DeepSeek boosts China’s chipmakers in race for affordable AI tech

Photo : ANI

For many years, Chinese chip companies like Huawei have found it hard to compete with Nvidia in making strong computer chips for artificial intelligence (AI). AI training needs a lot of computer power, where big amounts of data are used to teach AI how to make smart choices. Nvidia is the top company in this area because its special chips are very fast and powerful.

Stay informed with the latest news. Follow DXB News Network on WhatsApp Channel  

But now, DeepSeek’s AI models are giving Chinese chipmakers a new chance. DeepSeek does not need too much computer power. Instead, it focuses on “inference,” which means using AI to give answers or make decisions after learning from data. Because of this, Chinese companies can use DeepSeek to improve their AI work, even though they cannot buy the latest U.S. chips due to trade rules.

Experts believe DeepSeek will help reduce the gap between Chinese chips and American chips. Chinese chips may not be the best for AI training, but they are becoming good for AI inference, which does not need as much power. This makes them useful for many industries.

Many big Chinese chip companies, such as Huawei, Hygon, Tencent-backed EnFlame, Tsingmicro, and Moore Threads, have announced that their chips will support DeepSeek. They hope this will make their products better. However, these companies have not shared full details yet. Huawei has not made any comment, and the other companies have not answered questions.

Because DeepSeek is open-source and cheaper, more companies in China are expected to start using AI in their business. This could help China deal with U.S. trade bans on high-power chips. Even before DeepSeek became popular, many Chinese companies preferred using AI inference instead of training, as it needs less computing power. For example, Huawei’s Ascend 910B chip is already being used by companies like ByteDance for inference tasks.

In China, many industries, including car makers and telecom companies, are now planning to use DeepSeek’s AI models in their work. Experts say this is a good match for Chinese chipmakers. Even though Chinese chips are not as strong as Nvidia’s for training AI, they are good enough for inference work, which is easier and does not need as much power.

Tech expert Lian Jye Su from research company Omdia says that AI inference is more flexible and does not need very strong chips. He also says that Chinese chipmakers understand local industry needs better, which will help them use DeepSeek AI models in the right way.

China is working hard to build its own strong technology. DeepSeek’s AI models may help Chinese chip companies grow and improve their AI business, even with the limits on buying U.S. chips.