Google's Pixel 11 makes Gemini the centerpiece of the AI phone war

Google spent the day showing off its Pixel 11 lineup — and the pitch is almost entirely AI. The company put its most agentic Gemini yet at the center of the phones, unveiled a sign-language translation model that finally brings the tech to consumer products, and confirmed the memory shortage is pushing prices up across the industry.

Google's Pixel 11 makes Gemini the centerpiece of the AI phone war. Alphabet unveiled four phones — the Pixel 11, 11 Pro, 11 Pro XL and 11 Pro Fold — with the Tensor G6 chip, but the real story is Gemini Intelligence, a suite of agentic features designed to understand what you're doing on the phone and take action across apps. It can connect information from messages, calendars and maps, then carry out next steps: checking availability, comparing travel options or starting a restaurant reservation, all without opening a chatbot. Devices chief Rick Osterloh told CNBC he sees this as a fundamental shift — "a totally new way to interact with a computer" — and that Gemini could "someday become the primary way that people use their phones." The timing is deliberate: Google is shipping weeks before Apple rolls out a rebuilt Siri powered by Gemini models, putting the two companies in the odd position of competing head-on while Google remains one of Apple's biggest AI suppliers. Osterloh insists the experiences will diverge — "Pixel and iPhone are going in very different directions" — pointing to scam detection, voice transcription and deeper Gemini integration across Android as Android-only advantages. One caution from the research side: Counterpoint says AI still ranks only sixth or seventh among consumers' reasons to upgrade, so capable agents alone may not move the needle yet. The launch also carries a price hike — the Pixel 11 starts at $899, up $100 — and Osterloh confirmed the memory shortage in RAM and flash is forcing further increases across the industry.

Google DeepMind put sign language AI into real products for the first time. The lab introduced SL2T, a sign-language-to-text translation model that now powers sign-to-text dictation in Gboard and Live Transcribe on the Pixel 11, starting with American Sign Language to English. Trained on more than 100,000 hours of data across over 50 sign languages — roughly a quarter of it ASL — the model scores 70 BLEURT zero-shot on the FLEURS-ASL benchmark, which Google says is far beyond anything previously reported. The design choices matter as much as the numbers: an on-device model tracks pose landmarks so only geometric coordinates, never video, leave the phone, and SL2T translates directly from those landmarks rather than through the intermediate "gloss" annotations used in prior work, which miss facial expressions and spatial grammar. Google says the feature lets Deaf users sign to search the web, draft messages or prompt Gemini anywhere they'd normally type, and it built the release with an advisory committee of Deaf organizations. More sign languages, more devices and sign-language generation are promised next.

What to watch: whether Apple's Gemini-powered Siri, due in weeks, actually mirrors what Google just shipped on Pixel — and whether either company can turn agentic AI into a reason people buy new phones.

AI is finally becoming the interface itself — will agents make you upgrade your phone? Tell us in the comments.

Sources: CNBC · TechCrunch · Engadget via Techmeme · The Verge · Google DeepMind · SiliconANGLE