Search ad record high, but website traffic is cut in half: Reddit, Reuters, and The Economist are collectively considering blocking Google crawlers

Reddit, Reuters, Politico, and The Economist are collectively considering blocking Google’s crawlers, because search traffic is already no longer coming back. And in the same quarter, Google Search advertising revenue hit a historic high.
(Background: Cloudflare hit with a killer move! Default “block AI crawlers” sparks a new war for paid content scraping)
(Additional context: For the first time in history! Google burned through AI computing power and saw “negative cash flow”; Q2 revenue looked strong, yet the stock fell 4.5%)

Table of contents

Toggle

• An unspoken agreement is breaking down
• The same crawler, two uses
• Moat and dead end, the same set of numbers

Google’s latest earnings show search ads raking in $63.27B, up 17% year over year and setting a record high; but at the same time, search traffic for multiple longstanding US news websites was cut in half across the board. This sounds contradictory—after all, both ends of a business shouldn’t be heaven and hell at the same time. But the logic behind it is actually simple: Google doesn’t send users out of the search results page—it keeps them in AI summaries, and under those summaries, the links are mostly not clicked.

An unspoken agreement is breaking down

Over the past two decades, there has been an unwritten understanding between the internet and Google: Google deploys crawlers to collect data and build an index; the deal is that a large amount of traffic gets sent back to websites. The understanding was never especially fair—Google always makes more than the sites—but at least both sides could survive.

Nilay Patel, editor-in-chief of The Verge, discussed this week’s streak of “Google Zero” news on the Vergecast episode that aired July 25, 2026, with host David Pierce.

“Google Zero” is a term Patel introduced in 2024, referring to the moment when Google search stops sending traffic to third-party websites. Simply put, once the search engine no longer brings any visitors to your site, your site is effectively nonexistent in Google’s eyes. Back then it sounded like alarmism; today, publishers have already written it into their strategy decks—treated as a planning assumption that traffic approaches zero.

Semrush’s statistics for US organic search traffic from June 2025 to June 2026 (meaning users come in by naturally clicking search results, not visitors brought by paid ads) are bleak: USA Today’s national site slid by nearly 50%, Business Insider fell by more than 85%, CNN dropped by about 25%, and Politico fell by 23%.

A study by Pew Research Center explains the reason: when search results don’t have AI summaries, 15% of users click through to traditional links; with AI summaries, that drops to 8%; and when users click the links “inside” the summary, only 1% do. Put plainly, Google writes the answer right at the very top of the search results, so users’ questions get solved on the spot—turning websites from a source of answers into an appendix nobody flips through.

The same crawler, two uses

Publishers can theoretically block Google’s crawlers, but the problem is that the same batch of crawlers is used to both retrieve data for building search indexes and to train AI models. To stop content from being taken by AI for free, the price is to give up even the remaining scrap of search traffic as well; the site disappears from search results entirely.

This is the most brutal structural contradiction in the whole affair: publishers don’t have an option to “block only AI while keeping search.” They only have “both” or “neither.”

Reddit was the first company to hit this wall. In 2024, Reddit and Google signed an annual licensing agreement worth about $60 million, allowing Google to use content from Reddit to train AI models. That agreement is nearing expiration. The two sides are still in talks about a renewal, but Reddit is considering not renewing, and is even blocking Google from crawling its content. After the news broke, Reddit’s stock fell by about 9% on July 22, 2026. The market’s reaction was very straightforward: for Reddit, having its content scraped away to train models may be more worth worrying about than losing search traffic.

USA Today Co. CEO Mike Reed chose a different posture. He told The New York Times, “It’s time to take a stand—enough is enough.” A report by The New York Times on July 20, 2026 said that, in addition to USA Today Co., Reuters, Politico, The Economist, and People Inc. are also considering limiting or blocking Google’s access rights. What these companies have in common is that they’re large enough to absorb the pain of traffic declines, yet they can’t afford the loss of having their content taken outright by AI.

Moat and dead end, the same set of numbers

Cloudflare CEO Matthew Prince warned long ago that AI summaries could harm the business model of the entire web. Now it doesn’t look like prophecy—it looks more like a status report. The internet used to run on “content for traffic, traffic for advertising,” but AI summaries remove that middle step: the content is still being crawled, studied, and used to answer questions, but the traffic it used to bring is gone.

The author believes this tug-of-war won’t end with a dramatic finale. A more likely direction is: big enough publishers negotiate individual licensing deals with Google one by one, like Reddit; small sites with no bargaining power can only watch as their traffic goes to zero… .