Tag: #space-computing23 results
Star Origin has raised tens of millions in seed and angel rounds to develop space-based computing satellites. The company aims to deploy GPU-based data centers in orbit to run large language models directly in space.
Space computing is emerging as a critical infrastructure, with major players like SpaceX and Google exploring orbital data centers. This shift marks the transition of space from an exploration target to a scalable infrastructure space for AI and computing.
Elon Musk says SpaceX and xAI will use Nvidia AI accelerators exclusively for training and inference because they consider Nvidia's architecture the best available. Musk also says an optimized Vera Rubin NVL72 system will be launched into space next year.
The Beijing Space Computing Innovation Center has been established to lead the development of space-based AI chips, large space models, and integrated space-ground computing networks. It operates under a 'company + alliance' model to drive commercialization and industrial standards for space computing.
Elon Musk sold 220,000 GPUs to Anthropic for Claude AI. Paid Claude users woke up to doubled 5-hour rate limits and faster speeds. The companies are partnering to build space-based computing infrastructure.
Xingce Future released the Xingxi 06Pro space-based computing platform for in-orbit AI large model deployment. They announced a 'sense-compute integration' plan, partnerships with Puxing Aerospace for 1024 smart satellites, and Tsinghua University for integrated constellations. Deployments include forest fire detection and grid monitoring, with a new strategic deal.
Intel has joined the Terafab project in collaboration with SpaceX, xAI, and Tesla. This move supports global acceleration in space computing layouts. It aims to build integrated earth-space computing networks for advanced AI infrastructure.
Elon Musk plans to invest $20 billion in a self-built chip factory. The facility aims to produce 200 billion chips annually, with 80% operating in space, likely for satellite networks. This scales custom silicon for orbital computing infrastructure.
Chengdu Guoxing Aerospace and Zhijiang Lab launched 12 computing satellites, world's first space AI constellation with 80B param models and 5 POPS total compute. Enables on-orbit inference for massive satellite data. Plans scale to thousands for strategic AI sovereignty.
Nvidia CEO Jensen Huang highlights orbital computing's potential for direct space data processing as a logical AI step. He identifies space cooling via radiation as the biggest challenge, requiring large costly radiators. Nvidia has deployed CUDA-based systems on satellites for AI imaging tasks.