DeepSeek Delays AI Model Launch Due to Huawei Chip Supply Problems

DeepSeek Delays AI Model Launch Due to Huawei Chip Supply Problems


A big AI company from China, DeepSeek, has stopped the launch of its new AI model. The reason? Problems with the chips made by Huawei.


What Happened?

DeepSeek planned to use Huawei Ascend 910 chips to run (or “infer”) its AI model after training it. But these chips had many technical issues. The company said the Huawei chips were slower than expected and caused stability problems when running large AI tasks.


Why This Matters

Artificial Intelligence models need very powerful chips. If the chips are slow, the AI cannot answer fast, and it costs more money to run.

DeepSeek wanted to avoid using Nvidia chips because the U.S. has made export rules that limit how many high-end Nvidia chips can be sold to China. But now, because of Huawei’s problems, DeepSeek is still using Nvidia chips for training the model.

You can read about the Huawei Ascend chip technology on Huawei’s official site.


U.S.–China Tech Tensions

This delay shows how the tech fight between the U.S. and China is still affecting AI companies.

  • The U.S. wants to control sales of top AI chips to China.

  • China wants to make its own chips so it does not need U.S. technology.

  • Huawei is one of the main companies trying to make high-end AI chips for Chinese firms.


Impact on AI Industry

DeepSeek is one of the fastest-growing AI companies in China. It became well-known after launching earlier models that could answer questions, write articles, and make code — much like ChatGPT.

A delay means:

  • New features will take longer to reach customers.

  • China’s AI progress might slow down.

  • Nvidia might still keep a strong role in AI training, even in China.


The Technical Challenge

Training an AI model means teaching it with large amounts of text, images, and code. This is done on powerful chips like Nvidia’s A100 or H100.

Running the model (inference) means using it after training. Huawei wanted its Ascend chips to replace Nvidia for this stage, but DeepSeek found that:

  • The chips overheated during heavy use.

  • Software support was not as strong as Nvidia’s CUDA platform.

  • Some AI features did not work as expected.


Future Plans

DeepSeek has not given a new date for launching the AI model. The company said it will keep working with both Huawei and Nvidia until it can make a stable and fast AI product.

This news was also covered by Reuters and South China Morning Post, which explain how the global AI chip supply chain is facing big changes.


Why Users Should Care

If you use AI tools, this news might seem far away. But AI models from China could compete with U.S. models, giving users more choices and lower prices. If delays keep happening, U.S. AI companies like OpenAI and Anthropic may stay ahead in the race.


Quick Summary Table

Topic Details
Company DeepSeek (China)
Issue Delay in AI model launch
Cause Huawei Ascend chip problems
Current Solution Using Nvidia chips for training
Bigger Picture U.S.–China AI tech war

Outbound Links Used

Previous Post Next Post

যোগাযোগ ফর্ম