Agentic AI Is About to Hand The Power Back to Brands

The standard narrative about AI and retail is that AI will disrupt brands by commoditizing search, making product discovery algorithm-driven rather than brand-driven, and shifting power further toward the platforms that control consumer attention.

Bernstein SocGen analysts published a note to clients this week challenging that narrative directly.

Their argument, as reported by Investing.com on May 17, 2026, is that agentic AI will be net positive for apparel brands, and net negative for the multi-brand retailers that currently carry them. The disruption is real, but its direction is the opposite of what most brands fear.

The Current Model and Its Costs

To understand why agentic AI helps brands, it helps to understand what the current model costs them.

When an apparel brand sells through a multi-brand retailer like Macy's or Dick's Sporting Goods, it receives approximately 50-60% of the actual gross merchandise value of its products.

The retailer captures the remaining 40-50%. On top of that margin concession, brands pay for marketing and placement fees on those retail sites, currently running 15-25% of traffic, to ensure their products are discoverable within the retailer's ecosystem.

In exchange for giving up margin and paying placement fees, brands get two things: reach to a wider customer base, and discovery that makes future sales possible. The multi-brand retailer is, in effect, renting brands to their audience.

For premium brands the cost of this arrangement is even steeper. The delta between receiving 100% of GMV through a DTC channel and 60% of GMV through wholesale is proportionally larger at higher price points, more than enough to cover the higher operating costs of running a direct-to-consumer business.

Bernstein estimates the operating margin delta between DTC online and wholesale is 10 percentage points or higher for most brands, and 15-20 percentage points for premium brands.

What Agentic AI Changes

The current model depends on one assumption: that consumers begin their product search and discovery journey on multi-brand retail platforms. Google search leads to Amazon. Department store websites are where shoppers browse. Multi-brand retailers own the top of the funnel.

Agentic AI disrupts that assumption at its foundation.

As AI agents become the primary mechanism for search, discovery, and purchase decision-making, a transition Bernstein analysts expect to be the norm for most US adults by the 2030s, the starting point of the consumer journey shifts from multi-brand retail platforms to AI agents.

And AI agents, unlike multi-brand retailers, have no structural incentive to route consumers to Macy's or Dick's rather than directly to a brand's own DTC site.

When an agent helps a consumer find the right running shoe, it is optimizing for the consumer's stated preference, not for the placement fees a retailer has paid to ensure its platform gets the traffic.

A brand with a strong DTC presence, competitive pricing, and good product data can be surfaced directly by an agent without ceding margin to an intermediary.

For brands, the implications compound. Selling DTC rather than through a multi-brand retailer means receiving 100% of GMV instead of 60%. It means owning the customer relationship and the data that relationship generates.

It means controlling pricing, merchandising, cross-sell, upsell, and the transaction itself. And it means choosing where to allocate marketing spend across multiple acquisition channels based on fully loaded costs, conversion rates, and basket sizes rather than being structurally dependent on any single intermediary.

Bernstein analysts put a number on the magnitude of the shift. If 10% of consumers use agentic search that routes them to brand DTC sites instead of multi-brand retail, brands see approximately 150 basis points of tailwind to operating margin.

The retailers on the other side of this equation are the multi-brand intermediaries that have built their business model around being the discovery layer for brands, and capturing 40-50% of GMV as the price of that discovery.

Macy's, Dick's Sporting Goods, and Amazon's marketplace business are the clearest examples. They currently provide brands with reach and discovery in exchange for margin.

If agentic AI provides that reach and discovery at a lower cost, or routes consumers directly to brand DTC sites without requiring retailer intermediation, the value proposition of the multi-brand retail model weakens significantly.

According to Bernstein analysts, agentic AI will make transactions routed through agents more margin-accretive for brands compared to those made via wholesale.

The traffic brands currently pay multi-brand retailers to deliver will increasingly be delivered by AI agents instead, and AI agents will compete for the same marketing dollars that multi-brand retailers currently capture.

For DTC brands and direct-to-consumer retailers the picture is more complex. They face the same traffic displacement risk as multi-brand retailers, but without the margin capture advantage that made the multi-brand model attractive in the first place.

Bernstein analysts describe agentic AI as the primary mechanism for search, discovery, and purchase decision-making for most US adults by the 2030s, a five to seven year horizon.

That timeline gives brands a meaningful window to build the DTC infrastructure and data capabilities that agentic traffic will require.

It also gives multi-brand retailers a window to develop a response, though what that response looks like in a world where AI agents route consumers directly to brand sites is not yet clear.