Baidu Unveils New AI Chips to Strengthen China’s Push for Tech Self-Sufficiency
Baidu has taken a significant leap in China’s drive toward technological self-reliance by unveiling two new artificial intelligence chips at its annual developer and client conference. As competition intensifies between domestic innovators like Huawei and global giants constrained by export restrictions, Baidu is positioning itself as a central contributor to the country’s next generation of AI computing. The announcement highlights not only Baidu’s technical strengths but also China’s broader strategy to reduce dependence on foreign hardware for advanced machine learning and cloud applications.
Introducing the M100 and M300 chipsets
At the event, Baidu showcased the M100 and M300, two chips designed by its semiconductor subsidiary Kunlunxin Technology. Each targets a distinct but crucial part of AI development.
The M100 aims to dramatically improve inference efficiency for models built on the mixture-of-experts approach, a technique increasingly used to optimise large language models while reducing computational load. Baidu plans to release the M100 in early 2026, positioning it as a cost-effective, high-performance engine for running AI applications at scale.
The M300, scheduled to debut in 2027, is built for the much heavier task of training multimodal AI systems that handle text, images, video and audio simultaneously. With capacity to support models containing trillions of parameters, the chip is designed to power China’s most advanced research into artificial general intelligence and large frontier models.
Supporting China’s goal of independent AI computing power
Shen Dou, Baidu’s executive vice-president and president of its cloud division, emphasised that both chips were designed to deliver “powerful, low-cost and controllable” computing resources. In China’s tech discourse, “controllable” has become a key concept, reflecting the desire to avoid vulnerabilities caused by foreign technology restrictions. With access to high-end GPUs from companies such as Nvidia increasingly limited, the development of domestic alternatives has become a national priority.
Baidu’s chips are intended to work seamlessly across its cloud platforms, enabling Chinese enterprises and research institutions to build and deploy AI systems without depending on imported hardware. This aligns closely with government goals to achieve secure and stable supply chains in critical technologies.
A deepening rivalry within China’s AI hardware race
Huawei’s Ascend chips have dominated headlines in recent years, accelerating the company’s reputation as China’s leading AI hardware supplier. Baidu’s latest release signals that competition among domestic players is intensifying, which could accelerate innovation and broaden the range of computing solutions available within the country.
Unlike Huawei, which integrates its chips heavily into telecommunications and consumer devices, Baidu’s strength lies in cloud computing, AI services and foundational models. The M100 and M300 reflect this orientation, targeting the full lifecycle of AI development from training to deployment. Their arrival adds further momentum to China’s efforts to build a robust, diversified and globally competitive semiconductor ecosystem.
A stronger foundation for China’s AI future
The unveiling of Baidu’s new chips comes at a pivotal moment. China is rapidly expanding its AI infrastructure, from national computing power hubs to provincial cloud clusters. As demand for efficient and scalable AI accelerates, the availability of domestically produced, high-performance chips will be central to supporting everything from industrial automation to digital government services.
By developing processors tailored specifically for modern AI workloads, Baidu is contributing essential building blocks to the nation’s long-term technological ambition. The M100 and M300 represent more than hardware advances; they signal China’s increasing confidence in its ability to shape the future of AI on its own terms.
