Nike Just Cut Its Online Sellers in China.
In America, It’s Begging Them Back.

What Nike actually did — precisely

Let’s be exact, because the headline version gets it wrong. From 1 January 2027, Nike will terminate the online sales authorisations of its third-party distributors in mainland China — including its two largest, Topsports and Pou Sheng. Online, Nike’s shoes will be sold only through Nike’s own official flagship stores on Tmall, JD.com and Douyin, plus Nike’s own apps. The market reacted instantly: Topsports’ shares fell more than 23% in a day, because its Nike online business was around 22% of its revenue — roughly 5.7 billion yuan, about US$836 million, gone.

But note the two words that matter: online, and China. Offline, Nike is keeping those same partners to run physical stores and local service. This isn’t “Nike will only sell at Nike stores.” It’s Nike pulling the online channel in-house while leaving the physical one with its distributors. That distinction is the whole story.

The paradox: Nike is going both directions at once

Here is what makes this worth writing about. While Nike cuts online distributors in China, in the United States it is doing the exact opposite. Under CEO Elliott Hill, Nike spent this year rebuilding the wholesale relationships it had torn up — rejoining Foot Locker, Dick’s, and Amazon. Its own numbers show the reversal: last quarter wholesale rose 8% while Nike’s own direct sales fell 8%. In America, Nike decided going direct was a mistake and crawled back to the middleman.

So which is it? Is the distributor the future or the past? The answer — and this is the useful part — is that Nike isn’t confused at all. It has simply learned to tell two kinds of middleman apart, and it’s cutting one while re-hiring the other, in the same month, on purpose.

The one rule behind both decisions

Strip it down and the rule is brutally simple: cut the middleman where they only extract; keep the middleman where they create.

In China’s online channel, the third-party distributors had stopped creating and started extracting. Thousands of storefronts meant price chaos, constant discounting that wrecked Nike’s full-price sales, counterfeits, and a fragmented brand experience — all while taking a margin. Online, they weren’t bringing Nike customers Nike couldn’t reach itself; the platform (Tmall, Douyin, JD) already delivers the customers. The distributor had become a toll booth on a road Nike could pave itself. So Nike cut them.

Offline is the opposite. A physical store does real, hard, local work: staff who fit shoes, service, returns, presence in a mall in a second-tier city, relationships Nike can’t run from headquarters. That middleman still creates value Nike can’t easily replace — so Nike keeps them. Same company. One principle. Two answers.

And this reframes the American reversal too. Nike’s earlier mistake wasn’t going direct — it was going direct everywhere at once, cutting the creating middlemen along with the extracting ones. Hill’s correction and the China cut are the same lesson learned from both sides: be surgical about which middleman you remove.

Why the online middleman is the one that goes first

This is where your instinct about AI commerce is exactly right. The reason Nike can pull the online channel in-house now, when a blanket direct push failed a few years ago, is that the online distributor’s real job — discovery, getting the product in front of the right buyer — has been automated away from under them.

The platform’s recommendation engine, the brand’s own data, and increasingly AI shopping agents now do the finding. When discovery is handled by the algorithm and, soon, by an AI agent buying on a customer’s behalf, the online reseller’s only remaining contribution is markup and mess. That’s why the online middleman is the first to be cut in every category, not just Nike’s: it’s the function AI replaces earliest. The offline middleman — the human in the store — is the last, because service and physical trust are the hardest things to automate. Margin recapture is the motive; AI-driven discovery is what finally makes it possible.

Coming to Malaysia: which side of the razor are you on?

This lands directly on Malaysian distribution, and sooner than the comfortable timelines suggest. If you distribute a global brand here, the question is no longer “will they go direct?” It’s “which of my functions can they now do without me?”

If you are an online reseller whose value is buying stock and listing it on Shopee or Lazada at a markup, you are Topsports’ online business. The platform already does your discovery; an AI agent will soon do the rest; and the brand can see it. You are the toll booth on the road they’re about to pave. If you are a distributor who does real local work — service, fitting, credit, physical reach, after-sales, genuine market cultivation — you are Topsports’ offline business, and you are far safer, because that’s the function AI replaces last.

The survival move isn’t to fight the brand going direct. It’s to be the middleman that creates rather than the one that extracts — and to know, honestly, which one you are before the brand decides for you.

My bet: the American reversal is timing, not strategy

Here is my own read, and it’s a prediction, so weigh it as one. I don’t believe Nike’s return to wholesale in the United States is the destination. I think it’s a bridge.

Look at the sequence. Nike dismantled its distribution and went direct before the machine that replaces the distributor’s real job — discovery — was actually ready. Recommendation engines helped, but the true replacement, an AI shopping agent that finds and buys on a customer’s behalf, hasn’t arrived at scale yet. So Nike got the order wrong: it removed the middleman before the thing that does the middleman’s work existed, and had to retreat. The American wholesale comeback isn’t proof that direct was wrong. It’s proof it was early.

China is the template Nike is running while it waits. And once AI agents mature — once discovery genuinely works without a human reseller in the loop — I expect Nike to do globally what it’s doing in China’s online channel: reclaim direct, everywhere online. In my view the wholesale partners it’s rehiring in America today are a temporary bridge, carrying the online business until the technology that makes direct work finally lands. Offline stays with partners far longer, for the reasons above. But online, I read the current US reversal as a pause, not a change of direction.

If I’m right, the lesson sharpens: don’t read Nike’s wholesale comeback as proof the online middleman is safe. Read it as proof the online middleman got one more cycle — because the tech wasn’t ready yet. It’s getting ready.

And it might still fail — because I won’t pretend it’s a sure thing

I’m not going to tell you Nike’s China bet is guaranteed. Cutting US$836 million of distribution overnight is a real gamble; Topsports did more online than pure extraction — membership operations, service, scale — and Nike now has to do all of that itself, well, at once. Plenty of analysts think it could bleed before it heals, exactly as the blanket US direct push did. The rule is sound; the execution is where companies die. Which is the same warning I’d give any Malaysian brand tempted to copy the headline: cutting the middleman is easy; replacing what the good ones actually did is the hard part everyone underestimates.

The bottom line

Nike firing its online middleman in China while rehiring its offline one in America isn’t a contradiction — it’s the clearest map we have of where distribution is going. AI takes the online middleman’s job first because discovery is what it automates first. The physical middleman survives longest because service is what it automates last. Every distributor, and every brand, is about to be sorted onto one side of that line.

Published by IMA AI — July 2026. We build the rails for agent-era commerce — verified identity and direct brand-to-customer transactions — which is exactly the online layer where the extracting middleman disappears. We also work with the distributors who create, because AI hasn’t replaced that yet.