Search looks less like queries, more like conversation
No limiar de uma transformação que seus próprios executivos comparam à maior reinvenção da busca em 25 anos, o Google anunciou uma parceria bilionária com a Blackstone e uma nova geração de ferramentas de inteligência artificial que prometem responder perguntas sem encaminhar o usuário a lugar algum. O mercado reagiu com cautela, e editoras de dois continentes ergueram a voz: quem alimentou a máquina com seu trabalho agora vê os leitores ficarem para trás. A questão que paira não é apenas tecnológica — é sobre quem colhe os frutos do conhecimento coletivo e quem arca com o custo de produzi-lo.
- O Google anunciou uma parceria de US$ 5 bilhões com a Blackstone para construir data centers de IA com 500 MW de capacidade até 2027, sinalizando que a empresa aposta sua sobrevivência na infraestrutura que controla.
- As ações do Google caíram imediatamente, com investidores preocupados com o peso do investimento em infraestrutura frente a um modelo de receita ainda incerto — e a CoreWeave recuou 3,82% no mesmo movimento.
- 58% das buscas agora terminam sem nenhum clique em site externo, dado que virou munição jurídica: a Penske Media já entrou com ação nos EUA, e editoras europeias escalaram queixa formal à Comissão Europeia.
- O Conselho Europeu de Editores acusa o Google de usar conteúdo jornalístico para treinar e alimentar resumos de IA sem qualquer compensação às redações que produziram esse material.
- Com 2,5 bilhões de usuários no AI Overviews e 900 milhões no Gemini, o Google consolida escala impressionante — mas cada resposta direta que oferece é um clique a menos para quem escreveu a informação original.
O Google anunciou na terça-feira uma reformulação profunda de seu mecanismo de busca e uma grande parceria de infraestrutura, movimentos que abalaram investidores e acenderam alertas entre editoras e reguladores em múltiplos continentes. As ações da empresa recuaram enquanto o mercado absorvia a dimensão da transformação — descrita por um executivo da própria companhia como a maior reinvenção da caixa de busca em 25 anos.
No centro do anúncio está uma parceria de US$ 5 bilhões com a Blackstone para construir data centers dedicados à inteligência artificial, com capacidade de 500 megawatts prevista para 2027. As instalações serão equipadas com os chips Tensor Processing Units, desenvolvidos pelo Google para cargas de trabalho de IA em larga escala. Benjamin Sloss liderará o novo empreendimento como CEO.
No evento anual de desenvolvedores, Sundar Pichai apresentou uma série de ferramentas: o Docs Live, que transcreve fala em texto; o Ask YouTube, que permite consultar vídeos como numa conversa; e o Gemini 3.5 Flash, versão mais rápida e barata do modelo principal. Demis Hassabis, do DeepMind, descreveu o Gemini Omni — treinado em vídeo — como um passo em direção à simulação do mundo físico.
Os números são expressivos: 900 milhões de usuários mensais no Gemini, 2,5 bilhões no AI Overviews e cerca de 1 bilhão no AI Mode. Mas um dado inquieta: 58% das buscas agora terminam sem nenhum clique em site externo. Esse número embasou uma ação judicial movida pela Penske Media nos EUA e resume a tensão central da transformação — um buscador que responde diretamente significa menos visitas às editoras cujo conteúdo treinou a IA.
O desafio mais duradouro pode vir dos reguladores. O Conselho Europeu de Editores escalou sua queixa à Comissão Europeia, acusando o Google de usar conteúdo jornalístico para alimentar resumos de IA sem compensar quem o produziu. Organizações de mídia na Europa e nos EUA argumentam que o Google extrai valor de seu trabalho enquanto torna cada vez menos provável que os leitores cheguem até elas — um acerto de contas que pode custar muito mais do que qualquer queda no mercado.
Google announced a sweeping overhaul of its search engine and a major infrastructure partnership on Tuesday, moves that rattled investors and set off alarms among publishers and regulators across multiple continents. The company's stock fell as markets absorbed the scale of the transformation—a shift so fundamental that one of Google's own executives called it the biggest reimagining of the search box in a quarter-century.
At the heart of the announcement was a five-billion-dollar partnership with Blackstone, the investment giant, to build data centers purpose-built for artificial intelligence. Blackstone will fund the construction of facilities capable of delivering 500 megawatts of computing power by 2027, equipped with Google's custom-designed Tensor Processing Units—specialized chips built to run AI workloads at scale. Benjamin Sloss will lead the new venture as chief executive. The partnership represents a bet that the future of search, and of Google's business, depends on owning the infrastructure that powers it.
