- Dotika
- Posts
- Gemini AI powers conversational Google Maps
Gemini AI powers conversational Google Maps
ALSO : Google plans space AI data centers
Alexandre HOTTON
November 05, 2025
News of the day
1. Google Maps is integrating Gemini AI for a more conversational navigation experience, offering recommendations and precise directions using a vast database → Read more
2. Google's Project Suncatcher aims to build AI data centers in space, using solar-powered satellites and optical links to overcome energy and logistical issues → Read more
3. Replika founder launches Wabi, a platform for creating and sharing apps via prompts, raising $20M pre-seed. It aims to be the 'YouTube of apps'. → Read more
4. Microsoft's new synthetic marketplace tests AI agents, revealing vulnerabilities to manipulation and decision-making failures when overwhelmed by options.→ Read more
Our take
Hi Dotikers!
This week, Google struck hard with two announcements that perfectly illustrate the AI industry's voracious ambition and especially its main challenge: energy.
On one side, Google Maps integrates Gemini AI for its 2 billion users, transforming the app into a conversational assistant. You'll now be able to ask it vocally "an affordable Japanese restaurant on my route", ask follow-up questions, then simply say "let's go" to launch navigation. But the real game-changer is visual landmark navigation: no more "turn in 150 meters", Gemini will say "turn after Thai Siam restaurant" by analyzing 250 million places and Street View images. The goal? Make Gemini a showcase against ChatGPT by making it indispensable in daily life.
On the other side, the real bombshell: Project Suncatcher. Google wants to create satellite constellations equipped with TPU chips to run AI directly in space, powered by solar energy. The calculation is radical but logical: in the right orbit, a solar panel is 8 times more productive than on Earth, with near-continuous production. TPU chips have already survived space radiation tests, and the first prototypes launch in early 2027 with Planet. With the expected drop in launch costs (under $200/kg by the 2030s), the cost of a space data center could become comparable to a terrestrial data center.
The underlying message? According to Gartner, 40% of AI data centers will be limited by electricity shortages by 2027. While Google makes AI accessible in your pocket with Maps, it's already planning how to power tomorrow's models: at 650 km altitude, where energy is unlimited and there's no competition for resources.
A.
Meme of the day
Reply