Hardware & InferenceBusiness & Market 🇨🇳 27.07.2026 07:04

Major AI Companies Massively Develop Their Own Chips

DeepSeekDeepSeek Moonshot AIMoonshot AI ByteDanceByteDance Alibaba/QwenAlibaba/Qwen
Leading large language model developers such as DeepSeek, Moonshot AI (Kimi), ByteDance (Doubao/豆包), and Alibaba (Qwen) have begun actively creating their own specialized microchips for AI training and inference. This allows them to reduce dependence on external suppliers like NVIDIA and lower costs.
According to a report on the OFweek portal, Chinese companies developing large language models (LLMs) are increasingly turning to creating their own artificial intelligence chips. This trend, dubbed "collective chip manufacturing" (造芯), is driven by the need to reduce reliance on foreign suppliers such as NVIDIA and optimize costs for model training and inference. Specifically, DeepSeek, Moonshot AI (developer of the Kimi model), ByteDance (creator of the Doubao model), and Alibaba (developer of Qwen) are among the companies investing in developing their own ASIC chips or specialized accelerators. It is noted that proprietary chips allow companies to better tailor hardware to their model architectures, improving efficiency and reducing power consumption. Experts believe this trend could reshape the semiconductor market landscape, intensifying competition with traditional chip manufacturers.
Source: DeepSeek (GNews) — original
Our earlier posts on this topic ↓
Fresh news