OTWopentechwire
Tech Intelligence, Openly Wired
Policy

Microsoft's Nadella Sees Agent-Driven Commerce Upending Web Monetisation

The chief executive told a New York audience that platforms refusing to accommodate autonomous software may cede ground to rivals willing to accept a transactional revenue model.

MH
Marcus Halloran
Developer Tools Reporter · Singapore
Nov 20, 2026
7 min read
Microsoft's Nadella Sees Agent-Driven Commerce Upending Web Monetisation
Credit: Spencer Platt / Getty Images

A Shift in Traffic Economics

Satya Nadella, chief executive of Microsoft, told an audience in New York on Wednesday that autonomous AI agents will force a fundamental recalibration of how internet platforms generate revenue. Speaking at The New York Times DealBook Summit, Nadella drew a parallel to the upheaval that accompanied the shift from directory portals to algorithmic search two decades ago, when Yahoo ceded dominance to Google and the locus of value capture moved with it.

The executive's central thesis is that platforms built to monetise human attention through display advertising will find that model under strain as software agents, rather than people, conduct an increasing share of browsing and purchasing. In that environment, he suggested, revenue will migrate toward cost-per-click arrangements or transaction fees, rewarding platforms that facilitate agent activity rather than those that attempt to preserve older structures.

"The current aggregators who have a business model which was for the human web will get, in some sense, reshuffled," Nadella said, according to remarks from the event. "And if you were monetizing ads because of the eyeballs, you now have to probably get down to essentially maybe CPC wins; maybe it's more transactional."

Amazon's Block and the Shopify Alternative

The discussion turned to Amazon's decision to prevent AI agents from accessing its retail site, a move that stops autonomous software from browsing product catalogues or completing purchases on behalf of users. Nadella framed that choice as emblematic of a broader tension: incumbents defending existing revenue streams against a technology that threatens to reroute value.

He pointed to Shopify as a potential beneficiary. If Amazon maintains its block while a competitor opens its platform to agents, the latter could capture a new category of demand. The implication is that merchants and consumers alike may gravitate toward whichever infrastructure proves most compatible with agent-mediated commerce, regardless of historical brand strength or network effects.

"If you don't accept the agent, somebody else will," Nadella said. The remark underscores a view that access, rather than scale alone, may become the decisive competitive variable in a market where software intermediaries handle discovery and fulfilment.

The Precedent of Search

Nadella's analogy to the portal-to-search transition is instructive. In the late 1990s, Yahoo and similar directories organised the web through human curation, capturing advertising revenue by serving as the default starting point for navigation. Google's algorithmic approach bypassed that model, delivering results directly and monetising through text ads tied to intent. The shift redistributed billions in market capitalisation and redefined the economics of attention.

The agent-driven web, in Nadella's telling, represents a comparable inflection. Where search engines respond to queries typed by users, agents operate with greater autonomy, executing multi-step tasks such as price comparison, inventory checks, and purchase completion without requiring continuous human input. That changes the unit of monetisation from pageviews or impressions to completed actions, and it changes the entity being served from a person to a piece of software.

For platforms whose revenue depends on users lingering to view ads, the shift poses an obvious challenge. An agent optimising for speed and cost will not pause to consider banner placements or sponsored listings in the same way a human might. It will prioritise platforms that expose clean APIs, transparent pricing, and low-friction checkout, attributes that do not always align with advertising-led business models.

Regional Implications for E-Commerce and Advertising

The dynamics Nadella described have particular resonance in markets where e-commerce and digital advertising are still consolidating. In Southeast Asia, platforms such as Shopee, Lazada, and Tokopedia have built substantial businesses around app-based browsing and in-feed advertising. If agent-mediated shopping gains traction, those platforms will face a choice similar to Amazon's: open their infrastructure to autonomous software and accept a shift toward transactional revenue, or maintain control over the user experience and risk losing volume to more accessible competitors.

In India, where Flipkart and Amazon India compete alongside a fragmented landscape of vertical marketplaces, the arrival of capable agents could accelerate consolidation or, conversely, empower smaller platforms that adopt agent-friendly architectures early. The same applies to payments and logistics providers, which may find themselves competing to become the default rails for agent-initiated transactions.

