Microsoft CEO Satya Nadella has predicted that the rapid adoption of artificial intelligence agents could fundamentally change how businesses make money on the internet. Speaking at The New York Times DealBook Summit on November 18, 2026, Nadella argued that the shift from human browsing to AI-assisted interactions could disrupt advertising-based business models, change how websites attract customers and create new ways for digital platforms to earn revenue.

Nadella’s argument centres on the emergence of an “agentic web”, where AI agents increasingly search for products, compare prices, research information and complete transactions on behalf of users. If people rely less on visiting websites directly, businesses that currently earn money from page views and advertising impressions may need to rethink their strategies. Nadella suggested that companies could increasingly move towards cost-per-click arrangements and transaction-based revenue models, warning that platforms that refuse to accommodate AI agents could lose opportunities to competitors.

Key Takeaways

  • Microsoft CEO Satya Nadella says AI agents could reshape the economics of the internet.
  • The shift could challenge websites that depend on human traffic and advertising impressions.
  • Advertising businesses may increasingly move towards cost-per-click and transaction-based models.
  • Nadella compared the transition to the historical shift from web portals such as Yahoo to search engines such as Google.
  • He discussed Amazon’s restrictions on AI agents accessing its retail website.
  • Platforms that allow AI agents to browse and complete transactions could attract business from competitors that block them.
  • The transition could affect publishers, online retailers, search engines, advertising platforms and digital marketplaces.
  • Nadella did not provide a timeline or financial estimate for the expected transformation.

What Did Satya Nadella Say About AI Agents and the Internet?

Speaking at the 2026 New York Times DealBook Summit, Nadella discussed how autonomous AI agents could change the relationship between users, websites and digital businesses.

His central argument was that the internet’s current economic structure was largely designed around human users.

People visit websites, read articles, view advertisements, compare products and click links. These activities generate traffic and advertising revenue for publishers, search engines, retailers and other online businesses.

However, AI agents could increasingly perform these activities on behalf of users.

Instead of opening several shopping websites, a person might ask an AI agent to find a particular product, compare prices, check delivery conditions and recommend the best available option.

The user may never visit most of the websites involved in the process.

Nadella argued that this change could force companies to reconsider how they attract customers and generate revenue.

He said that existing aggregators built around the economics of the human web would be reshuffled as AI agents become more important.

He also suggested that businesses dependent on advertising impressions may need to move towards cost-per-click or more transactional models.

What Is the Agentic Web?

The agentic web refers to an emerging internet environment in which AI agents perform tasks on behalf of users rather than simply helping them find information.

Traditional internet usage generally follows a process:

  1. A person opens a search engine.
  2. They enter a query.
  3. They visit different websites.
  4. They compare information or products.
  5. They make a decision.
  6. They complete a transaction.

With AI agents, much of this process could happen automatically.

For example, someone planning to purchase a laptop might ask an AI assistant to find suitable models under a specific budget.

The agent could potentially:

  • Search multiple retailers.
  • Compare technical specifications.
  • Check available discounts.
  • Examine customer reviews.
  • Compare delivery options.
  • Identify the most suitable product.
  • Help complete the purchase.

The user interacts primarily with the agent rather than visiting every website.

This could change how online businesses compete for customers.

Why AI Agents Could Disrupt Advertising-Based Business Models

Advertising is one of the foundations of the internet’s current economic structure.

Many websites generate revenue by attracting visitors and displaying advertisements. Search engines sell advertising placements, publishers monetise page views and social platforms earn money from user attention.

AI agents could weaken some of these models by reducing the number of direct website visits.

1. Fewer Human Visits to Websites

If an AI agent retrieves information and summarises it directly for a user, that person may not need to visit the original website.

For example, someone asking for a summary of five news articles could receive the information in an AI interface without opening all five publishers’ websites.

This could reduce referral traffic and the number of advertising impressions generated by those visits.

However, the effect will vary by content category. Some users will still visit original websites to read full articles, watch videos or verify important information.

2. Advertising Impressions Could Lose Some Value

Traditional online advertising frequently relies on the number of people viewing an advertisement.

When AI agents perform more browsing, the number of human eyeballs reaching individual pages could decline.

That creates a challenge for advertising businesses that depend heavily on impressions.

Nadella suggested that companies may need to focus more on cost-per-click and transactional outcomes rather than simply monetising page views.

This could change how advertisers evaluate the value of digital platforms.

3. Transaction-Based Revenue Could Become More Important

In an agent-driven internet, businesses may increasingly compete to become the platform through which AI agents complete transactions.

For example, an AI agent might compare products across several retailers and complete the purchase through one platform.

The retailer that successfully supports the agent’s workflow could receive the transaction, even if the customer never browses its website manually.

This could create new opportunities for platforms that make their product catalogues, pricing information and purchasing systems accessible to AI agents.

Nadella Compares the Shift to Yahoo and Google’s Rise

Nadella compared the potential transformation to an earlier change in the internet industry.

During the early development of the web, portals such as Yahoo helped users discover websites and navigate online information.

Search engines subsequently changed how people found content.

Google became a major gateway to the internet by helping users search for information across millions of websites.

That transition changed which companies controlled online discovery and how businesses attracted traffic.

Nadella believes the rise of AI agents could create another similar transition.

Earlier internet transitionEmerging AI transition
Users navigated websites through portalsUsers delegate discovery to AI agents
Search engines became major traffic gatewaysAI agents could become decision-making gateways
Businesses competed for search visibilityBusinesses may compete for agent visibility and compatibility
Advertising monetised search and website trafficTransaction-based and agent-oriented monetisation could grow

The comparison highlights a potential change in the internet’s distribution system.

However, the final structure of an agent-driven internet is not yet established. Search engines, websites, social networks and AI platforms may continue to coexist.

