a16z warns "AI agents don't watch ads" will place orders directly: $291 billion online advertising market will be destroyed

a16z Crypto’s co-founder of Merit Systems, Sam Ragsdale, pointed out that AI agents won’t be distracted by ads, meaning the “distraction economy” model that sustains the entire free internet is breaking down; ChatGPT and Gemini have introduced direct checkout features within conversations, but they remain walled gardens. The true breakthrough lies in open protocols like x402 and MPP.

(Background: a16z’s latest insight: Is traditional e-commerce dead? AI-native platforms are redefining “shopping.”)

(Additional context: OpenAI’s first AI agent, “Operator,” is here! Helping you shop, book tickets, order food… solving tedious online tasks.)

The global online advertising market is expected to reach about $291 billion in 2025, with Google holding a dominant position. However, a fundamental threat is emerging: AI agents don’t look at ads.

In a recent insight from a16z Crypto, Ragsdale warned that as AI agents take over shopping processes, the “attention interception” mechanism that ad-based revenue relies on will become completely ineffective. Ads make money because humans browsing the web are distracted by banners and pop-ups; but large language models (LLMs) and AI agents “won’t be distracted,” and this is shaking the very foundation of the advertising industry.

From 1997 to 2024: A Business Model Powered by Distraction

Ragsdale characterizes nearly 30 years of internet history as the “distraction business model” era. Starting in 1997, advertisers paid not for product exposure but for a user’s fleeting attention—and that attention was precisely the “interruption” created by ads.

This model gave rise to free search engines, social platforms, news sites, and built today’s open internet. However, Ragsdale points out a paradoxical closed loop: it was ads that created the free, open web, which in turn generated the 10 trillion characters of data needed to train LLMs. The rise of LLMs ultimately signals the end of the ad era. He describes ads as “a clever shortcut,” but by 2026, “this shortcut is heading toward death.”

ChatGPT and Gemini Have Already Taken the Lead, But Still Walled Gardens

AI platforms are aware of this trend. ChatGPT and Gemini have launched “instant checkout” features for US users, allowing direct shopping within the chat window without jumping to external e-commerce sites.

This seems like the first step toward AI-driven commerce, but Ragsdale notes it remains a walled garden—merchants must go through strict vetting to be included, and consumer options are subject to platform control. For merchants, it’s just replacing Google ad bidding with another access mechanism; for consumers, the “freedom” of AI agents is still limited by platform boundaries.

Open Protocols Are the Real Solution: x402 and MPP

Ragsdale uses a precise analogy: “AI agents that can only shop from pre-approved merchants are like employees with company credit cards who can only spend at three specific vendors; but AI agents with open protocols are like entrepreneurs holding bank accounts.”

He advocates that breaking down walled gardens depends on enabling AI agents to explore products via open protocols. Currently, two main directions are emerging: one is Coinbase’s x402 protocol, based on HTTP 402 status code, allowing AI agents to complete payments without human intervention; the other is Tempo’s collaboration with Stripe on the Machine Payments Protocol (MPP), designed for machine-to-machine automated transactions.

Implications for the Crypto Ecosystem

This structural shift in the advertising industry opens a critical door for crypto and Web3. AI agents require permissionless, programmable payment infrastructure—and that’s exactly where blockchain protocols excel. The rise of protocols like x402 means that future transactions by AI agents could be settled on decentralized payment rails, rather than within proprietary, closed payment systems of large platforms.

Ads funded the internet; the internet trained AI; AI is now killing ads—and once this cycle completes, the underlying architecture of e-commerce and payments will be reshaped.