China's e-commerce giants, Alibaba and JD.com, have already invested heavily in automation and algorithmic recommendation. Their challenge will be adapting those systems to serve external agents rather than only their own recommendation engines. The question is whether they will treat third-party agents as partners or as threats to the closed-loop ecosystems they have spent years constructing.

Business Model Scramble

Nadella's phrase "if you're in love with a particular business model, you have to kind of scramble" captures the strategic bind facing many large platforms. Advertising-supported services have enjoyed decades of compounding returns to scale: more users attract more advertisers, which funds better content or lower prices, which attracts more users. Agents disrupt that loop by decoupling consumption from attention.

A transactional model, by contrast, aligns revenue with completed actions. Platforms earn a percentage of each sale, subscription, or booking, regardless of how much time a user or agent spends on the site. That structure is common in travel (Booking.com, Expedia), gig marketplaces (Uber, Grab), and some categories of software distribution (app stores), but it is less prevalent in content, social media, and general retail, where advertising has historically been more lucrative.

The transition is unlikely to be binary. Hybrid models will emerge, with platforms offering agent-accessible APIs for transactional revenue while maintaining ad-supported experiences for human users. The risk for incumbents is that agents capture the highest-intent, highest-value interactions, leaving the advertising side with lower-quality inventory and diminished pricing power.

Technical and Regulatory Frictions

Opening a platform to agents is not merely a matter of policy; it requires infrastructure. APIs must be stable, well-documented, and capable of handling the request volumes that hundreds or thousands of agents might generate. Authentication, rate limiting, fraud detection, and dispute resolution all become more complex when the counterparty is software rather than a person.

Regulatory questions also loom. If an agent makes a purchase that a user later disputes, who bears liability? If an agent scrapes pricing data at scale, does that constitute fair use or a violation of terms of service? Jurisdictions differ in how they treat automated access, and platforms operating across borders will need to navigate a patchwork of rules.

In the European Union, the Digital Markets Act already imposes interoperability obligations on designated gatekeepers. If agents become a significant mode of commerce, regulators may extend those obligations to require that large platforms provide agent access on reasonable terms. That would accelerate the shift Nadella anticipates, but it would also introduce compliance costs and potential friction with competition authorities in other regions.

The Shopify Advantage

Nadella's reference to Shopify is revealing. Unlike Amazon, which operates a retail marketplace with its own inventory and advertising business, Shopify provides infrastructure for independent merchants. Its revenue comes primarily from subscription fees and payment processing, a model already aligned with transactional economics.

For Shopify, enabling agents is a natural extension of its existing API-first architecture. Merchants on the platform could benefit from agent-driven discovery and purchasing without Shopify needing to renegotiate its core business model. If agents become a meaningful source of demand, Shopify's decentralised structure may prove more adaptable than the vertically integrated marketplaces that dominate retail today.

The same logic applies to other infrastructure providers. Payment processors, logistics networks, and identity verification services that expose clean APIs and charge per transaction are well-positioned to serve agent-mediated commerce. Platforms that bundle those services with advertising or subscription models may find themselves at a disadvantage.

What Comes Next

Nadella's remarks do not constitute a prediction of immediate disruption. Agents capable of autonomously navigating complex purchasing decisions remain experimental, and consumer trust in delegating financial transactions to software is still nascent. But the direction of investment and product development is clear. Microsoft itself has embedded agent functionality into its Copilot suite, and competitors from Google to Anthropic are building similar capabilities.

The question for platforms is not whether agents will arrive, but whether they will be welcomed or resisted. Those that choose resistance may preserve short-term revenue but risk long-term displacement. Those that adapt early will need to rebuild monetisation models, retrain sales teams, and renegotiate relationships with advertisers and merchants. Neither path is without cost.

What is certain is that the economics of aggregation, which have defined the internet for two decades, are entering a period of contestation. The platforms that emerge strongest will be those that treat agents not as a threat to be blocked, but as a new category of customer to be served.

Read next
Policy

Nvidia Exits China Market as Acting CEO Reverses Huang-Era Chip Deal

Wei Zhang · 6 min
Policy

Nvidia Installs Board Committee for Export Controls After Huang Departure

Kenji Watanabe · 6 min
Policy

Nvidia Creates Board Committee to Police Chip Exports After CEO Exit

Hana Park · 6 min
Spot something wrong? Email corrections@opentechwire.com. We log every correction publicly.