For a quarter of a century, Google Search sat at the center of the internet under a simple, mutually beneficial social contract. Users typed in a query, Google indexed hundreds of billions of web pages to show the most relevant links, and users clicked through to the websites that held the answers.

In this ecosystem, handling roughly 8.5 billion searches a day and over 90% of global traffic, everyone got paid. Publishers monetized the visitor traffic, while Google took the largest cut by selling ads both on the search results page and on the publishers’ sites themselves—growing its search advertising business to over $260 billion a year.

Today, that 25-year agreement is broken. With the aggressive rollout of AI across Google Search, roughly 60% of searches now end without a user ever clicking a link. While NPR has called the resulting traffic collapse an “extinction-level event” for online news, the real story isn’t just the destruction of traditional search—it is the closed-loop business model Google is building to replace it.

Engineered to Eliminate the Click

The shift began with a seemingly harmless feature: AI Overviews. Sitting atop the search results for billions of queries and reaching 2.5 billion monthly users, Gemini immediately generates an answer on the page—and neither users nor publishers can opt out of the system.

From there, Google rapidly escalated its on-page retention features:

• AI Mode: A full conversational interface that answers questions, cites sources, and keeps the user inside the chat window.
• On-the-Fly Custom Interfaces: Announced at Google I/O in May 2026, Search can dynamically build custom interfaces pulling in images, tables, and structured data around a query.
• Information Agents: Automated agents that track topics over time and push updates directly to users before they even initiate a search.

Each of these features is specifically engineered to reduce clicking, effectively demoting publishers from a final destination to a raw data source.

The Great Traffic Collapse

When users stop clicking, the economic foundation of the open web fractures. According to data from Chartbeat, which tracks over 2,500 news sites globally, Google search page views fell by 34% between December 2024 and December 2025.

However, that 34% average hides a stark divide. The smaller the publisher, the harder they were hit:

Publisher Tier / Company	Reported Traffic Impact	Additional Context
Large Publishers	-22% search traffic	Digital Content Next members (NYT, Condé Nast, Vox) lost 1% to 25%, with some queries dropping over 75%.
Medium Publishers	-47% search traffic	Squeezed out as larger companies dominate remaining link visibility.
Small Publishers	-60% search traffic	Independent sites like Giant Freaking Robot shut down entirely; HouseFresh suffered a traffic collapse and laid off staff.
DMG Media (Daily Mail)	Up to -89% on specific queries	Documented steep query-level visibility drops.
HubSpot	-70% to -80% estimated drop	Affected despite being one of the most SEO-optimized companies in the world.
People Inc.	Fell from ~60% to mid-20%	Parent of People, Food & Wine, and Investopedia; CEO now calls Google a “competitor, not a partner”.
Chegg	-49% traffic decrease	Filed a lawsuit against Google following the traffic collapse.

The Replacement Model: Licensing Instead of Traffic

Rather than sending readers to websites, Google and its AI peers are transitioning to a wholesale licensing model—paying select publishers for the right to feed their content directly into AI systems. Google has a standing real-time news deal with the Associated Press for Gemini and is running a licensing pilot with roughly 20 national publishers.

Notably, Google did not initiate this shift out of goodwill. In 2024, the company showed almost no interest in licensing negotiations because it could summarize web content for free. Talks only resumed in 2025 and turned serious in mid-2026 when regulators stepped in—specifically the UK’s Competition and Markets Authority forcing opt-out controls in June 2026 and the EU setting a July 2026 deadline under the Digital Markets Act.

By the time Google moved, it was merely copying a wholesale market already established by competitors:

• OpenAI: Signed News Corp for a reported $250 million over five years, alongside deals with The Guardian, Axios, and The Washington Post.
• Microsoft: Built a pay-per-use content marketplace for Copilot, joined by the AP and USA Today.
• Amazon & Meta: Amazon signed The New York Times, Condé Nast, and Hearst, while Meta partnered with News Corp, Le Figaro, and Süddeutsche Zeitung.
• Perplexity: Set up a $42.5 million fund to share ad revenue directly with publishers whenever their content is cited.

In this new economy, content is no longer sold “retail” to human readers via clicks; it is sold “wholesale” to AI models as raw material.

Why the Licensing Economy Fails the Web

While six- or nine-figure licensing checks sound substantial on paper, they fail to replace the revenue lost from direct traffic.

1. Pennies on the Dollar: Even the most lucrative deal on the market—News Corp’s $250 million over five years ($50 million annually)—represents just 0.5% of News Corp’s ~$10 billion annual revenue.
2. Small Publishers Are Locked Out: Google is negotiating private deals with around 20 major national outlets, leaving independent journalists, local news, and niche experts with nothing. Yet those smaller publishers—who already lost 60% of their traffic—still have their work crawled, summarized, and fed into the AI unpaid.
3. Suppliers, Not Partners: As a $2 trillion company, Google already owns a proven revenue-sharing framework: the YouTube Partner Program, which shares ad cuts with creators at scale. It could apply that same flat-cut model to cited websites tomorrow. Instead, by choosing private, per-deal licensing over open revenue sharing, Google avoids making publishers true partners—keeping them on the hook as replaceable suppliers while acting as the sole gatekeeper of the internet’s chain of command.

Three Visions for the Future of the Internet

As this model cements itself, the internet is evolving into three distinct realities:

1. A Snake Eating Its Own Tail (The Data Drought)

When users stop clicking, ad revenue and subscriptions dry up, removing the financial incentive for humans to research and publish original work online. As human-made content dwindles, AI models are forced to train on and recycle AI-generated outputs. Over time, this feedback loop degrades the accuracy and quality of AI answers—meaning Google is slowly consuming the very human supply chain its search product depends on to remain useful.

2. A Web with a Landlord

Instead of the old web where Google presented 10 links and the reader judged which source to trust, the AI now decides what gets cited and what disappears. A tiny tier of national outlets is paid on contract to continuously feed the machine, while the rest of the web remains invisible and unpaid. Twenty-five years ago, Google’s deal with publishers was “We will send you readers.” Today, the deal is “We decide if you exist.”

3. The Rise of the Direct Audience

The only part of the internet that survives Google’s closed loop is the territory an AI intermediary cannot stand in front of: direct human relationships. Email newsletters, direct subscriptions, social media communities, and owned audiences are becoming more vital than ever. Because a loyal audience seeking a specific human voice or creator cannot be replaced by an AI summary, the future of digital media belongs to direct creator-to-audience channels.