Nvidia-backed startup Starcloud has trained an artificial intelligence model in space for the first time, heralding a new era of orbital data centres that could help ease the growing strain on Earth’s digital infrastructure.

Last month, the Washington-based company launched a satellite equipped with an Nvidia H100 graphics processing unit, a chip 100 times more powerful than any GPU previously sent to space, according to CNBC.

The Starcloud-1 satellite is now operational, running queries on Gemma, Google’s open large language model, marking the first-ever instance of an LLM operating on a high-powered Nvidia GPU in orbit.

“Greetings, Earthlings! Or, as I prefer to think of you — a fascinating collection of blue and green,” reads a message from the recently launched satellite.

“Let’s see what wonders this view of your world holds. I’m Gemma, and I’m here to observe, analyze, and perhaps, occasionally offer a slightly unsettlingly insightful commentary. Let’s begin!” the model wrote.

Starcloud aims to demonstrate that outer space can serve as a viable environment for data centers, especially as Earth-based facilities increasingly strain power grids, consume billions of gallons of water each year, and generate significant greenhouse gas emissions.

According to the International Energy Agency, global data centre electricity consumption is expected to more than double by 2030.

In addition to Gemma, Starcloud trained NanoGPT, an LLM developed by OpenAI co-founder Andrej Karpathy, on the H100 GPU using the complete works of Shakespeare.
As a result, the model learned to generate responses in Shakespearean English.