Key Points

• AI Shopping agents threaten the lucrative retail advertising business
• Amazon and Walmart dominate a market worth $74.4 billion
• Automated purchasing may change the control over product recommendations.

Amazon and Walmart have built massive advertising businesses around a simple idea: companies are willing to pay to appear in front of customers when they are about to spend money. But what if customers no longer browse on their own, and instead let artificial intelligence make purchasing decisions for them?

This issue is now threatening a market in the United States that is estimated to be worth $74.4 billion annually. According to a survey by Reuters on November 4th, Amazon and Walmart together control approximately 88% of retail media advertising spending in the U.S.

The potential impact comes from AI shopping agents. These tools are capable of comparing products, evaluating prices, and completing purchases without requiring customers to scroll through sponsored lists.

Why does Amazon make money before customers even buy anything?

When someone searches for headphones on Amazon, the products that appear first are not necessarily the cheapest or those with the highest ratings. Some of them are sponsored listings purchased by manufacturers, with the aim of attracting buyers.

The value of these advertisements lies in the fact that they appear when consumers already have the intention to make a purchase.

Retailers can earn advertising revenue regardless of whether a particular click ultimately leads to a sale, depending on the advertising arrangement.

Amazon has turned this model into a considerable business. Advertising typically accounts for about 10% of its total sales, but by the third quarter of 2026, the growth rate slowed down to 11%.

When customers leave the decision-making regarding purchases to software, economic logic becomes more complex.

Imagine you ask an AI assistant to help you buy a pair of the best wireless headphones for less than $150. Instead of browsing dozens of products, the assistant would first compare the specifications, availability, and prices before selecting one product.

Customers may even complete a purchase without being able to see any advertisements that the retailer has previously sold.

Why does Amazon block AI proxies, while Walmart welcomes them?

Amazon and Walmart have adopted different approaches to this potential impact.

In September, Amazon blocked the access of Muse shopping proxies to its marketplace by Meta, citing that the software was operating without authorization and raising concerns regarding customer data.

The company also challenged the shopping agents of Perplexity in court, although a previous ban had been overturned on appeal.

Meanwhile, on October 21st, Walmart launched infrastructure that allows third-party shopping agents to interact with customer accounts under clear permissions.

These different strategies reflect the broader challenges surrounding proxy payments: retailers must decide to what extent they should grant control to external software while protecting customer relationships.

Morgan Stanley estimates that by 2030, AI agents may handle 10% to 20% of U.S. e-commerce spending.

However, this prediction does not mean that advertising revenue will decline in the same proportion.

Will advertisements become “invisible” to consumers?

A more likely response from the industry might be to change the way ads reach buyers, rather than completely abolishing ads.

Amazon has already integrated ads into its own AI shopping tool, while Walmart is testing sponsored content within its shopping assistant Sparky.

Brands may increasingly pay to have their product information taken into consideration by the AI system, thus forming a new form of commercial influence.

Companies such as Procter & Gamble, Nestle, and L'Oréal have already begun to adjust their product descriptions to adapt to the discovery methods driven by AI.

This shift has brought about new issues of transparency, especially in cases where autonomous purchasing involves financial decisions and human intervention is limited.

AI Financial agents also face similar challenges, where permissions and accountability may be even more important than the complexity of the software itself.