Huawei doubles down on scale: Atlas 950 SuperCluster packs even more chips than before
Huawei introduced its new AI supercomputer, the Atlas 950 SuperCluster, at the Connect 2025 conference. The system uses more than 524,000 Ascend-950DT chips and, according to Huawei, can deliver up to 524 FP8 exaFLOPS for training and 1 FP4 zettaFLOP for inference. That makes it capable of handling models with trillions of parameters, according to Huawei.
In contrast to Nvidia's Rubin systems, Huawei continues to focus on scale over individual chip performance. As reported by Tom's Hardware, the setup requires about 64,000 square meters of floor space. Huawei is already planning a follow-up: the Atlas 960, slated for 2027, will include more than one million chips.
AI News Without the Hype – Curated by Humans
Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.
Subscribe nowRead on for the full picture.
Subscribe for hype-free coverage.
- Full access to every article on THE DECODER
- No ads
- Join the comments and community discussions
- A weekly AI news recap via mail
- 6x/year: "AI Radar" — deep dives on the AI topics that matter most
- Daily AI news, always up to date
- Our full ten-year archive
- Covered by a team with 10+ years in AI