Make us preferred on Google

DeepSeek develops its own advanced AI chip to minimise reliance on Nvidia

The transition marks a major update for DeepSeek that has primarily focused on building cutting-edge AI models

DeepSeek develops its own advanced AI chip to minimise reliance on Nvidia
DeepSeek develops its own advanced AI chip to minimise reliance on Nvidia

DeepSeek is reportedly developing its own cutting-edge artificial intelligence (AI)-powered chip, aiming to reduce its reliance on Nvidia and Huawei chips.

As per Reuters, the new chip is particularly designed for AI inference, the stage where a trained AI-powered model creates responses for users.

Unlike training chips, inference chips are generally more affordable and consume reduced power, making them perfect for growing AI applications.

A major strategic shift

The significant transition marks a major update for DeepSeek that has primarily focused on building cutting-edge AI models instead of developing hardware.

A report suggested the Chinese manufacturer has been preparing a chip for nearly a year and has quietly increased hiring of chip-design engineers while holding talks with memory suppliers, chip-design partners, and more.

Competition in China's AI chip market

If successful, DeepSeek may also join some famous Chinese tech firms, including Alibaba, Baidu, and Huawei, of which they are currently working to prepare their own AI chips to power its models.

Huawei has provided benefits from US export restrictions on the AI-chip manufacturing giant’s advanced chips, capturing half of China's projected $50 billion AI chip market.

However, competition between the tech giants continues to intensify as more companies invest in domestic chip development.

Impact of US export controls

DeepSeek’s efforts are also driven by the US restrictions that prevent Chinese companies from buying Nvidia’s most cutting-edge AI-powered chips.

Previously, Nvidia trained its highly famous R1 reasoning model using Nvidia's H800 chips before banning those exports.

DeepSeek has recently adapted its AI models to run on Huawei's Ascend processors.

Challenges ahead

Building a competitive AI-powered chip remains a challenging task.

Chip design requires years of development, large financial investments, and access to advanced manufacturing technology. US restrictions on cutting-edge chip production and high-bandwidth memory also pose major obstacles.

Despite these challenges, DeepSeek's reported chip project underscores China's growing push for AI self-reliance and enhanced control over the technology powering future AI systems.