| Mobile Web

Search results for Nvidia H100

Industry

SpaceX orbital data centre plan may need up to 42 launches a day

SpaceX\'s plan for an orbital data centre envisioned as future AI infrastructure is technically feasible, but major obstacles remain before commercialisation, an analysis said. The biggest challenge is building an unprecedented satellite constellation and launch system. The concept includes operating 1 million satellites producing 120 GW and using tens of millions to 100 million GPUs. Analysts cited heavy satellite mass, cooling limits in space, huge manufacturing and network costs, and communications latency as constraints.

AI & Enterprise

Tesla moves into modular AI data centre hardware with Megapod trademark filing

Tesla is pushing into modular artificial intelligence data centre hardware after filing a trademark for “Megapod” in the United States. The application describes a modular data centre hardware system for AI computing, including servers, networking equipment, power distribution and cooling. Tesla also describes a self-contained modular computing unit with downloadable software for monitoring and optimisation. The move would put Tesla into competition with Nvidia-backed platforms, while Tesla remains a major Nvidia customer and has faced delays in its own AI chip plans.

AI & Enterprise

Microsoft says ChatGPT power use per query may be up to 20 times lower than estimates

Microsoft released an analysis saying the electricity used to send one question to a large language model can be far lower than widely cited figures. Assuming a 300-token response, it put the median at 0.31 watt-hours, with the middle 50 percent ranging from 0.16 to 0.60 watt-hours. It said earlier estimates may have overstated usage by 4 to 20 times by not reflecting batch processing and real-world GPU utilisation.