DeepSeek is one of the popular artificial intelligence companies, and is ready to offer their AI models that are equivalent to various other models on the market, and developed at a lower cost. Most recently, in an effort to reduce dependence on NVIDIA chips, DeepSeek is reported to be developing their own AI chips.
Previously, DeepSeek used NVIDIA and Huawei chips to train and operate their AI models. The initial chip being developed by DeepSeek will focus on inference support, which is to allow it to perform calculations and computing to generate answers for users.
Currently, chips to train AI models have not yet begun to be developed, and depend on other chips on the market.
Various AI companies have recently been ready to develop their own chips in order to reduce dependence on external parties, and have more control over their models, while at the same time optimizing operating costs. This move by DeepSeek, if successful, is also expected to deal a huge blow to the United States. Previously, some of the latest DeepSeek v4 models were said to be trained using Huawei chips.
With this healthy development, let's all look forward to what DeepSeek will offer next in continuing to offer powerful models at more competitive prices.
