AI shopping assistants disrupt the e-commerce landscape! Bernstein: It may be a wise move for platforms such as Amazon.com, Inc. (AMZN.US) and Uber Technologies, Inc. (UBER.US) to temporarily refuse access.

As AI agents gradually gain the ability to search, shop, and complete transactions autonomously, traditional internet platforms face a key decision: should they actively embrace third-party AI shopping assistants, or restrict their access to protect their own user relationships and business models? In a research report released on October 23, Bernstein pointed out that for internet platforms with relatively high market share such as Amazon.com, Inc. (AMZN.US), DoorDash (DASH.US), and Uber Technologies, Inc. (UBER.US), refusing or restricting third-party AI agent access at this stage may be a completely reasonable business strategy. Bernstein analysts Nikhil Devnani and Mark Shmulik believe that although AI agents are expected to fundamentally change the way consumers search online, discover products, and complete transactions in the long run, the market's current concerns about their disruption of traditional internet platform business models may be excessive. AI agents are still at a very early stage of development, with technical capabilities, consumer usage habits, and commercial cooperation models all yet to mature, and there is no need for existing platforms to rush to help potential competitors build new transaction entry points. The report notes that whether consumers will ultimately be willing to shift from traditional e-commerce and service platforms to AI agents depends not on the technical performance of AI models, but on whether these products can provide a shopping experience clearly superior to existing methods. What consumers really care about is price, trust, convenience, ease of operation, and whether tasks can be completed accurately and promptly. For consumers already accustomed to using mature platforms such as Amazon.com, Inc., AI shopping assistants must not only complete the same tasks but also demonstrate sufficiently clear advantages in reliability and convenience to prompt users to change long-established consumption habits. Bernstein emphasized that existing internet platforms, after years or even decades of development, have established mature brand credibility, product search systems, transaction processes, and user service capabilities. Taking Amazon.com, Inc. as an example, the process for consumers to search for and purchase daily necessities on its platform is already highly mature, and it is not easy for third-party AI agents to replace this experience. At the same time, the difficulty of applying AI agents in different consumption scenarios also varies significantly. The firm summarizes the main uses of consumer AI into three categories: completing tasks for users, purchasing goods for users, and providing entertainment services. Among them, shopping is considered one of the areas where large-scale behavioral change is hardest to achieve. Bernstein noted that even though e-commerce has developed for about 30 years, its penetration rate in the overall U.S. retail market is still only about 20%, and the penetration rate in the global market is even lower. This shows that changes in consumer shopping habits usually take a long time, and the emergence of AI technology does not mean that all transactions will quickly shift to new channels. However, the early performance of Meta's (META.US) AI agent Muse has already shown the potential market size of consumer AI shopping. According to data disclosed by Meta on its Q3 2026 earnings call, Muse attracted 2.5 million daily active users in the North American market in just about a month and a half after launch. More notably, about 25% of Muse's weekly active users have tried using the AI agent for shopping, making it the product's most popular use case. Meta also revealed that the average purchase amount for Muse users exceeds $200, including travel-related spending, and some users with relatively high average revenue per user on the company's platform have also become early adopters of Muse. Bernstein believes these data indicate that AI agent shopping is gaining a certain degree of consumer acceptance. However, it is still difficult to determine whether early trials can long-term stable usage habits. The firm particularly emphasized that user retention and actual use cases will be important indicators for measuring the long-term value of consumer AI products. Even if AI agents can attract a large number of users in the short term, if they cannot continue to provide reliable and convenient services, it may be difficult to maintain user activity. On the technical level, Bernstein believes that one of the key factors determining whether AI agents can truly become widespread is their interoperability with external websites and service platforms. Meta's Muse accesses and operates websites through virtual machine technology, enabling the AI agent to perform tasks in a broader internet environment, and the integration of payment systems also helps enhance consumer trust. However, this approach still has obvious limitations. Bernstein analysts found in actual experience that although Muse could help users quickly find the materials required for passport applications, it also encountered Siasun Robot&Automation verification obstacles on websites such as Yelp and Wayfair (W.US), preventing tasks from being completed smoothly. For ordinary consumers, such problems may directly affect their trust in AI agents. The firm pointed out that shopping differs from general information queries: once an AI agent selects the wrong product, misses an order, or fails to complete a transaction during the purchase process, consumers may not only fail to save time but may also need to deal with additional problems. Therefore, official technical interfaces and commercial partnerships are crucial to the development of AI agents. Compared with directly operating websites designed for human users, completing product