Unverified50% confidenceFactExact time
The NVIDIA GB200 NVL72 rack delivers about 180 PFLOPS of dense AI compute at 145 kilowatts of power.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/19/2026
First Seen
Valid until: 2/17/2027
Sources
Huawei Ascend's Paper Compute Beats Nvidia by 1.7x—So Why Does DeepSeek Only Trust It for Inference?
bilibili智视界AI时代11/19/2026
Related Entities
Related Claims
UnverifiedThe Ascend 950PR is equipped with 128 GB of VRAM.73% similarUnverifiedThe P100 uses four HBM2 stacks with a total bandwidth of 732 GB/s, far exceeding the 320 GB/s of the consumer-grade GTX 1080 from the same period.73% similarUnverifiedEach of the 4 modified 3080Ti 16G graphics cards has a power consumption of approximately 115 watts, and the full-load power consumption of the complete machine is around 700+ watts.72% similarUnverifiedThe full weights of LLaMA 2 (70B parameter version) amount to approximately 130 GB.72% similarUnverifiedA full PyTorch installation typically exceeds 700 MB, and TensorFlow is in the same order of magnitude.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1190030API
curl https://kongchang.com/api/v1/knowledge/claims/1190030MCP
get_claim(id=1190030)