In a move that underscores the growing tension between stock market speculation and the technical realities of industrial software, Gstarsoft, a prominent Chinese developer of Computer-Aided Design (CAD) solutions, has tempered expectations regarding its foray into the latest artificial intelligence frontiers. During a recent interaction on its investor relations platform, the company stated that it currently does not involve 'Physical AI' or the application of AI technologies within the simulation domain.
This clarification comes at a time when 'Physical AI'—a branch of intelligence that incorporates the laws of physics into machine learning models—has become the new North Star for global tech giants like NVIDIA. In the industrial software sector, the integration of AI into Computer-Aided Engineering (CAE) and simulation is viewed as a critical step toward automating complex design tasks, yet Gstarsoft's admission highlights that these advanced capabilities remain out of reach for many of China's domestic champions.
The discrepancy between investor appetite and corporate R&D highlights a broader trend within the Chinese tech ecosystem. As the central government pushes for 'autonomous and controllable' software to mitigate the risks of Western sanctions, retail and institutional investors often aggressively bid up companies with even tangential links to AI. Gstarsoft’s decision to manage expectations suggests a strategic focus on its core CAD business rather than chasing the high-risk, high-reward simulation market that requires deep expertise in multi-physics modeling.
While Gstarsoft remains a leader in China’s 2D and 3D CAD market, the simulation field represents a much higher barrier to entry. For Chinese industrial software to truly compete with Western incumbents like Ansys or Siemens, firms will eventually need to bridge the gap between static design tools and the dynamic, AI-enhanced simulation environments that Gstarsoft is currently avoiding. For now, the company’s stance serves as a sobering reminder that the transition from traditional software to 'AI-native' industrial tools will be a marathon, not a sprint.