier.

Why the Consensus Read Is Too Narrow

The standard framing this month treats these closures as a bear-market body count. More than 60 crypto companies and projects have announced closures in 2026, and RootData counts 99 project deaths year to date. Read that way, CoinEx is one line in a spreadsheet about a tough cycle, and the implied conclusion is that the survivors inherit the volume when conditions improve.

What that misses is that the mid-tier exchange wasn't only a trading venue. For an early-stage token, it was the distribution channel — the place a new asset went to acquire its first non-airdrop holders, its first price history, and the third-party volume print that every subsequent listing application and every exchange BD conversation was built on. Tier-2 and tier-3 books were where a token became legible to the rest of the market.

That's the second distribution channel to close on founders this quarter. We wrote in Brief #20 about digital asset treasury companies mattering to teams as a way to place supply without touching an order book, and about that route narrowing to bitcoin. The mid-tier CEX is the same category of loss, and it's larger, because far more projects used it.

Neither channel is coming back on the old terms. Both were subsidised by a market structure that no longer exists — in the DAT case by a persistent mNAV premium, in the exchange case by retail spot volume broad enough to fund a long tail of venues.

What the Listing Math Looks Like Now

Price the sequence out and the problem shows up immediately.

A tier-1 listing is commonly quoted between $150,000 and $500,000 and up, frequently settled in tokens rather than cash (MotionTrade listing cost breakdown). Tier-3 venues quote nearer 50,000 USDT and will negotiate. Then the recurring line: a professional market maker runs roughly $5,000 to $50,000 a month per venue in retainer, and wants $100,000 to $500,000 of token and stablecoin inventory to hold a credible book. Every additional pair multiplies the retainer and the inventory. None of it is one-time.

Under the old ladder that math worked because the cheap rungs did real work. You paid 30,000 USDT and a small retainer, you got a price feed, a CoinMarketCap listing, an order book a KOL could point at, and six months of history. Then you took that history to a larger venue. The tier-2 fee was an acquisition cost for credibility, and credibility compounded.

Take those rungs away and the same spend buys something much thinner. A tier-3 listing in late 2026 gets you a book on a venue whose own survival you're now underwriting, at a venue-level daily volume that may be a fraction of what your token already does on a decentralized exchange. Of the token launches we've reviewed at PRIM3 over the past year, the ones that held their ranges best treated the first centralized listing as a distribution cost with an ongoing retainer attached, not as the moment the project arrived. The ones that struggled had modelled listing spend as capex and discovered in month four that it was opex.

And the market maker arrangement is where this compounds badly. A loan-structured deal hands inventory to a counterparty against a strike, and when the venue itself is winding down, that inventory has to be unwound into a book that's being switched off on a published date. We've seen enough of those unwinds to have a view. Plan the exit at the same time you plan the listing.

What This Means for Founders

Start with the venue, not the fee. Before any listing conversation goes past a first call, run the venue through the same diligence you'd run on a counterparty: published daily volume, proof-of-reserves cadence, which licences it actually holds in which jurisdictions, and how much of its book is its own incentive programmes. A venue doing well under $100 million a day across a wide geographic footprint in 2026 is carrying the exact cost structure that just closed four of its peers. That's not a prediction about any specific exchange. It's a reason to size the cheque as if the pair might not be there in eighteen months.

Then flip the sequence. The advice that a DEX-first launch beats a paid tier-3 listing has been circulating all year, and it's correct now for a reason most versions of it skip: on-chain venues don't publish wind-down calendars. Launch on the dominant DEX for your chain, build three to six months of verifiable trading and holder history, and take that data to a centralized venue from a position where you're bringing volume rather than buying it. In our portfolio, Bondex built its distribution around on-chain activity it could evidence, which is the posture that travels well in this structure.

Budget the listing as a subscription. If a pair costs $8,000 a month in retainer to keep honest, that's roughly $96,000 a year of runway committed to one venue — and the question of whether that pair is producing anything needs to be asked every quarter, not once at TGE. Kill pairs that don't. Most projects never do, and quietly fund dead books for years.

Last, keep token distribution structurally separate from any single venue's health. That's the CET lesson, and it's not a narrow one. CET was a well-run exchange token backed by a solvent business with over 100% reserves, and it's being retired at the number it was issued at, because its utility ended with the venue. A launch token whose float, price discovery, and holder base all route through one centralized book inherits the same dependency, and we flagged the design-versus-market-event distinction in Brief #16. It applies here too.

Where PRIM3 Is Placing Bets

PRIM3 has worked on TGE and listing strategy across 40+ token launches, so this one is close to the operating work rather than the commentary.

Three things we weight heaviest now when we underwrite an early-stage token-issuing team. First, whether the token's distribution plan survives its primary venue disappearing — if the answer needs a slide, it doesn't. Second, whether liquidity provisioning is modelled as a recurring operating line with a named owner and a quarterly review, rather than a TGE-week expense. Third, whether there's revenue that doesn't route through the token at all, because that's the part of the business a venue closure can't reach.

We're more interested this quarter in the infrastructure underneath launches than in the launches themselves. On-chain order books, liquidity routing, market-making automation that doesn't need a $500,000 inventory commitment per venue, and settlement rails like Kima Network that treat chains as interchangeable. The volume left the middle of the centralized market. The tooling that lets a small team run credible liquidity without a mid-tier CEX has not been built out to match, and that gap is where we think the next set of cheques goes.

One more thing worth watching. We argued in Brief #5 that trust incidents were driving a sustained volume shift on-chain. What's happening now is quieter and more permanent than a trust event, because a venue that exits cleanly doesn't come back when sentiment improves.

That's where PRIM3 is placing capital this quarter. More on our thesis at prim3.vc.

FAQ

Should a Web3 project still list on a tier-2 exchange in 2026? Only when the venue clears a solvency and licensing test the project runs itself. Four mid-tier venues wound down between July 1 and September 29, 2026. A tier-2 or tier-3 listing now buys a maintained order book rather than a distribution event, and it should be underwritten against the venue's own daily volume before any fee is paid.

What does a CEX listing actually cost a token project in 2026? Tier-1 venues commonly quote $150,000 to $500,000 or more, often settled in tokens. Tier-3 venues quote closer to 50,000 USDT and negotiate. On top of that, a market maker charges roughly $5,000 to $50,000 a month per venue and needs $100,000 to $500,000 in inventory to hold a book. The retainer and the inventory recur.

Is a DEX-first token launch better than a CEX listing in 2026? For most early-stage projects, yes. DEX spot volume reached a record 24% of centralized exchange volume in July 2026, and on-chain venues don't publish wind-down calendars. A DEX-first launch also produces the verifiable history that a later centralized application gets built on, which reverses the old sequence of paying for a listing to manufacture that history.