Where things stand now
The current position is easier to state as a list than as a paragraph. The ledger holds twenty-two platform objects. Two of them, ACBuy and AllChinaBuy, are two names for the same shopping operation and both are recorded as closed. PandaBuy, which shut permanently after the 2024 case, sits in the same group as an earlier example of the identical pattern. Mulebuy is filed as a historical record. Eleven objects carry a directory page, and seven more are listed without any verifiable operating claim either way.
The measured part of the picture is narrower and cleaner than the paperwork. In the 2026-W39 window, thirty-one wrapped addresses were collected from public sheet pages and requested as they were written. All thirty-one failed at the network layer, which means the wrapper itself did not answer at all. Rebuild those same thirty-one into the source addresses they encode and all thirty-one returned a normal response. The status counts printed on the weekly snapshots page for that period read ninety-three resolves, zero moved, zero gone, zero shielded, zero unrecorded.
Two account-level questions stay open in the ledger rather than being answered by assumption. The first is whether balances held when purchasing stopped will be settled for every account, and on what timetable. The second is whether warehouse items still sitting under account-specific deadlines will be released or processed. Both were raised in platform notices and neither has a published resolution, so the ledger records them as open questions instead of guessing at an outcome. They are recorded as open because they are open, which is a status rather than a hedge.
One word in the ledger needs its definition kept close by. Closed here means the observed behaviour of the service: no ordering since June 2026, no working storefront since August. It does not mean a company statement was issued, because none was. That distinction matters for anyone still waiting on an announcement before deciding what to do with a sheet. The behaviour has been settled for months. Anyone waiting for a press release before rebuilding a sheet is waiting on something that has not been promised.
The three forces at play
The first force is the wrapper layer, and it belongs to the platform. A wrapped address was only ever a doorway that a platform built and controlled. When the platform stops serving, the doorway stops opening, and nothing the holder does will make it answer again. Thirty-one of thirty-one failures in this period is what that looks like when it is measured rather than described. The force is already spent, and it is not coming back through any action available to a buyer.
The second force is the marketplace listing, and it belongs to Weidian, Taobao or 1688 rather than to any agent. Those listings were never hosted by ACBuy and were never deleted because ACBuy stopped. The rebuild test shows the effect plainly: thirty-one source addresses that had been sitting behind dead wrappers replied normally once they were requested directly. The catalogue survived the platform that was pointing at it, which is the single most useful fact in this entire file. That is the part of the story worth keeping when everything in front of it looks broken.
The third force is the record the holder keeps, and it is the only one of the three that responds to effort. It covers the sheet as saved, the notes beside each row, the rebuild of each address, the order and warehouse evidence kept from before June, and the deadline attached to anything still in storage. The carry-over steps on the platform ledger pages exist for this force alone. The first two forces move without permission. The third one moves when someone works on it.
The three forces also move on different clocks, which is the practical reason a weekly check beats an annual one. The wrapper layer died once and stays dead. The marketplace layer changes for ordinary marketplace reasons, an item delisted here, a listing moved there, and it is the only layer where a status can shift between periods. The holder layer changes whenever anyone opens the file. The snapshot exists so the moving layers get a fresh reading on a schedule instead of whenever somebody remembers.
Three ways this could go
The quiet route is that nothing more is ever announced. Ordering never resumes, the storefront stays down, and the platform simply stops being part of the present tense. That is the least dramatic option and also the one the observed record already matches most closely, since no statement has appeared in the months since ordering stopped. For a sheet holder it changes nothing: the wrapper stays dead and the rebuild stays the only route to the item. Nothing about that route requires a decision from the reader today.
The partial route keeps one business alive and the other closed. Something like this has already been documented once, when the June notices suspended purchasing from 12 June to 30 June while warehouse and parcel operations continued under the same announcement. A version of that arrangement lasting indefinitely would leave accounts with storage deadlines, shipping options and unresolved balances, but no way to place a new order. Deadlines would be the live question under it, and those sit on the account rather than on the storefront.
The reinstatement route would mean a relaunch, either under the same name or under a new one. Even then, previously issued addresses would only start working again if the wrapper format were preserved exactly as it was, including the source marker that tells a rebuild where the listing lives. This ledger holds no evidence pointing either way, and a successor brand reusing an old name would still have to prove that its addresses match the old format before any sheet benefits.
None of the three routes is a forecast, and the ledger does not publish one. What each route does have is an observable marker that cannot be faked for long: a storefront that answers, an official statement with a date in it, or a previously issued wrapped address that resolves on a fresh check. Until one of those appears, the honest reading of the record is that the platform is closed on behaviour and unannounced in writing. Everything else is speculation dressed up as a timeline.
What to watch and when
Four markers are worth a look whenever the question comes up again. Does the storefront answer at all. Has any official statement appeared, on the platform or through the representatives who spoke in July. Has the status mix in the weekly snapshot changed from the current all-resolves pattern. Does a wrapped address from an old sheet resolve on a fresh check. Three of the four are observable in under a minute, and the fourth is printed as a dated status on the platform ledger page.
The checking method page sets out the parts that stay fixed between checks: an eight-second window per address, a concurrency limit of two per host, no cookies and no account credentials, and only the response status plus the host it landed on recorded. Those settings are published so a reader can tell what a status does and does not cover. A resolve is a reading taken at a moment, not a promise about next month, and not a stock notice about the item.
Meanwhile the work that pays off regardless of which route wins is the same work. Rebuild the addresses in the sheet to their source listings. Open each rebuilt address once to confirm it answers. Keep the original wrapped address in the notes so the trail stays readable. Paste only rebuilt addresses into an order box at a platform taking orders today. None of that depends on an announcement, which is why it is worth doing before one arrives. It also turns a saved file into a checked one, which is the difference between a record and a guess.
A single observation would overturn the current reading, and it is worth stating what it looks like: a fresh check in which previously dead wrapped addresses resolve again, or a dated statement describing what happens to balances and stored items. Neither has appeared. Until one does, the pattern from PandaBuy in 2024 remains the closer precedent, and it points the same way this period numbers do: the wrapper is gone, the catalogue is not. Thirty-one rebuilds that answered and thirty-one wrappers that did not is the whole case in one line.