Expired50% confidenceFactExact time

AI training clusters (such as NVIDIA H100/H200 racks) can achieve power densities exceeding 100kW per rack, whereas traditional server racks are typically only 5–10kW, a difference of 10–20 times

1

Sources

50%

Confidence

Medium-term (~90 days)

Relevance

7/4/2026

First Seen

Valid until: 10/2/2026(expired)

Sources

OpenAI CFO深度访谈：IPO态度、算力短缺与Jony Ive新硬件全解析

bilibili阅词6/2/2026

Related Claims

UnverifiedTraditional rack power is approximately 5–10 kW, while AI racks equipped with H100 GPUs can reach 40–80 kW, with liquid-cooled racks even exceeding 100 kW.83% similar UnverifiedA single large AI training cluster can have a sustained power demand of 200–500 megawatts, equivalent to the electricity consumption of a mid-sized city.76% similar Unverified传统空气冷却的物理极限约在每机架25-30千瓦，而AI工作负载的功率密度已突破100千瓦/机架70% similar Unverified装满B200芯片的机架功率可达120千瓦，是传统服务器机架的十倍以上68% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/68452

API

curl https://kongchang.com/api/v1/knowledge/claims/68452

MCP

get_claim(id=68452)