DeepSeek Looks To Sidestep Huawei As It Begins In-House AI Chip Journey While Dodging Development Complications, Trade Sanctions, Delays & More

Omar Sohail
DeepSeek is developing its own AI inference chip
The move mimics what similar companies are doing, but execution is a Herculean task / Image made using Gemini

AI chips from Huawei and NVIDIA are currently being leveraged to train and run DeepSeek’s popular models, but the days of this arrangement could be numbered as the Chinese startup has plans in motion to launch its in-house inference silicon. According to people familiar with the firm’s plans, the chip won’t be utilized for training new models, but for generating responses for users on a pre-trained one.

Of course, the journey will be arduous because not only does it take years to launch a viable solution, but DeepSeek also has to contend with being limited to existing chip manufacturing equipment due to U.S. sanctions, not to mention the costs of developing a custom solution from scratch. In short, success is far from guaranteed.

Related Story Intel To Feature 12 Xe3P iGPU Across Multiple Nova Lake Mobile & A Single Desktop CPU, Entry-Level Chips Based on 18A-P Node

Huawei has managed to carve out a massive market in China, with DeepSeek one of its customers, but the latter seeks greater control over the hardware

Chinese companies like Alibaba and Baidu are already weakening Huawei’s grip on the AI chip market as they have commenced the journey to design and mass producing their own in-house silicon. Reuters reports that DeepSeek aims to reduce its reliance on NVIDIA and Huawei, obtaining more control over its hardware.

While the plan is in its early stages, DeepSeek has already begun hiring talented chip design engineers in recent months for the same reason. However, the report mentions that to keep a low profile, these hirings were performed privately rather than posted on public platforms. As much as DeepSeek should be commended on carving its own path, its progress will be mired by a truckload of obstacles.

For instance, an in-house AI chip takes years and billions in incurred costs, particularly during the tape-out process. Most importantly, DeepSeek will be limited to relying on SMIC’s older and inefficient 7nm process for mass producing the unnamed AI chip, meaning that the latter will be technologically inferior to NVIDIA’s solutions that use TSMC’s latest and greatest manufacturing processes.

Sourcing Dutch firm ASML’s advanced EUV machines is out of the question for DeepSeek, as a Chinese entity would immediately qualify for the U.S. trade sanctions list. To tackle mounting costs, DeepSeek was previously reported to have raised $7 billion in a funding round that took its valuation to between $52-$59 billion, contrary to its earlier strategy of rejecting external investment.

News Source: Reuters

Omar Sohail Photo

About the author: Omar Sohail is a reporter and analyst for Wccftech's mobile section, specializing in the technology and business of the mobile industry. His expertise lies in the intricate hardware supply chain, covering developments in semiconductor manufacturing, chip lithography, and camera sensor technology.

Follow Wccftech on Google to get more of our news coverage in your feeds.

Button