Google goes all-in on Gemini
Google led the day with a sweeping push to put its generative AI into everything you touch. The standout is a fully redesigned Google Home Speaker, landing June 29 for $99.99, that finally retires the decade-old Assistant in favor of a chattier, smarter "Gemini for Home." It arrives alongside a bigger software wave: Google officially shipped Android 17 and Wear OS 7, rolling out first to Pixel phones and watches. The releases bring bubble-style multitasking, deeper Gemini integration, and a June Pixel Drop loaded with Google's latest AI models and hardware perks. Taken together, it's the clearest sign yet that Gemini, not the old Assistant, is now the front door to Google's ecosystem.
The wearables and AR bet
Snap finally made its long-teased AR ambitions concrete, unveiling Specs, its first augmented-reality glasses aimed at everyday consumers. The catch is the price: $2,195, with preorders open now and a launch this fall in the US, UK, and France. It's a bold, expensive swing at a category that has humbled bigger players — and a bet that consumers are ready to wear computing on their faces.
Robots that program themselves
Nvidia stole the robotics spotlight twice. The company showed off a new generation of AI robots that teach themselves delicate manual tasks — including snapping a graphics card into a motherboard, effectively building the very PCs that train them. In parallel, new Nvidia research lets robots improve their own grasping by writing their own code, a promising route around one of robotics' oldest weaknesses: handling oddly shaped objects without fumbling. France's Genesis AI added to the momentum with Eno, a general-purpose humanoid that deliberately avoids looking human, with a commercial rollout planned for 2026. Underpinning all of it is unglamorous grunt work: a company called XDOF is now being paid by multiple AI labs to collect the real-world training data smarter robots depend on.
Quantum inches toward useful
The quantum drumbeat grew louder. Amazon's AI chief Peter DeSantis predicts "commercially useful" quantum computers within five to seven years — a notably concrete timeline from a major cloud player. Supporting that future, Atom Computing and Nu Quantum announced a partnership to build the networking technology needed to link quantum machines together, a key step toward scaling beyond single devices.
Also worth knowing
Researchers at Oregon State built a tiny brain-inspired device that lets an AI sensor see, remember, and process in one place — a potential efficiency leap for edge AI. Uber is expanding its premium robotaxi service to Houston by mid-2027, its second US city beyond San Francisco. And for PC enthusiasts, Nvidia's RTX Remix 1.5 adds compression that shrinks ray-traced mod file sizes by up to 37%, making modern makeovers of classic games far easier to share.
The through-line today: AI is moving off the screen and into the physical world — your speaker, your watch, your face, and the robots quietly learning to build the machines that power it all.