Pichai accepts AI Mode displacing classic Google search

Sundar Pichai has stopped pretending the ten blue links are sacred. Search Engine Journal reports that Google's chief executive, asked about AI Mode cannibalising classic search, said he is comfortable with users migrating to the generative experience and treats the shift as the point, not a problem. The interview, conducted around Google's recent product cycle, marks the clearest public concession yet that the company's core product is being rebuilt around model output rather than ranked documents.

This matters because Google spent two years insisting AI Overviews would complement, not replace, the familiar results page. Pichai's tone has changed. He now frames AI Mode as the destination, with classic search a legacy interface Google is willing to let shrink. For anyone whose marketing plan still rests on organic blue-link traffic, that is the boardroom quote to circulate.

The mechanics of the shift are already visible. AI Mode answers questions inside Google's interface, summarising sources and offering follow-ups, with citations relegated to expandable panels. Click-through rates on informational queries have been falling across the industry since Overviews rolled out; Pichai's comments suggest Google has modelled the traffic loss to publishers and decided it is an acceptable cost. The company's incentive is to keep users inside Gemini-powered surfaces, where it controls the ad unit, the follow-up query, and the data exhaust.

For B2B brands the implication is sharper than for consumer publishers, who at least have scale to negotiate licensing. A reinsurer, a development bank, or a standards body has perhaps a few dozen pages that matter for the queries its buyers actually run. When AI Mode synthesises an answer about climate risk disclosure or trade finance compliance, it picks two or three sources and names them in a sidebar most users will not open. Being one of those sources is now the entire game. Being ranked fourth on a results page that fewer people see is consolation.

Financial services firms should read Pichai's comfort as a deadline. The compliance-heavy content they produce, white papers, regulatory explainers, methodology notes, is exactly the material AI Mode prefers to cite: dense, sourced, attributable. But citation goes to the institution the model already trusts on that topic, and trust is built through corroboration across the open web, not through gated PDFs. Firms that keep their best thinking behind a lead-gen form are invisible to the system now deciding which three names appear next to the answer.

Multilaterals face a different version of the same problem. UN agencies, the World Bank, the OECD, and bodies like ISO and IEEE still command authority in their domains, and models lean on them heavily for definitional and statistical queries. That advantage is fragile. AI Mode rewards freshness and structured data alongside authority; a 2019 PDF on a slow-loading subdomain loses to a 2024 blog post from a think tank that publishes in clean HTML with schema markup. The institutional voice is being out-engineered by faster competitors who understand the retrieval layer.

Industrial groups have the opposite problem: plenty of product pages, almost no analytical content the model finds useful. When a procurement lead asks AI Mode about low-carbon cement specifications or grid-scale battery chemistries, the citations go to trade press, consultancies, and academic sources. The manufacturer making the product rarely appears. Pichai's shift means that gap is no longer a missed SEO opportunity; it is a structural absence from the buyer's first and possibly only research step.

Philanthropic and policy institutions should note the second-order effect. AI Mode flattens the distinction between a foundation's own research and the secondary coverage of it. If a model cites the Financial Times summarising a Gates Foundation report rather than the report itself, the foundation loses both the citation and the framing. Owning the synthesis, publishing the explainer the model wants to quote, becomes more valuable than the underlying study.

Pichai's calm is rational. Google's revenue does not depend on classic search surviving; it depends on Google remaining the layer where queries are answered. The losers are organisations that built their digital presence for a results page that is being quietly retired. The question on the next marketing planning cycle is not how to rank, but whether the model would name you if a buyer asked.