Amazon and Walmart Are Racing to Control Retail's AI Decision Layer

By Hector Herrera | June 2, 2026 | Retail

Nearly half of all online shoppers now use AI tools at some point during their purchase journey, and ChatGPT's share of product research has jumped from 2 percent to 30 percent in just two years. Amazon and Walmart have each identified what that shift means: whoever controls the AI system that determines what consumers see, consider, and ultimately buy will own retail's most valuable competitive position in the coming decade. That race is now producing real strategic divergence between the two largest US retailers.

What the "Decision Layer" Actually Is

The phrase "decision layer" refers to the AI system that mediates between a consumer and their purchase. In the Google Search era, that layer was the search algorithm — the system that decided which products appeared for which queries. Today, the decision layer is fracturing.

According to PYMNTS analysis, consumers are now beginning purchase journeys across a range of AI surfaces: Amazon's own search and Rufus assistant, Google Shopping, ChatGPT, Perplexity, Gemini, and individual retailer apps with embedded AI. The 2 percent to 30 percent shift in ChatGPT product research is the most striking data point — it represents consumer purchase intent that neither Amazon's algorithm nor Walmart's marketing spend currently controls.

That's the gap both companies are sprinting to close.

Amazon's Strategy: Own the Entire Stack

Amazon's approach is built around vertical control — keeping consumers inside the Amazon ecosystem by giving them better AI tools than any external competitor can offer.

Key investments:

• Rufus — Amazon's embedded AI shopping assistant, launched broadly in 2024 and expanded significantly in 2025, allows consumers to ask conversational questions about products, get comparisons, and receive recommendations without leaving the Amazon app. The goal is to make leaving Amazon unnecessary during a purchase decision.

• AI-native search — Amazon has rebuilt its product search to handle large language model-style conversational queries, not just keyword matching. A consumer who types "what's the best espresso machine for a small kitchen" now gets a different experience than one who types "espresso machine."

• Fulfillment AI — Automated warehouse robotics, AI-driven delivery routing, and predictive inventory positioning are compressing delivery windows and costs in ways that reinforce why consumers start their search on Amazon in the first place. Speed and reliability are themselves part of the decision layer.

The underlying logic is containment: if Amazon's AI is more useful than ChatGPT for product decisions, users have no reason to leave. Amazon doesn't need to control ChatGPT — it needs to make ChatGPT irrelevant to the Amazon shopping experience.

Walmart's Counter: Openness and Partnership

Walmart's strategy is structurally different. Rather than trying to contain the purchase decision inside a single ecosystem, Walmart has built partnerships that place its products across every major AI surface — betting that the decision layer will remain fragmented and that presence across all surfaces matters more than dominance in any one.

Key partnerships and moves:

• Gemini integration — Walmart partnered with Google to allow Gemini to surface Walmart products within AI-powered search, discovery, and shopping flows. A consumer asking Gemini for product recommendations can be routed to Walmart inventory directly.

• Microsoft Copilot — Walmart's product catalog is accessible through Microsoft's Copilot ecosystem, reaching enterprise users and Windows users who may never visit Walmart.com directly.

• Store-based fulfillment as the structural moat — Walmart's physical footprint, with locations within 10 miles of roughly 90 percent of the US population, gives it a logistics advantage that no purely digital competitor can easily replicate. Same-day delivery at scale is a capability Amazon has invested billions to approach; Walmart has it structurally built in.

Walmart's bet: the AI decision layer will not consolidate around a single winner the way search consolidated around Google. In a fragmented AI landscape, being findable and purchasable across every major AI surface is more valuable than owning any one of them.

The Numbers Behind the Race

The financial stakes explain the investment intensity. US e-commerce is a multi-trillion-dollar market, and AI-assisted purchase journeys show measurably higher conversion rates than traditional search-driven flows. For both Amazon and Walmart, even a small shift in AI-mediated purchase share translates to billions in annual revenue.

The 2-to-30 percent ChatGPT product research jump matters because it is purchase intent flowing through a channel neither retailer fully controls. When a consumer asks ChatGPT "what's the best air purifier for allergies," the answer is shaped by OpenAI's training data, its commercial partnerships, and its content deals — not by Amazon's or Walmart's marketing. Both retailers are treating this as a structural leak in their respective decision layers.

The Amazon-Walmart race is also being watched closely by mid-tier retailers who lack the resources to build proprietary AI at scale. For those retailers, the outcome of this battle will determine which platforms they need to partner with to remain visible in AI-mediated shopping — a dynamic that could consolidate retail AI into a few dominant surfaces faster than expected.

What to Watch

Watch Amazon's response to ChatGPT's product research growth — the company may accelerate Rufus capabilities and pursue AI partnership deals that ensure Amazon products appear prominently in third-party AI recommendations. The OpenAI-Amazon relationship is one to track specifically; an exclusive or preferred placement deal in ChatGPT product search would be a significant defensive move.

On Walmart's side, watch the Gemini partnership for depth signals: if Gemini can handle end-to-end Walmart checkout — not just discovery but transaction — it would be a meaningful proof-of-concept for the federated AI decision layer model.

The industry question: does the decision layer consolidate (as search did, around Google) or remain fragmented across multiple AI platforms? That answer will define retail competitive dynamics through the end of the decade.

Source: PYMNTS, June 2026