How AI Is Threatening Platforms’ Revenue Streams

Over time, the sheer volume of online activity and the convenience of centralized digital services have turned us into a platform-dependent society, shaping markets and labor while seizing ever-greater shares of our time and daily activity. Consumers overwhelmingly rely on the likes of Google for search, WhatsApp for communication, YouTube for entertainment, Instagram for social connection, and Amazon for commerce. Platforms have turned these interactions into a massive profit machine. User attention, data, and online engagement have become the currency of the digital age. Amazon’s advertising businesses alone reached $56 billion in 2024, growing 18% year-on-year, making it Amazon’s fastest growing and most profitable segment. The World Advertising Research Center projects the number could exceed $79 billion by 2026. But here’s the catch: almost every dollar of that revenue depends on human eyeballs seeing sponsored products and advertisements and people making decisions based on what they find on the platforms they frequent. These are no longer safe assumptions. That’s because people are increasingly relying on personal AI agents to make choices for them rather than just present decision-support data (as platforms and search engines do). AI agents are rational, not emotional. They don’t see ads. They don’t impulse buy. They don’t get locked into ecosystems. As machines start making purchase decisions for humans, they threaten all the revenue streams on which platforms rely: advertising clicks, transaction fees, subscriptions, and service charges for third party providers. This article examines the nature of the AI threat and argues that not only will AI agents compete away the revenue streams of the giant digital platforms, but they will also render irrelevant the data on which the platforms built their competitive advantage. The platforms’ ability to predict user preference from user behavior on their websites will be superseded by AI’s ability to discern user intent by accessing the full range of the user’s interactions. Let’s begin by looking at how AI will affect platform revenue streams. The Erosion of Revenue Historically, platforms made money through advertising, transaction fees, subscriptions, and ecosystem services such as cloud computing, logistics, and payments. As users delegate search, choice, and execution to AI agents, the interaction surfaces on which these revenues depend erode, shifting value capture from platforms back to users. Advertising revenue The single most important revenue source for platforms remains advertising. In 2024, advertising represented approximately 75% of Google’s revenue and 97% of Meta’s revenue; as noted earlier, it was also Amazon’s most profitable segment. As we witness a structural shift away from individual platform interfaces (websites, apps, search engines) tasked with finding the best option for a consumer, platforms lose the ability to monetize attention by selling targeted ad space based on user data. Instead, users will turn to AI agents that understand them more holistically, consume information directly, evaluate options autonomously, and either make agentic recommendations or autonomously act on behalf of users. AI agents not only remove the cognitive burden of users navigating endless choices, but also eliminate the incentives of ad-driven platforms. Platforms lose the ability to monetize attention because AI agents no longer “see” or click on ads. The result is what we call “zero-click commerce”: transactions that proceed from intent to fulfillment without any interface interaction where advertising could intervene. This fundamentally breaks the two-sided market equilibrium: platforms can no longer subsidize users with free services when the advertiser side cannot reach those users. Transaction fees Platforms take a cut of each transaction they mediate (e.g., Amazon Marketplace fees, Uber/Lyft ride commissions, Airbnb service fees). To maximize their ROI, platforms have long created internal referral loops (e.g., Google steering users from Search to Maps to YouTube, Amazon redirecting shoppers into Prime and Whole Foods) and moats to trap attention, data, and transactions within their own ecosystems. These make it difficult, time-consuming, or costly for users to compare alternatives or switch providers. Agentic AI breaks this logic completely. Agents search across platforms, unbundle offerings, compare prices instantly, and route buyers to whichever provider meets the user’s needs. The economic consequence is clear: a race to the bottom in fees. When their moats are rendered worthless, the classic platform dynamic of “winner-take-all” reverses into “everyone-loses-together” because AI agents operate across all networks simultaneously. Traditional marketplaces, such as Amazon, Uber, Airbnb, and Booking.com, lose control over discovery, pricing, transaction flow and most importantly, collecting transaction fees. Subscriptions and membership fees Depending on the provider, subscriptions can be a major source of platform revenue. Amazon alone has approximately 250 million paying Prime members, bringing in $44.37 billion in subscription