pragmatism now trumps the dream of doing everything in-house
Numa das mais simbólicas viragens da história recente da tecnologia, a Apple revelou que a sua assistente de voz Siri será reconstruída sobre os modelos de inteligência artificial da Google — a mesma empresa que durante anos serviu de contraponto à narrativa de privacidade da Apple. O anúncio, feito na última conferência anual de programadores e marcado pela despedida emocional de Tim Cook do palco principal, expõe a tensão permanente entre ideais corporativos e as exigências pragmáticas de um mercado em aceleração. A Apple aposta que chegar mais tarde, mas com dois mil e quinhentos milhões de dispositivos prontos a receber a tecnologia, vale mais do que ter chegado primeiro com promessas por cumprir.
- A Apple admite, implicitamente, que a sua aposta interna em IA falhou — e que o custo dessa falha foi de 250 milhões de dólares pagos em acordo judicial a clientes que se sentiram enganados.
- A parceria com a Google cria uma tensão filosófica difícil de ignorar: a empresa que construiu a sua identidade sobre a privacidade passa a depender da infraestrutura de dados do seu maior rival.
- Tim Cook despediu-se visivelmente emocionado do palco da WWDC, encerrando quinze anos de liderança numa cena que misturou triunfo tecnológico com melancolia humana.
- A SiriAI promete chegar ao outono na maioria dos mercados, mas regulações na China e na Europa adiam o acesso a centenas de milhões de utilizadores potenciais.
- Analistas argumentam que a estratégia tardia da Apple pode ser, paradoxalmente, uma vantagem: evitar investimentos massivos em infraestrutura enquanto colhe os frutos de uma adoção de IA já madura.
A Apple anunciou na segunda-feira uma reconstrução completa da Siri, transformando-a num sistema de inteligência artificial conversacional alimentado pela tecnologia Gemini da Google. A nova ferramenta, rebatizada de SiriAI, chegará no outono à maioria dos mercados — com exceção da China e da Europa, onde obstáculos regulatórios atrasam o lançamento. Entre as capacidades prometidas estão a redação de emails, pesquisa entre aplicações e interpretação do conteúdo visível no ecrã, mas apenas em dispositivos mais recentes com os chips mais avançados da empresa.
A parceria representa uma concessão significativa. A Apple construiu parte da sua identidade na promessa de que protege os dados dos utilizadores de forma diferente da Google — que a privacidade, e não a vigilância, é o alicerce do seu ecossistema. No palco da conferência de programadores, Craig Federighi tentou reencadrar a decisão como princípio e não como compromisso, insistindo que a empresa estava a abordar a IA com ponderação e consideração pelas pessoas. A Apple sublinhou que as funções de IA funcionarão dentro do seu ecossistema fechado, e não nos servidores da Google.
O anúncio ganhou um peso adicional por marcar a última aparição de Tim Cook no keynote principal. Cook, que lidera a Apple desde 2011, cederá o lugar de diretor executivo a John Ternus, responsável pela engenharia de hardware, a partir de 1 de setembro. No momento em que saiu ao encontro dos programadores reunidos no Apple Park, Cook não conseguiu esconder a emoção, num instante que simbolizou uma transição mais ampla — não apenas de liderança, mas de era.
O lançamento da SiriAI chega também como lembrete de um tropeção recente. Há dois anos, na mesma conferência, Cook prometera uma integração massiva de IA nos produtos Apple. Essa promessa nunca se materializou, custando à empresa um acordo judicial de 250 milhões de dólares com clientes norte-americanos. Ainda assim, muitos analistas veem a chegada tardia como menos prejudicial do que parece: com mais de dois mil e quinhentos milhões de dispositivos ativos em todo o mundo, a Apple tem uma base instalada enorme pronta a adotar funcionalidades de IA quando estas atingirem maturidade. Ao licenciar tecnologia já provada em vez de construir a sua própria infraestrutura, a empresa evita os investimentos de centenas de milhares de milhões que os seus concorrentes enfrentam — e posiciona-se para colher os frutos sem suportar o custo total da corrida.
