AI agents compete for the $30 trillion consumer market! Morgan Stanley: Meta(META.US), Apple Inc.(AAPL.US), Amazon.com, Inc.(AMZN.US) and others are poised to benefit; Alphabet Inc. Class C(GOOGL.US) faces a business model test

Morgan Stanley has released a new research report stating that with Meta Platforms (META.US) launching its Muse agent and AI agents such as Instinct rapidly gaining market attention, "agentic AI" is evolving from a simple information Q&A tool into an "action-oriented assistant" capable of searching, comparing, making decisions, and even completing transactions on behalf of users. Morgan Stanley estimates that the global consumer spending that could potentially be digitized through agent products in the future amounts to approximately $30 trillion, covering e-commerce, travel, advertising, food delivery, ride-hailing, and other sectors, with its potential market size roughly comparable to the market opportunity for generative AI in enterprise knowledge work. Morgan Stanley believes that this agent wave may not only create new traffic entry points and transaction channels but also redistribute the profit pools that the internet industry has long established. Platforms with massive user distribution channels, unique data, and software-hardware ecosystems are in a relatively favorable position, with Meta, Apple Inc. (AAPL.US), and Alphabet Inc. Class C (GOOGL.US) possessing clear foundational advantages; among downstream companies, Amazon.com, Inc. (AMZN.US), eBay (EBAY.US), and Shopify (SHOP.US) stand out for their positioning in the agent era. At the same time, industries with higher standardization and where consumers place greater emphasis on price, such as ride-hailing, face potential disruptions that warrant closer attention. It is worth noting that Morgan Stanley remains optimistic about the overall outlook for the North American internet industry and lists Meta, Apple Inc., Amazon.com, Inc., eBay, and Shopify as a group of companies it considers favorably positioned in the agent wave. **Meta Muse Becomes a New Variable: Every 100 Million Users Could Contribute About 1% Incremental EPS by 2028** This research report pays particular attention to Meta's newly launched Muse. Morgan Stanley believes Muse is a significant development in the agent market: after launch, it ranked in the top five of app download charts for nine consecutive days, can leverage Facebook, Instagram, and other Meta data resources to provide personalized interactions, and has integrated mobile web browsing and operation capabilities into usage scenarios, enabling it to directly execute certain transaction tasks through the browser. At the same time, Muse offers a relatively high free usage quota100 million free tokens per weekwhich Morgan Stanley estimates may be sufficient to support users in making more than 100 queries per day. Compared with relying solely on subscription fees, Morgan Stanley believes the commercialization opportunity truly worth watching for Muse may come from user behavior and transactions themselves. Currently, Muse can charge through a subscription model, but the firm expects Meta may ultimately charge commissions on transactions facilitated by Muse. Morgan Stanley also conducted a quantitative analysis of this business model. Assuming that by 2028 Muse has 100 million users, each user initiates 5 queries per day, 10% of which are monetizable queries, and Meta generates approximately $0.07 in revenue per commercial query, then Muse could contribute approximately $1.3 billion in revenue and about $0.35 in incremental earnings per share annually for Meta, equivalent to adding about 1% to 2028 EPS. This model has considerable scale effects. Morgan Stanley's sensitivity analysis shows that if Muse's user scale and the number of commercial queries generated per user increase further, its transaction-related revenue could reach over $8 billion; under a more aggressive scenario, Muse could even bring more than 15% incremental upside to Meta's 2028 EPS. However, this result is highly dependent on assumptions such as user scale, commercial query frequency, and monetization efficiency, and is not the firm's base case forecast. **The Key to Agent Competition Is Not Just the Model: Distribution Capability and Exclusive Data Become Core Chips** Morgan Stanley believes that two important factors determining the final competitive landscape of horizontal AI agents are scaled distribution capability and large-scale unique datasets. This is also why Meta, Alphabet Inc. Class C, and Apple Inc. are considered by the firm to have a solid foundation. Apple Inc. and Alphabet Inc. Class C respectively control the iOS and Android ecosystems, and their deep integration of software and hardware makes it easier for both companies to deploy agent applications and functions across the entire device ecosystem; Meta's differentiated advantage comes from the large amount of first-party data accumulated by platforms such as Facebook and Instagram, which helps agents understand user preferences and provide more personalized services. However, technical capability alone is not enough to guarantee ultimate victory. Morgan Stanley points out that promoting agent adoption essentially also involves changes in consumer behavior habits, so product promotion, user education, and the ability to clearly demonstrate practical value are equally important. Muse has been launched for less than 10 days, but initial user performance has already released some positive signals. **E-commerce Landscape May Be Reshuffled: Amazon.com, Inc., eBay, and Shopify Are Relatively Well Positioned** As agents gradually intervene in shopping decisions and transactions, e-commerce platforms may become one of the most directly affected areas. Morgan Stanley believes that among the e-commerce companies it covers, Amazon.com, Inc. and eBay are relatively well positioned in the agent era, while companies such as Peloton (PTON.US), Chewy (CHWY.US), and FIGS (FIGS.US) face relatively higher potential risks. Shopify plays a different role. Morgan Stanley believes that Shopify is not just a direct-to-consumer shopping platform; its more important value lies in providing infrastructure for a large number of dispersed merchants. Therefore, when