Google's leadership unveiled the new search experience at the company's annual developer conference. Sundar Pichai, the chief executive of Alphabet, introduced a suite of AI-driven tools: Docs Live, which transcribes speech into written documents; Ask YouTube, which lets users query video content as if having a conversation; and Gemini 3.5 Flash, a faster, cheaper version of Google's flagship AI model designed for routine automated tasks. The company also rolled out Gemini Omni, a model trained on video, which Demis Hassabis of DeepMind described as a step toward simulating the physical world itself.
The numbers behind these products are staggering. Gemini, Google's chatbot, now has 900 million monthly users. The AI Overviews feature in search—which generates summaries instead of listing links—reaches 2.5 billion users each month. A separate AI Mode has accumulated roughly one billion. Yet there is a troubling statistic buried in the data: 58 percent of searches now end without a click to any website. That figure has become the basis of a lawsuit filed by Penske Media in the United States, and it signals the core tension at the heart of Google's transformation. The company is building a search engine that answers questions directly, which means fewer people visit the publishers whose content trained the AI in the first place.
Nick Fox, a vice president at Google, framed the shift as inevitable progress. He called the introduction of AI agents—software that can make reservations, contact professionals, and compare prices across a universal shopping basket—the "biggest reinvention of the search box in 25 years." Pichai echoed the sentiment in remarks to journalists, noting that search "looks less and less like a series of individual queries and more and more like a continuous conversation." The company also cut prices on its premium AI subscription, dropping the AI Ultra plan from $250 to $200 per month and introducing a $100 tier for developers and workers.
The market's reaction was swift. Google's stock declined as investors weighed the capital intensity of the infrastructure buildout against the uncertain revenue model of a search engine that increasingly answers questions without directing traffic elsewhere. CoreWeave, a competitor in cloud computing, fell 3.82 percent in sympathy.
But the steeper challenge may come from regulators and publishers. The European Council of Editors has escalated its complaint to the European Commission, accusing Google of harvesting journalistic content to fuel AI summaries without compensating the writers and outlets that produced the original reporting. Media organizations across Europe and the United States have raised similar objections, arguing that Google is extracting value from their work while simultaneously making it less likely that readers will visit their sites. The company faces potential legal action and regulatory intervention on multiple fronts—a reckoning that may prove far more costly than any stock decline.
Notable Quotes
The introduction of AI agents represents the biggest reinvention of the search box in 25 years— Nick Fox, Google vice president
Search looks less and less like a series of individual queries and more and more like a continuous conversation— Sundar Pichai, CEO of Alphabet
The Hearth Conversation Another angle on the story
Why did Google's stock fall if the company is announcing such ambitious new products?
Because the market sees two problems at once. First, the infrastructure costs are enormous—five billion dollars is just the beginning. Second, and more troubling, the new search model makes money differently. If 58 percent of searches end without anyone clicking through to a website, that's 58 percent of the traffic that used to flow to publishers, and to Google's advertising partners on those sites.
So Google is cannibalizing its own business model?
Not exactly. Google makes money from search ads, not from clicks. But the publishers whose content trained Gemini are losing traffic and revenue. That's why they're fighting back.
The European Council of Editors is suing. What's their argument?
They're saying Google is using journalistic content without permission or payment. The AI summaries that appear in search are built on reporting that journalists spent time and money to produce. Google gets the value; the publishers get nothing.
Can Google win that fight?
Legally, maybe. But politically and reputationally, it's harder. If regulators in Europe decide Google is unfairly extracting value from publishers, they could force the company to pay licensing fees or change how the AI summaries work. That would cut into the economics of the whole strategy.
What about the Blackstone partnership? Is that a sign of confidence?
It's a sign that Google believes this is the future and is willing to spend heavily to own the infrastructure. But it's also a sign of how capital-intensive AI is becoming. Five billion dollars for data centers is real money, even for Google. That's money that could have gone to shareholders or other projects.
So what happens next?
Watch the regulatory cases in Europe. If publishers win compensation rights, the entire economics of AI search change. Watch also whether the 900 million Gemini users actually generate revenue proportional to the infrastructure costs. And watch whether other search engines—or other companies—can build competitive AI search without the same legal and regulatory headwinds.