fees in 2024. For platform providers, these subscription models depend on users spending meaningful time inside the ecosystem. While some only offer one service, others, like Costco Online or Amazon offer integrated subscription models designed to increase lock-in, loyalty, and cross-service engagement (e.g., shipping, video, music, discounts, etc.). Users, once subscribed, tend to remain loyal to these providers. The vulnerability here is psychological, not technical. Subscriptions rely on mental biases; once we pay for Prime, we feel compelled to “get our money’s worth” by shopping more on Amazon, despite it not always being the most economical choice. Unlike humans, agents evaluate whether Prime’s benefits justify its cost for each transaction without the psychological pull of sunk costs. When AI agents interact instead of users, many of the perceived benefits of subscriptions (fast shipping, ad-free viewing, storage, premium recommendations) lose relevance. An agent can instantly compare the total cost (including shipping) across all providers, making “free Prime shipping” irrelevant if another provider offers lower total costs. The result is more transparent choice, lower switching costs, better value, and far less susceptibility to marketing influence. Ecosystem services Platforms often offer ancillary services as a way of growing and monetizing user dependence on their ecosystems. Major profit centers include cloud services (AWS, Google Cloud), logistics and fulfillment services (Fulfillment by Amazon), payment and fintech services (Apple Pay, Google Pay), and data services (APIs, analytics tools). AI agents disaggregate these services and optimize across providers, not within ecosystems. They choose the cheapest cloud service, the best logistics provider, and the most cost-efficient payment option, even if it means moving away from dominant players or platform bundles. As a result, the financial bundling advantage of platform ecosystems breaks down as AI agents deconstruct walled gardens into interchangeable components. The Erosion of Competitive Advantage Platform personalization has long been celebrated as a competitive advantage. Amazon’s recommendation engine, Netflix’s content curation, Spotify’s Discover Weekly, each uses the behavioral signals they capture to predict what users want. But this traditional personalization has inherent limitations. It relies on observed behavior within a single platform: what you clicked, watched, purchased, or lingered on. It does not capture your behavior outside the platform, which means that the picture platforms build is fragmented and inferred. AI agents operate on an entirely different level. When users invite an AI agent into their digital lives, granting access to their inbox, calendar, cloud drives, and private conversations, they share information they would never voluntarily disclose to a brand. Developments such as OpenAI’s integration with Gmail, Google Calendar, and Contacts, or Microsoft Copilot’s ability to analyze email threads and meeting schedules, mark a fundamental shift in the data asymmetry between consumers and commerce. The agent doesn’t infer preferences from clicks; it knows them from context. It understands that your budget is tight this month from your bank notifications, that you are initiating a romantic relationship from your calendar entries and that you’ve been stressed about a work deadline from your email tone. The intimacy is asymmetric in another sense too. Users confide in AI agents about their anxieties, insecurities, aspirations, and constraints: the kind of “vulnerable intimacy” they would never share with a retailer’s chatbot. An AI agent knows when someone is stretched thin financially, feeling self-conscious about a purchase, or shopping under time pressure. This creates a level of personalization that platforms, for all their data, cannot match. AI users don’t just feel understood, they feel “seen”. For platform executives, this represents a profound challenge. The data that once conferred competitive advantage – the ability to recommend, target, and predict – is now dwarfed by the holistic understanding AI agents accumulate through trusted, ongoing relationships with users. Platforms see behavior; agents discern intent. Platforms personalize within their walls; agents hyper-personalize across the user’s entire life. Inevitably loyalty will flow to the AI agent, not the platform. What Can Platforms Do? Platform operators now face a set of unattractive options. They can: Resist The most immediate weapons at hand for platforms are legal actions and technical barriers. These can buy the platforms some time. A good example is Amazon’s lawsuit last year against Perplexity, owner of the AI shopping assistant Comet AI. Amazon claimed that Perplexity concealed its agents so that they could access data on Amazon’s website without the online retail giant’s approval. This March, a U.S. District found in Amazon’s favor and ordered a preliminary injunction forbidding Perplexity agents access to Amazon’s website. Amazon had argued that Perplexity’s actions of its agents compromised Amazon’s efforts to guarantee a “safe shopping experience” for the site’s users. While