Recently, Starcloud, an innovative orbital data center startup supported by Nvidia, made a groundbreaking announcement. Last month's midpoint saw the successful launch of a satellite outfitted with Nvidia's state-of-the-art H100 chip. This marked a significant milestone, as it completed humanity's pioneering experiment in training a large language model in the vast expanse of space.

Presently, the satellite is actively running applications of Google's open-source model, Gemma. Additionally, it has embarked on training the NanoGPT model, a creation of former OpenAI member Andrej Karpathy. Using the complete works of Shakespeare as its training dataset, the model has now acquired the ability to express itself in a Shakespearean-style of English, a testament to the remarkable capabilities of AI in space.

Starcloud has ambitious plans for the future. It envisions constructing a 5-gigawatt orbital data center, harnessing the power of solar panels. This forward-thinking approach promises construction and operational costs that are substantially lower than those of comparable ground-based facilities, making it a cost-effective and sustainable solution.

Furthermore, the company has set its sights on launching its next-generation satellite in October 2026. This advanced satellite will be equipped with multiple H100 chips and will incorporate Nvidia's cutting-edge Blackwell platform. This strategic move aims to further expand the scale and scope of AI training in space, opening up new frontiers in artificial intelligence and space exploration.