Apple announced on Monday that it had rebuilt Siri from the ground up, transforming the voice assistant into a conversational AI system powered by Google's Gemini technology—a striking reversal for a company that has long positioned itself as a privacy-first alternative to its rivals. The new tool, rebranded as SiriAI, will arrive in the fall across most of the world, though regulatory hurdles will keep it from China and Europe for now. It promises to handle email drafting, search across apps, and interpret what appears on your screen, but only on newer Apple devices equipped with the company's latest chips.
The partnership represents a significant concession. Apple has built its brand partly on the claim that it protects user data differently than Google does—that privacy, not surveillance, is the foundation of its ecosystem. Yet here the company was, standing on stage at its annual developer conference, explaining that it had licensed Google's AI models and adapted them for its own use. Apple's software chief, Craig Federighi, tried to reframe this as principle rather than compromise, arguing that the company was pursuing AI thoughtfully, with real consideration for people, rather than chasing the technology for its own sake. The company emphasized repeatedly that these AI functions would run only within Apple's closed ecosystem, not on Google's servers.
The announcement carried particular weight because it marked Tim Cook's final appearance at the keynote address. Cook, who has led Apple since 2011, will hand over the role of chief executive to John Ternus, the head of hardware engineering, starting September 1st. During the presentation, Cook stepped out to greet the assembled developers at Apple Park, visibly emotional, wiping away tears as the crowd applauded. It was a moment of transition—not just for the company, but for its public face.
The timing of this AI rollout also served as a reminder of Apple's stumble in the race to deploy generative AI at scale. Two years earlier, at the same conference, Cook had announced that Apple was on the verge of integrating AI massively into its products, riding the wave of excitement that followed ChatGPT's release. That promise never materialized. The redesigned Siri that Apple had touted never arrived. The delay cost the company dearly: it settled a class-action lawsuit from American customers this year for $250 million, acknowledging the broken commitment.
Yet many analysts see Apple's late arrival as less of a liability than it might appear. With more than 2.5 billion active devices worldwide, Apple has an enormous installed base ready to adopt AI features once they reach maturity. The company is making a calculated bet: by licensing proven technology from Google rather than building and maintaining its own massive AI infrastructure, Apple can avoid the hundreds of billions in annual spending that competitors face while still capturing the benefits when consumer AI adoption becomes routine. John Gruber, a widely read technology commentator, framed it this way: Apple is positioned to reap the rewards without bearing the full cost of the infrastructure race.
The company also used the conference to highlight expanded parental controls, a move that reflects growing pressure from civil society on American tech giants to better protect minors online. Even as Apple plays catch-up in AI, it continues to position itself as the company that thinks about consequences—about the human stakes of powerful technology. Whether that positioning holds up, given the company's dependence on Google's models, remains an open question. What is clear is that Apple's strategy has shifted: pragmatism now trumps the dream of doing everything in-house.
Notable Quotes
AI is an incredibly powerful technology, capable of shaping society profoundly. But some seem to charge ahead as if chasing AI for its own sake, without real consideration for people.— Craig Federighi, Apple's software chief
Apple is making an enormous bet on AI: not having to spend hundreds of billions per year on AI infrastructure to reap the benefits.— John Gruber, technology commentator
The Hearth Conversation Another angle on the story
Why would Apple, a company that has spent years telling customers it's different from Google on privacy, turn around and build its AI on Google's technology?
Because the alternative—building competitive AI from scratch—would have cost Apple tens of billions a year in infrastructure alone. They looked at the math and chose partnership over pride.
But doesn't that undermine the privacy argument? If the AI is based on Google's models, isn't Google still involved somehow?
Apple's claim is that the models run only on Apple's devices and servers, not on Google's infrastructure. So the data stays in Apple's ecosystem. Whether that distinction matters to users is another question.
Tim Cook cried at his last keynote. Was this announcement emotional for him personally, or was it just theater?
It was his final appearance in that role after fifteen years. The emotion seemed genuine—he was saying goodbye to something that mattered to him. The AI announcement itself was just business.
Apple promised AI two years ago and failed. Why should anyone believe this will actually ship in the fall?
Because this time Apple isn't trying to build it alone. They're using proven technology from Google and adapting it. The risk is lower, the timeline is shorter. That's the whole point of the partnership.