legal actions of this sort, which enable platforms to introduce technical barriers to access by AI agents from outside the firm may afford a breathing space they are unlikely to halt the shift to AI agent-enabled shopping in the longer term. Adapt In tandem with their efforts to buy time, platforms are building their own AI agents to protect their customer relationships. Amazon’s “Buy for Me” and represents an attempt to control the agentic layer rather than cede it to third parties. Google has also introduced AI agents that call stores on your behalf, while Visa and Mastercard are building authentication protocols for autonomous AI purchases. This strategy carries risks: platforms may cannibalize their own advertising revenue by accelerating the shift away from human browsing. And they will still not be immune to the disruption by competing AI agents. The space is rapidly filling out, and it is not clear how much variation there will be across agents in terms of quality delivered to the user. If there is, it may well emerge across customer need, with people gravitating towards different agents for healthcare, finance, or shopping based on where they developed trust. Reinvent Ultimately, any long-term strategy for platforms must be predicated on a realization that their era of dominance is over, and that, in the future, they will compete to be selected by AI agents, not by human users. Such platforms reinvent their entire strategy to become agent-ready and invest in API-first architectures, machine-readable product data, real-time pricing feeds, and verification services that AI agents can consume programmatically. Some are already doing so. In January 2026, Google and Shopify co-developed the Universal Commerce Protocol (UCP), an open standard—endorsed by more than 20 partners including Target, Walmart, Visa, and Mastercard—that gives AI agents a common language to discover products, initiate transactions, and manage orders from any merchant. The question isn’t whether this shift will happen, but whether platforms will be prepared for it. It is not only the platform operators who must adjust. All organizations whose strategies are predicted on the platform economy will have to change. CEOs and strategy leaders need to rethink their platform-based business models and digital transformation roadmaps. They are likely not future proof. CMOs need to consider how today’s customer journeys will change when AI agents, not humans make the choices. User interface specialists need to think about pivoting away from user-facing screens to API-first, agent-first conversations. Technical leaders need to prepare tech stacks for a world of AI infrastructure owned by others rather than their own proprietary platforms. The list goes on. Every leader’s job will be changing, and consumers, boards and investors will reward those who move quickly to AI agent-first strategies. Waiting is No Longer a Good Strategy Christmas 2025 may be remembered as the tipping point for agentic commerce. What was experimental throughout the year became common practice during retail’s most emotionally intense period. Salesforce reported that AI agents influenced $67 billion in global Cyber Week sales—20% of all purchases. Adobe found that AI traffic to retail sites surged 805% year-on-year on Black Friday and 670% on Cyber Monday. Shoppers arriving via AI platforms converted 38% more often than those coming from traditional sources, including social media. Perhaps most telling: Mastercard found that nearly half of Gen Z and Millennials happily handed their holiday shopping off to AI agents. The behavioral shift is real. Anthropic’s Economic Index shows that users delegating complete tasks to AI with minimal oversight jumped from 27% in late 2024 to 39% by August 2025, with automation (49%) now exceeding augmentation (47%) for the first time. Among enterprise API users, 77% of interactions are now fully automated. People aren’t cautiously testing AI agents anymore. They’re trusting them to act. And while not all incursions into the platform space have worked out—in September 2025, OpenAI launched an in-chat AI-enabled checkout in ChatGPT with Shopify only to shut it down six months later—other comparable initiatives are succeeding. Walmart’s AI shopping assistant “Sparky” drives order values 35% higher than unassisted purchases, according to CEO John Furner on the company’s Q4 FY26 earnings call, with half of all app users having tried it. Macy’s “Ask Macy’s,” powered by Google’s Gemini, found that users spent 4.75 times more than non-users during several weeks of testing with half the retailer’s website traffic. . . . Leaders who dismiss the shift from platforms to AI-shopping agents as “hype” or “too early” are trapped in the assumption that the platform era will hold even as AI rewrites the rules platforms depend on. Their organizations will not just lose competitiveness; they’ll lose the very foundation and economic logic upon which their business models rest. Without a sustainable revenue stream, the end of the platform economy is looming. The only question is who will prove resilient enough to adjust.

How AI Is Threatening Platforms’ Revenue Streams

Summary.

Partner Center