Curve New York opens at the Javits Center on August 2, into a wholesale market that has spent this year pulling orders forward and cutting their depth. Neither move is irrational on its own. Together they do not shrink the forecasting problem in intimates and swim, they relocate it: whatever the buyer declines to commit to, someone upstream must still decide whether to build it. The show puts more than 120 lingerie, swim and loungewear brands in a room for three days to sell Spring/Summer 2027. Most will leave counting orders. They should be counting the obligations that arrived with them.

The earlier ordering is tariff-driven, and it is datable. US retailers brought orders from China forward by four to six weeks to land inventory ahead of the tariff step-up, Reuters reported on June 30, citing shipping executives. Maersk told the wire that space on the China-to-US route had been tightening since mid-May on “stronger customer demand and earlier seasonal bookings,” which is the freight market describing a calendar rewritten upstream of it. Sheng Lu, who tracks apparel trade data at the University of Delaware, records the same behavior from the sourcing side: firms now make more frequent adjustments to their original shipping schedules as policy moves. Intimates and swim leave the same ports on the same vessels, and they do not get their own calendar.

The deadline those buyers were racing arrived three days ago. The temporary 10 percent global surcharge levied under Section 122 expired at 12:01 a.m. on July 24 and was replaced in the same minute by forced-labor Section 301 duties. The distinction that matters is not the rate but the clock. Section 122 was capped at 150 days by statute; its replacement expires when an administration decides it does. A buyer who waits is therefore no longer waiting for a cost to lapse, which removes the last structural reason to hold the order.

Not every vendor faces that duty. Goods qualifying under USMCA or CAFTA-DR are exempt from the new Section 301 rate, and four Asian producing countries have a future duty-free quota that has not yet taken effect. Sheng Lu’s trade data put Asia’s share of both cotton and MMF apparel imports near 73 percent, though, which is where most of what fills a Curve booth gets cut and sewn.

The thinning has a survey behind it and a playbook in front of it. NuORDER by Lightspeed’s 2026 State of B2B eCommerce report, drawn from 200 senior wholesale decision-makers, found 54 percent of brands prioritizing cost reduction and margin protection over higher-risk bets. Its authors read the same data as retailers moving away from massive pre-season orders toward smaller, more frequent ones. RepSpark, which sells wholesale software to brands, advises buyers to “pre-book the proven core, and reserve open-to-buy for at-once replenishment,” and warns against “spending the entire budget on pre-book and having nothing left to chase what is actually selling.” As buying discipline that is sound. As an account of who ends up holding the goods it is silent.

The forecast does not disappear when the buyer stops making it.

Every open-to-buy dollar held back is a bet that the vendor will have stock when the retailer wants it. RepSpark states the dependency plainly in the same guide: “At-once depends entirely on the brand having stock available when you want it.” At-once buying, it adds, “reduces markdown risk because you are reacting to real demand rather than a forecast,” which is true, and true only for the party doing the reacting. The unit still had to be cut, shipped and cleared months before anyone reacted to anything, and under the new duty regime it cleared customs at a rate paid on goods no retailer had committed to buy. What the buyer has written is not simply a smaller order. It is a smaller order plus a call option on the vendor’s warehouse, struck at the wholesale price, with no premium attached.

Swim is where that option turns from expensive to unfundable. A bulk restock in swimwear takes six to twelve weeks once placed, against a North American selling season that peaks in spring and summer. An at-once order placed in May against strong sell-through cannot be manufactured into that season. It can only be filled from stock someone speculatively built the previous autumn. The reorder the buyer is holding budget for looks like a production decision deferred. In fact the vendor already made it, alone, without being told its size.

{{generate: A vast dim warehouse aisle stacked to the ceiling with cartons stencilled AT ONCE, each bearing a small customs duty stamp. A chambered nautilus in half-moon spectacles stands at the base with a clipboard, counting. Pinned to the nearest carton is a single sheet headed OPTION AGREEMENT with the buyer’s signature line conspicuously blank. Mood: inventory that belongs to no one who ordered it.}}

Absorbing that decision requires continuous reading of the retailer’s sell-through, and most vendors cannot do it. In the NuORDER survey, 74 percent of brands claimed strong sell-through visibility, which sounds like a solved problem. It is a self-report about access to numbers, not about the systems that would turn those numbers into a cut ticket. Only 9 percent of the same respondents run a fully integrated ERP. We noted this week that retail’s demand models remain fitted to a seasonal curve the calendar has stopped producing; the wholesale version of that failure is quieter, because it has an owner who never volunteered. The distance between those two percentages is the roster of vendors who believe they can see the reorder coming and are in fact reading a spreadsheet emailed at month end.

The strongest case against this reading is that earlier orders help vendors rather than hurt them. More lead time is more planning time, and a pre-book landed in August rather than October buys a real factory slot and a known duty rate instead of a rumored one. For the argument here to fail, two conditions would both have to hold: the smaller pre-book would need to be firm, and the at-once unit would need to cost the buyer more than the pre-booked one. The first is weakened by the very thing that moved the order forward, because an order placed against a customs deadline is placed before the season has taught anyone anything. The second is not how this category prices. In practice, wholesale lists rarely distinguish a pre-booked unit from an at-once one, so the flexibility buyers are advised to protect typically costs them nothing to protect.

Vendors at Curve have two instruments available and are using neither. They can price the option, charging more for the at-once unit than the pre-booked one and letting buyers decide what flexibility is worth. Or they can decline to write it, building only to committed depth and letting the empty June shelf belong to the retailer who chose not to fill it. If the pull-forward holds through the Fall/Winter 2027 buys in February and pre-book depth keeps falling, the vendors still standing in a year will be the ones that made that choice on purpose. The rest will find they made it anyway, in a warehouse, in October, at markdown.