merchants need to adjust product information, transaction processes, and technical architecture to adapt to an agent-led shopping environment, Shopify may become an important "infrastructure provider" helping these merchants complete their transformation. In other words, as consumers increasingly let AI "find products, compare prices, and place orders" for them, the focus of competition in the e-commerce industry may gradually expand from "who has the best website and app" to "who is easiest for AI agents to discover, understand, and complete transactions with." **Travel Platforms Are Not the Most Vulnerable: Unique Real-Time Inventory Constitutes an Important Barrier** Recently, market concerns about agents disrupting online travel agencies (OTAs) have clearly intensified. The report notes that as related concerns resurfaced, the stock prices of some online travel companies on the "agent competition battlefield" fell by about 6%-12% cumulatively over roughly the prior two weeks. But Morgan Stanley believes the market may be underestimating an important asset held by Booking Holdings (BKNG.US), Airbnb (ABNB.US), and Expedia (EXPE.US): unique, dispersed, large-scale, and constantly changing real-time inventory. Travel products such as hotel rooms, homestays, and flights are not as easily fully replaced by AI as standardized goods. Travel consumption typically requires extensive image browsing, destination exploration, itinerary planning, and product discovery, so consumers are not necessarily willing to completely bypass OTA platforms. Morgan Stanley believes that "exploration, planning, and product discovery" itself may constitute a competitive barrier in the travel industry. However, OTAs still need to watch two key indicators: first, whether direct traffic is being diverted by AI agents; second, whether these platforms can launch their own agent products to keep users and transactions within their own platforms. These two points may affect their long-term valuation levels. **Ride-Hailing Faces More Direct Impact: AI May Push Competition Further Toward "Price Comparison"** Compared with travel and food delivery, Morgan Stanley believes the ride-hailing industry faces higher agent disruption risk. The reason is that ride-hailing services are more standardized. For a consumer who simply wants to go from the office to the airport, the choice between Uber Technologies, Inc. (UBER.US) and Lyft (LYFT.US) may mainly depend on price and wait time. An agent can fully query multiple platforms simultaneously and then automatically choose the lowest-priced or fastest service. By contrast, ordering food delivery and booking travel products usually involve more image browsing, exploration, and personal preferences, so the value of consumers actively participating in decisions is higher. Morgan Stanley therefore believes that the more standardized, the smaller the ticket size, and the more price-dependent the transaction, the more susceptible it is to horizontal AI agents. However, this does not mean Uber Technologies, Inc., Lyft, or DoorDash (DASH.US) will lose value. Morgan Stanley points out that large-scale drivers, delivery personnel, and local delivery networks themselves are physical infrastructure that agents cannot easily replicate on their own. Even if consumers in the future hail rides or order food through Meta or other AI agents, horizontal agents may still need to connect to Uber Technologies, Inc., Lyft, and DoorDash's networks to actually complete the service. Therefore, the real question may shift from "whether consumers will continue to open the Uber Technologies, Inc. app" to "how much economic value Uber Technologies, Inc. can retain from an order once AI becomes the entry point." **Alphabet Inc. Class C Faces a Deeper Test: Agent Commission Models May Lower Traditional Search Monetization Rates** Morgan Stanley believes that one potential victim especially worth watching in the agent era is Alphabet Inc. Class C's existing business model. As large platforms such as Meta continue to launch agent products, if these products successfully attract large numbers of users and take commercially valuable queries and transactions away from Alphabet Inc. Class C's search engine, Alphabet Inc. Class C may have to proactively adjust its currently highly mature search advertising monetization system. The biggest difference lies in the "take rate." Morgan Stanley estimates that the current "effective take rate" corresponding to each transaction in traditional search advertisingthat is, advertising spend as a percentage of gross merchandise value (GMV)is about 5 to 10 times that of some agent transaction commission models earlier this year. This means that if consumers increasingly complete purchases directly through AI agents rather than clicking traditional search ads to enter merchant websites, Alphabet Inc. Class C may face the risk of lower monetization per transaction. Therefore, the firm believes Alphabet Inc. Class C must continue to accelerate the rollout of next-generation agent products, using its massive search user base and data advantages to defend its position upstream in the agent transaction funnel. But Alphabet Inc. Class C is not without buffer space. Currently, more than 80% of retailers' online traffic is still free or direct traffic. As AI agents become more widespread, if some of the originally free direct traffic converts into agent traffic that requires commission payments, even if the take rate per transaction declines, the additional paid transaction volume may partially offset the impact. Morgan Stanley estimates that if about 5 percentage points of free/direct traffic shifts to paid channels, it can offset a decline of about 14 percentage points in the effective take rate; but if the agent's ultimate take rate is only 5%, Alphabet Inc. Class C would need about 38 percentage points of traffic to shift from free channels to paid channels to fully break even. This also means that potential conflicts of interest in the agent era will become increasingly obvious: large platforms such as Alphabet Inc. Class C want to convert more originally free direct traffic into chargeable agent transactions, while retailers need to protect higher-margin direct traffic.