DeepSeek Developing In-House Inference Chip to Reduce Hardware Reliance

DeepSeek Developing an In-House Inference Chip

The race for artificial intelligence infrastructure in China has taken a sharp turn inward. Chinese startup DeepSeek is developing its own AI chip, a strategic shift that could eventually reduce the company’s reliance on processors supplied by Nvidia and Huawei. If successful, DeepSeek’s expansion into semiconductor development would mark a major strategic shift for a company widely hailed in China as the country’s AI champion, potentially adding to challenges faced by Chinese tech giant Huawei. While the project remains in an early stage, DeepSeek has quietly ramped up hiring for chip-design engineers in recent months, according to sources. The push places the company alongside global peers, while it simultaneously develops its own AI chip to reduce its reliance on Nvidia and Huawei chips, which it has depended on to train and run its globally popular models. Internal Hardware Realities and the Nvidia-Huawei Balance

DeepSeek’s hardware calculus came into sharper focus following reports and insights regarding the computational gap between Chinese artificial intelligence laboratories and their American counterparts. The CEO of the company has clear views on the enormous difference between Chinese AI Big Tech firms and their American counterparts: computing capacity.

Computing Capacity Gap

Detailing the hardware discrepancy, the CEO gives a data point to illustrate it: four Huawei chips equal one Nvidia chip.

The Ascent of Huawei Ascend and the V4 Model Integration

DeepSeek Releasing its V4 Model

DeepSeek has taken a new step in global technological competition by releasing its V4 model, designed specifically to operate with chips developed by Huawei, thereby reinforcing the country’s technological autonomy in the artificial intelligence sector.

DeepSeek, the Chinese startup whose low-cost AI model surprised the world last year, presented a preview of a new model adapted to Huawei’s chip technology, highlighting China’s growing self-sufficiency in the sector. The collaboration with Huawei on the new V4 model contrasts with DeepSeek’s previous reliance on Nvidia chips. Huawei stated that its processors were used in part of the training of V4-Flash, a lighter version of the model.

Global Market Realignments and Reopening Sales

Nvidia Announcing Sales of its H20 Chip to China

The shifting landscape coincides with broader commercial adjustments. Nvidia, the world’s largest company by market capitalization, announced that it will resume sales of its H20 artificial intelligence chip to China. In addition, it has presented a new model adapted to the regulatory requirements of the Asian giant. The announcement was reflected in financial markets, returning the Nasdaq to all-time highs. At the opening on Wall Street, Nvidia shares rose more than 5%, up to 172.2 dollars, a new all-time high, bringing market capitalization to a level never reached by another company. Since the beginning of the year, shares have gained more than 20%, leaving behind the impact caused by the emergence of China’s DeepSeek. Amid this vortex, Oppenheimer raised its price target for Nvidia stock from 175 to 200 dollars, while DZ Bank raised it from 160 to 195 dollars. The U.S. company noted it is submitting requests to the U.S. government to resume sales of the aforementioned H20 graphics processing unit to China, adding it expects to obtain the licenses soon. Shares of U.S.-based Nvidia slipped about 1.6% in premarket trading when news of DeepSeek’s chip push broke. Richard Windsor, an analyst at Radio Free Mobile, stated that Nvidia has zero presence in China and will remain that way, and added that DeepSeek has almost no chances of selling its chips outside of China unless it gains access to cutting-edge manufacturing technologies.