queries, order processing, and payments through application programming interfaces (APIs) can usually provide a more efficient and reliable transaction experience. Currently, Muse has established partnerships with companies such as Shopify (SHOP.US), Walmart Inc. (WMT.US), Instacart (CART.US), Expedia (EXPE.US), and Wayfair (W.US). Bernstein believes that for companies with relatively low market share that hope to challenge industry leaders, cooperating with AI agents may be clearly attractive. For example, Walmart Inc. competes with Amazon.com, Inc. in e-commerce, and by connecting to third-party AI shopping channels, it has the opportunity to gain additional traffic, expand market share, and increase revenue. But for platforms such as Amazon.com, Inc. that already have a huge user base and a mature transaction ecosystem, the situation is different. Bernstein noted that Amazon.com, Inc. has already adopted a strategy of restricting third-party AI agent access while actively developing its own Alexa shopping features. The business logic behind this choice is that once third-party AI agents become the main entry point for consumer shopping, traditional e-commerce platforms may gradually lose control over user relationships, the product discovery process, and transaction decisions. By then, consumers may no longer directly visit platforms such as Amazon.com, Inc., but instead compare products, choose merchants, and complete purchases through AI agents. Traditional platforms may be reduced to infrastructure providing goods, fulfillment, and other back-end services. This is precisely the "disintermediation" risk that internet platforms fear. Bernstein believes that for companies such as Amazon.com, Inc., DoorDash, and Uber Technologies, Inc. that occupy important positions in their respective markets, cooperating with third-party AI agents too early may amount to actively helping future competitors build user entry points. This is especially true for platforms that possess unique product or service supply and large amounts of offline resources, which third-party AI agents often find difficult to bypass to complete transactions directly. This supply-side advantage allows existing platforms to retain a certain degree of initiative when negotiating with AI agent developers. Therefore, Bernstein believes that these companies can choose to wait and see for the time being, and then decide whether to open up cooperation after the scale of consumer AI agent usage and the business model become clearer. This strategy can not only delay the potential risk of disintermediation but may also buy time for both sides to establish a more favorable commercial cooperation mechanism in the future. However, Bernstein also cautioned that refusing third-party AI agents does not mean existing platforms can permanently withstand technological change. If consumers ultimately shift to AI agents on a large scale, and external AI shopping channels begin to contribute a considerable proportion of new transaction value in a certain industry, then even market leaders may find it difficult to continue ignoring this channel. For industries such as travel, where the supply of goods and services has already become highly digitalized, AI agents may find it easier to connect directly with airlines, hotels, and other service providers, thereby bypassing traditional online travel platforms. By contrast, in areas such as food delivery and instant retail that are highly dependent on offline merchant networks, delivery capabilities, and platform services, existing aggregator platforms may have stronger defensive capabilities. Bernstein also pointed out that one of the important future competitions in the consumer AI market will take place between general-purpose AI agents and vertical-domain specialized AI agents. General-purpose AI agents seek to become a unified entry point for consumers to handle various online and offline matters, while platforms such as Amazon.com, Inc. can use their own user purchase history, payment information, membership relationships, and logistics services to develop AI shopping tools better suited to specific scenarios. The firm believes that whether Amazon.com, Inc. can successfully promote Alexa shopping features will become an important indicator for observing this competition. If even Amazon.com, Inc., with its huge user base and mature e-commerce ecosystem, cannot drive widespread consumer use of its own AI shopping agent, then smaller internet platforms seeking to build successful vertical AI agents may face even greater challenges. From an investment perspective, Bernstein currently maintains "Outperform" ratings on Amazon.com, Inc., Meta, DoorDash, Uber Technologies, Inc., Instacart, and Wayfair. It gives Amazon.com, Inc. a price target of $320, Meta a price target of $850, DoorDash a price target of $270, Uber Technologies, Inc. a price target of $95, Instacart a price target of $60, and Wayfair a price target of $125. These price targets and ratings are part of Bernstein's existing coverage views and are not new adjustments made by the firm as a result of this AI agent research. Overall, Bernstein believes that the long-term impact of AI agents on internet business models may be profound, but this change will not happen overnight. At a stage when consumer usage habits, technical reliability, and platform cooperation mechanisms are not yet mature, the choice by internet platforms with relatively high market share to restrict third-party AI agent access does not mean rejecting innovation, but may be a reasonable choice to protect their own commercial advantages and gain time for strategic adjustment. But the firm also emphasized that consumers will ultimately determine the industry landscape. If AI agents can continue to provide a shopping experience clearly superior to traditional platforms and gradually become consumers' preferred transaction channel, then existing internet giants must also adapt to this trend, or they will face the risk of losing the user entry point.