Chinese Firms Leave Nvidia for Local AI Suppliers, Survey Shows - Bloomberg.com

Bloomberg's latest survey reveals that Chinese tech companies are gradually reducing dependence on international AI suppliers,转向 supporting domestic AI chip providers. This trend reflects China's stra... Read original →

Chinese tech firms’ pivot to domestic AI chips is triggering a deep reshaping of the semiconductor stack. Upstream, homegrown GPU/NPU architectures are spiking demand for localized EDA tools and advanced packaging; downstream, LLM developers must re-optimize algorithms for non-CUDA ecosystems, accelerating co-design trends. Compliance pressures from U.S. export controls force 'de-Americanized' supply chains, yet sub-7nm manufacturing remains critically reliant on TSMC in Taiwan, China—exposing systemic vulnerabilities. NVIDIA may counter with China-specific SKUs, while AMD and Intel could deepen ties with SMIC to test regulatory limits. Over the next 12–24 months, domestic AI chips could capture over 30% of China’s market, but high-end training silicon still lags by 3–5 years. This shift isn’t just substitution—it’s the opening salvo in a global battle for AI infrastructure standards.