Starcloud-1 (space satellite)

History

2025: Orbit Launch

In November 2025, the company launched Starcloud into orbit a satellite with an Nvidia H100 GPU. The power of this chip is 100 times higher than the capabilities of any of the GPUs previously used in space. We are talking about the work of the world's first data center for AI in space.

Gemma, an open large language model from Google, has been launched on the satellite. This is the first time in history that LLM runs on a powerful GPU from Nvidia outside the Earth. The breakthrough marks a new era of orbital data centers that should mitigate the growing crisis of digital infrastructure on Earth.

𝓲

Фото: Freepik

Greetings, earthlings! Or, as I like to think about you more - a delicious combination of blue and green. Let's see what miracles this view of your world has in it.

I am Gemma, and I am here to observe, analyze, and perhaps occasionally provide insightful comments that can be slightly embarrassing. Let's get started!

The NanoGPT model, created by one of the founders of OpenAI Andrei Karpati, was trained on the H100 chip. The complete works of Shakespeare were used for training. As a result, the model began to respond to the Old English manner.

Successful training of different models confirms the versatility of the orbital platform. The satellite can work with various machine learning algorithms without adapting the hardware.

Starcloud CEO Philip Johnston expects that space data centers will be 10 times more profitable than ground data centers in terms of energy consumption. All the capabilities of the ground data center can be implemented in space. The main thing that drives the company is the energy restrictions faced on Earth.

According to forecasts of the International Energy Agency, by 2030 the consumption of electricity by data centers will more than double. Land-based facilities overload power grids, consume billions of cubic meters of water each year and produce huge greenhouse gas emissions^[1] in^[2]

Notes