Amazon’s AI Agent Restrictions Become Part of the Debate

Nadella’s comments also addressed the growing tension between online retailers and AI agents.

Amazon has taken steps to restrict certain third-party AI agents from accessing its retail website.

These restrictions matter because AI shopping agents need access to product information, pricing and purchasing systems to complete transactions.

If a retailer prevents agents from interacting with its website, users may face difficulties using those agents to compare products or complete purchases through that retailer.

Nadella suggested that businesses that resist the transition could create opportunities for competitors.

He used the example of a shopping platform such as Shopify potentially benefiting if it makes its services more accessible to AI agents.

His broader argument was that companies may need to reconsider existing business models as customer behaviour changes.

The discussion does not establish that Shopify has replaced Amazon or that retailers will abandon their existing platforms. It highlights the potential competitive consequences of different approaches to AI-agent access.

How AI Agents Could Change the Economics of Different Internet Businesses

The potential effects extend beyond online shopping.

Different parts of the internet economy rely on human attention, website traffic, referrals and transactions.

Business categoryExisting revenue modelPotential change from AI agents
News publishersAdvertising, subscriptions and page viewsGreater pressure on referral traffic and direct visits
Search enginesSponsored listings and advertising clicksMore agent-led discovery and possible transaction-based models
E-commerce platformsProduct sales and advertisingCompetition for AI-agent recommendations and purchases
Travel websitesBookings, commissions and advertisementsAI agents comparing and booking services
Review websitesAdvertising and affiliate commissionsAgents extracting information without direct visits
Digital marketplacesTransaction fees and seller servicesNew agent-oriented purchasing and referral systems
Online advertising networksImpressions and clicksGreater emphasis on measurable actions and conversions

These are potential developments rather than confirmed outcomes.

The impact will depend on how AI agents access websites, how platforms negotiate commercial relationships and whether users trust agents to complete transactions independently.

What This Could Mean for Digital Publishers

News publishers and content websites could be among the businesses affected by a shift towards agent-led browsing.

Many publishers depend on visitors arriving through search engines, social media and direct website visits.

If AI agents increasingly summarise information without sending users to the original articles, publishers could experience pressure on advertising impressions and referral traffic.

This creates a difficult question: how should content creators earn revenue when AI systems use their information but users do not necessarily visit their websites?

Possible business responses include:

  • Direct subscriptions.
  • Licensing content to AI companies.
  • Paid access to specialised information.
  • Commercial partnerships with AI platforms.
  • Membership-based business models.
  • Greater emphasis on original reporting and proprietary data.

Publishers may also seek new commercial arrangements that compensate them when AI agents retrieve or use their content.

However, the economics of such agreements remain unsettled.

The long-term relationship between AI companies and content publishers will depend on licensing negotiations, copyright disputes, regulation and changes in consumer behaviour.

Why Businesses May Need to Become AI-Agent Friendly

Nadella’s comments suggest that businesses could increasingly need to optimise their digital infrastructure for both humans and AI agents.

Traditional website design focuses on user experience, search visibility, advertisements and conversion rates.

An agent-driven environment could introduce additional requirements.

Companies may need to provide:

  • Structured product information.
  • Reliable and frequently updated pricing.
  • Machine-readable catalogues.
  • Clear availability and inventory information.
  • Secure purchasing interfaces.
  • APIs for approved agent access.
  • Transparent transaction policies.

For example, a hotel website that provides accurate room availability and booking information through accessible systems may be easier for AI agents to use than one that depends entirely on manual browsing.

The same principle could apply to retailers, travel companies, financial services and digital marketplaces.

Challenges Facing the Agentic Web

Although AI agents could create new commercial opportunities, several challenges need to be addressed before this model becomes widespread.

Trust and Accuracy

AI agents can misunderstand instructions, provide incorrect recommendations or select unsuitable products.

Users will need confidence that agents can make reliable decisions, especially when handling financial transactions.

Security and Fraud

Allowing automated systems to interact with websites and make purchases creates new security risks.

Businesses will need to distinguish legitimate agents from malicious bots and establish secure authentication mechanisms.

Website Access and Control

Website operators may not want unrestricted automated access to their services.

They may introduce rules around scraping, automated browsing, purchasing and data access.

Attribution and Compensation

If an AI agent uses information from a publisher or retailer, the original source may want attribution or payment.

The industry will need new commercial arrangements to address how value is shared.

Competition and Market Concentration

If a small number of AI platforms become the primary gateways through which users discover products and services, they could gain significant influence over online commerce.

This could create new competition and regulatory questions.

The Bigger Picture

Nadella’s prediction points towards a potential restructuring of the internet’s economic model. For much of its history, the web has been built around human attention: users visit websites, view advertisements, click links and make purchases. AI agents could increasingly perform these activities on their behalf, reducing the importance of direct browsing in some categories.

This does not necessarily mean advertising will disappear. Instead, businesses may experiment with different ways of monetising digital interactions, including referrals, completed transactions, subscriptions and commercial access to information. The companies that make their services useful and accessible to AI agents could gain opportunities, while those dependent on older traffic models may face pressure to adapt.

Looking Ahead

The next stage of the agentic web will depend on whether consumers become comfortable delegating shopping, research and other online activities to AI systems. It will also depend on how retailers, publishers, search engines and technology companies establish rules for agent access, attribution, security and payments. Nadella’s comments describe a possible direction for the internet, but the speed and scale of the transition remain uncertain.

For businesses, the important question is no longer only how to attract human visitors. They may increasingly need to consider how AI agents discover their services, interpret their information and complete transactions. If that shift accelerates, the companies controlling the next generation of digital discovery could influence how value is distributed across the web.

Get the day’s top stories in your inbox

One concise email. No spam, unsubscribe anytime.