Overview
- The National Development and Reform Commission said on May 22 that Chinese large AI models should increase compatibility with domestically produced compute chips and accelerate construction of embodied‑intelligence training infrastructure to keep AI development secure and controllable.
- AMD announced an investment of more than US$10 billion in Taiwan to expand advanced packaging and validate 2.5D bridge interconnects, a move AMD says will speed deployment of higher‑performance, energy‑efficient AI systems.
- Lenovo’s CEO and AMD’s CEO warned of a structural shift from training to inference workloads and tight CPU supply, with Lenovo forecasting roughly 70% of GPU server use moving to inference and AMD forecasting over 35% annual CPU market growth for the next five years.
- Commercial product news continues in parallel: Google showed new Android XR smart glasses using Qualcomm’s Snapdragon XR platform and fashion partners, and Xiaomi rolled out the YU7 family and reiterated strong early lock‑order and delivery figures.
- Industry analysts and reporting suggest these policy and investment moves could spur faster regional chip‑model alignment, greater vendor and cloud partnerships for alternative server chips, and practical impacts for customers facing longer lead times and higher costs for AI compute.