Nvidia-Backed Starcloud Trains First LLM In Space Amid Orbital Datacenter Buzz — CEO Calls It 'Significant' First Step

Starcloud, an Nvidia Corp.-backed (NASDAQ:NVDA) company, trained an AI model from orbit amid a buzz surrounding space-based AI datacenter satellites.

Starcloud Used Nvidia's H-100 Chip To Train AI From Space

The Washington-based startup launched the Nvidia H-100 GPU, which boasts 100 times the compute of other chips previously launched into orbit, CNBC reported on Wednesday. The company has been training Alphabet Inc.'s (NASDAQ:GOOGL) (NASDAQ:GOOG) Google's open LLM Gemma from orbit via its Starcloud-1 satellite.

Starcloud was also able to train an LLM created by former Tesla Inc. (NASDAQ:TSLA) AI lead, Andrej Karpathy, called NanoGPT, from orbit on Shakespeare’s works, the report suggests.

Starcloud describes itself as a company that wants to "build large data centers in space that are much more rapidly scalable and cost competitive" than their Earth-based counterparts. According to the White Paper released by the startup, the company wants to create a 2.4-mile-tall and 2.4-mile-wide, 5-gigawatt orbital data center with solar and cooling panels in orbit.

A Significant Step

Starcloud CEO and co-founder Philip Johnston hailed the achievement in a post on the social media platform X on Wednesday, calling it a "significant step on the road to moving almost all compute to space," and to stop draining Earth's resources.

We just trained the first LLM in space using an @Nvidia H100 on Starcloud-1! 🚀

We are also the first to run a version of @Google's Gemini in space!

This is a significant step on the road to moving almost all compute to space, to stop draining the energy resources of Earth and… pic.twitter.com/csc9MjDPco

— Philip Johnston (@PhilipJohnston) December 10, 2025

Elon Musk's Orbital Datacenter Goals

The news comes as SpaceX CEO Elon Musk has reiterated his ambition to put solar-powered AI satellites in space, which would be a more cost-effective solution than Earth-based data centers.

The billionaire recently shared that SpaceX was targeting 1 Megaton per year for AI satellites, which, according to Musk, "yields 100GW of AI added" in space. He also touted building factories on the lunar surface, which would facilitate AI satellite launches to deep space.

Musk also confirmed that SpaceX was gearing up for its IPO next year, targeting a valuation of $1.5 trillion. This comes after Musk had hinted at the possibility of an IPO for SpaceX during Tesla's annual shareholder meeting last month.

Nvidia vs Michael Burry

Meanwhile, "The Big Short" investor Michael Burry has raised questions about Nvidia CEO Jensen Huang‘s comments about the company’s Blackwell GPU shipments, urging users to share pictures of the GPUs being stored in "large quantities" within the U.S. and overseas.

Nvidia scores well on Momentum, while offering satisfactory Quality and Growth, but poor Value. It also has a favorable price trend in the Short, Medium and Long term. For more such insights, sign up for Benzinga Edge Stock Rankings today!

Price Action: NVDA declined 1.92% during After-hours to $180.25, according to Benzinga Pro data.

Check out more of Benzinga's Future Of Mobility coverage by following this link.

Read Next:

• Elon Musk Says He Would Rather Focus On His Businesses Than Work With DOGE Again: ‘Wouldn't Have Been Burning The Cars'

Photo courtesy: Chung-Hao Lee / Shutterstock