IT·Science News
Google’s AI satellite answers its first prompt: a step toward data centres in space?
Google’s AI chip-equipped satellite has generated its first response from space, The Verge reports. In a blog post on Friday, Google said it had successfully booted up the satellite’s Tensor Processing Units (TPUs), allowing the on-board Gemma 4 model to follow a prompt and write a haiku about being in space. The Verge’s report is filed under Google’s Project Suncatcher.
The idea of running data centres in space has prominent sceptics. At the India AI Impact Summit in Delhi earlier this year, OpenAI chief executive Sam Altman told The Indian Express that space-based data centres, an idea promoted by Elon Musk, are impractical because of the cost of transport to space, German tech site heise online reported. He said nobody is discussing how to repair broken GPUs in orbit, that GPUs still fail often, and that he does not expect space data centres to pay off this decade.
“Honestly, I think with the current environment, the idea of data centers in space is ridiculous.” — Sam Altman, via heise online
How each outlet framed it
- The Verge — the milestone: a working AI model answering from orbit, with a short poem as proof.
- heise online — the doubts: cost, repairs and hardware failures, in the words of a rival AI company’s chief.
What we don't know yet: how much computing the satellite can do and how long its chips will last in orbit. Google’s first result is a single short response.