DigestAI news desk

Cut through the AI noise.

Hardware & Compute

SemiAnalysis reveals China’s 24GW datacenter capacity, surpassing EMEA

China’s datacenter capacity now exceeds 24 gigawatts, surpassing EMEA and the rest of Asia combined, according to SemiAnalysis’ new China Datacenter Model. The report tracks over 1,000 facilities across 60+ operators, revealing a market shaped by four eras: telecom-dominated retail hosting (pre-2015), a cloud land grab (2015–2021), a digestion phase (2022–2023), and the current AI supercycle…

1 source

Key points

  • China’s datacenter capacity now exceeds 24GW, surpassing EMEA and the rest of Asia combined, per SemiAnalysis’ new model
  • ByteDance occupies ~20% of delivered capacity, renting nearly all of it, making it the largest tenant for wholesale colocation providers
  • AI-driven buildout is concentrated in *Eastern Data, Western Compute* hubs, with Inner Mongolia leading due to cheap power and quotas

The AI-driven buildout is concentrated in designated hubs under the Eastern Data, Western Compute policy, with Inner Mongolia emerging as a key destination due to cheap power and abundant quotas. Hyperscalers like Alibaba, Tencent, and Baidu are expanding self-built campuses and leasing capacity at unprecedented speeds—GDS and VNET signed 1.3GW of wholesale orders in the first half of 2026 alone. Despite high vacancy rates in legacy retail racks, wholesale AI demand is filling new facilities at 70%+ utilization, while construction times have dropped to under 12 months. The report also highlights China’s cost advantages, with capex falling as modular and prefabricated designs accelerate deployment.

The story so far

2 episodes →
  1. SemiAnalysis reveals China’s 24GW datacenter capacity, surpassing EMEAthis story
Read the original at SemiAnalysis · by EverlynOpen source ↗
Topics · follow one to build your own front page

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Hardware & Compute

All →

Related stories