DeepSeek plans 160,000-chip Huawei Ascend cluster in Inner Mongolia for inference
DeepSeek is planning to deploy at least 160,000 Huawei Ascend-950DT accelerators at a massive data center under construction in Inner Mongolia, according to Bloomberg sources cited across multiple outlets. The cluster would be dedicated to AI inference rather than model training, and if completed would rank among the largest known deployments of Huawei AI chips worldwide. The move signals deepening reliance on domestic silicon as China pursues alternatives to Nvidia hardware for large-scale inference workloads. However, reporting from The Decoder cautions that Huawei production bottlenecks and memory shortages could delay full chip delivery by more than a year, meaning the announced scale may materialize gradually rather than all at once.
DeepSeek plans 160,000-chip Huawei Ascend cluster in Inner Mongolia for inference
DeepSeek plans to deploy at least 160,000 Huawei Ascend-950DT chips at a data center in Inner Mongolia, with the cluster limited to inference rather than training per Bloomberg. If built, it would be the largest known Huawei AI chip cluster, though production limits and memory shortages may delay full delivery for over a year.
Key takeaway
DeepSeek's planned 160,000-chip Huawei inference cluster in Inner Mongolia would be China's largest known Ascend deployment if built, testing domestic AI silicon at unprecedented scale.
What happened
According to Bloomberg sources reported by Techmeme and Bloomberg Technology, DeepSeek plans to deploy at least 160,000 of Huawei Technologies Co.'s top accelerators at a massive data center it is building in Inner Mongolia.
The Decoder reports the chips are Huawei Ascend-950DT units and that the cluster is intended for inference only, not training; if completed, it would be the largest known Huawei AI chip cluster.
Evidence
DeepSeek plans to deploy at least 160,000 Huawei accelerators at an Inner Mongolia data center
Bloomberg Technology · attributed
DeepSeek plans to deploy at least 160,000 of Huawei Technologies Co.'s top accelerators at a massive data center it's building in Inner Mongolia
The cluster would use Huawei Ascend-950DT chips for inference only, not training
The Decoder · attributed
Deepseek wants to put 160,000 Huawei Ascend-950DT chips into an Inner Mongolia data center for inference only, not training
If built, the deployment would be among the largest known Huawei AI chip clusters
Bloomberg Technology · attributed
which could create one of the largest known clusters of Huawei AI chips and advance China's efforts to replace Nvidia Corp.
Huawei production bottlenecks may delay full chip delivery for over a year
The Decoder · attributed
But production bottlenecks mean Huawei probably can't deliver for over a year
Bloomberg sources first reported the 160K+ Ascend 950DT chip plan
Techmeme · attributed
Sources: DeepSeek plans to use 160K+ of Huawei's Ascend 950DT chips at an Inner Mongolia data center, which would create one of the largest Huawei chip clusters (Mackenzie Hawkins/Bloomberg)
Why it matters
If realized, the cluster would advance China's push to replace Nvidia in large-scale inference and stress-test Huawei's manufacturing capacity under real deployment pressure.
Limits and uncertainties
The Decoder reports production bottlenecks and memory shortages may delay full delivery for over a year.
The plan is attributed to Bloomberg sources; the full 160,000-chip build-out is not confirmed as already completed.
Practical implications
Operators tracking China AI infrastructure should monitor Huawei Ascend-950DT delivery timelines and phased cluster rollouts.
Inference-focused builders should watch whether DeepSeek publishes performance or reliability benchmarks on domestic Huawei hardware.
What to watch
Huawei chip delivery pace and any official DeepSeek confirmation of Inner Mongolia cluster timelines.
Whether the facility reaches the full 160,000-chip target or scales in phases amid reported production constraints.
