TECH – The Chinese Academy of Sciences (CAS) introduced QiMeng—an AI‑driven automated chip design system—on June 10, 2025, marking a major strategic milestone in the nation’s ambition to accelerate semiconductor innovation and reduce dependence on foreign electronic design tools. QiMeng, named after the concept of “enlightenment,” employs large‑language model technology to autonomously engineer integrated circuits, completely bypassing the traditional reliance on human programmers.
Developed collaboratively by the State Key Laboratory of Processor and Intelligent Software Research Centre at CAS, alongside the University of the Chinese Academy of Sciences, QiMeng was detailed in a recent research paper and released as open‑source on GitHub. The modular architecture comprises three tiers: the Large Processor Chip Model (LPCM) layer, a central hardware/software design agent layer, and an upper-layer suite of chip design applications.
CAS researchers reported that QiMeng’s output rivals chip performance and efficiency traditionally achieved by teams of skilled engineers. They cited the autonomous design of a processor chip for self-driving vehicles in mere days—a task that typically takes human developers weeks. Industry sources further claim QiMeng automatically completed the front-end design of a 32‑bit RISC‑V CPU in just five hours, achieving Intel 486-level performance with over four million logic gates. Its successor iteration, QiMeng No. 2, reportedly rivals ARM Cortex‑A53, integrating approximately 17 million logic gates.
Read More: Meta Debuts V‑JEPA 2: AI “World Model” for Robotics
“QiMeng functions like an automated architect and builder for computer chips,” a lead developer commented, highlighting its transformative capability. The system’s layered design, spanning the domain‑specific LPCM foundation to intelligent design agents and application-level outputs, allows seamless end‑to‑end chip engineering.
QiMeng’s debut coincides with heightened U.S. restrictions on EDA software exports, which have disrupted China’s access to vital design tools from firms like Synopsys and Cadence. By daring to open-source this system, China signals a proactive strategy toward achieving semiconductor autonomy. While acknowledging hurdles, such as ensuring fabrication stability, yield control, thermal management, and long-term reliability, QiMeng’s creators believe targeted applications in edge computing, custom AI processors, and domain-specific chips will unlock immediate benefits.
This development arises amid broader industry momentum: in 2022, the U.S., Japan, and the Netherlands imposed export controls to slow China’s IC sector expansion. China responded by ramping up domestic innovation and investment in research infrastructure. If QiMeng can scale and integrate into lead-edge fabrication nodes—such as 3nm—its open-source status might enable a global AI-assisted design ecosystem beyond U.S. control
