Compute/press release/2026-09-22

DeepSeek Adopts Huawei Chips for AI Model Training

Chinese AI company DeepSeek is reportedly increasing its adoption of Huawei's Ascend chips for training its artificial intelligence models. This strategic move, confirmed by DeepSeek CEO Tong Yi to The Information, aims to reduce the company's reliance on Nvidia GPUs, which are impacted by U.S. export controls. While some reports highlight a broad shift towards domestic alternatives, the primary motivation is seen as bypassing these restrictions. This development underscores the growing trend among Chinese tech firms to seek indigenous solutions for advanced semiconductor technology. DeepSeek plans to use Huawei chips for its next generation of models.

4 articles from 4 outlets covered this story. Their coverage differs on 2 points. The underlying claim is sourced from a press release.

What do all outlets agree on?

4 outlets covered “DeepSeek Adopts Huawei Chips for AI Model Training”. All of them report the following:

DeepSeek is a Chinese AI company
DeepSeek will use Huawei chips for AI model training
This represents a shift towards domestic chips

Did outlets disagree about this?

Yes. Coverage of “DeepSeek Adopts Huawei Chips for AI Model Training” differs on 2 points. Each account below is how a different outlet described the same event:

The extent of DeepSeek's shift to Huawei chips

The Standard (HK) (describes it as using "more domestic chips"), Huawei Central (describes it as using "Huawei chips widely for AI training over Nvidia"), theinformation.com (frames it as "Bets Big on Huawei Chips")

The primary motivation for DeepSeek's shift

theinformation.com (explicitly states "to Bypass U.S. Export Controls"), while 조선일보, The Standard (HK), and Huawei Central focus on the general move to domestic chips without explicitly stating the bypass of U.S. controls as the primary driver