China claims breakthrough in advanced chipmaking technologies (HBM, advanced packaging) that Western export controls were designed to restrict.
The bottleneck in advanced chipmaking has long sat behind a single Dutch gatekeeper: ASML. Now a team at the Chinese Academy of Sciences' Shanghai Institute of Optics and Fine Mechanics, led by Lin Nan, has published results for a laser-produced plasma EUV platform that achieves internationally competitive parameters. In a March 2025 paper, the team reported a peak conversion efficiency of 3.42%, surpassing earlier academic results from ETH Zurich (1.8% in 2021) and a Dutch research center (3.2% in 2019). Though commercial EUV light sources remain around 5.5% efficient, the progress narrows ASML's lead at a critical moment: the US has barred ASML from selling its most advanced EUV tools to China since 2019, and EUV lithography is essential for chip nodes below 7 nanometers.
Lin previously worked at ASML in the Netherlands as a competence owner for light sources used in metrology—a background that explains the close attention US officials give to such technical papers. The research team's approach uses a solid-state 1-micron laser rather than the CO2 lasers employed in commercial systems.
Power output remains the gap between prototype and production. China's EUV methods currently generate around 100 to 150 watts of light, versus 600 watts for ASML's current commercial benchmark. That difference is consequential: higher power enables higher wafer throughput, which determines fab profitability. Reuters reported in December 2025 that researchers in Shenzhen had assembled an operational prototype EUV lithography machine in early 2025 by reverse-engineering ASML technology, using salvaged parts from older tools and other export-restricted components. The machine was not yet producing functional chips, but the ambition was explicit—to enable China to eventually manufacture advanced chips on entirely domestically produced equipment.
On the software front, Alibaba previewed its Qwen3.8-Max-Preview on July 19, 2026 at the World AI Conference in Shanghai, describing it as a multimodal model with 2.4 trillion total parameters using a sparse Mixture-of-Experts design. Alibaba has not disclosed benchmark results, license terms, or the number of active parameters at inference. A day earlier, Alibaba's chip unit T-Head announced it would open-source SAIL, the software stack behind its Zhenwu chips, positioning it as an alternative to Nvidia's CUDA ecosystem. T-Head had unveiled the Zhenwu M890 in May 2026, claiming 0.6 PFLOPs FP16 performance with 144 GB of HBM3 memory, and reported 560,000 units shipped to more than 400 corporate customers by April 2026.