
Shanghai Enflame Technology is nearing a STAR Market listing, bringing another Chinese AI chipmaker closer to public markets while losses, Tencent concentration, and a small share in Nvidia-led accelerators remain unresolved.
Huawei is preparing to launch commercial HarmonyOS-powered desktop PCs in September, marking another step in its effort to extend its self-developed operating system from smartphones and tablets into enterprise computing.
ByteDance is in talks to buy AI chips from Shanghai-based Iluvatar CoreX and is also considering using Baidu's Kunlunxin chips, as the TikTok parent expands its domestic chip options amid rising inference demand from its AI chatbot Doubao.
TSMC is accelerating its advanced packaging roadmap for AI chips, expanding CoWoS capacity while publicly disclosing new progress in glass substrate technology. The company is also signaling that next-generation packaging competition is shifting from CoWoS toward CoPoS as it builds out a fuller ecosystem ahead of rivals.
The semiconductor supply chain is facing another raw material shock — this time from tungsten hexafluoride, or WF6, a specialty gas used in chip manufacturing. Planned production adjustments or exits by some Japanese suppliers in the second half of 2026 have intensified concerns over tighter global supply, sending prices sharply higher and raising the risk of disruption into 2027.
Japan's Rapidus has signed a memorandum of understanding with the UK Semiconductor Centre, a government-backed body established in 2025 to support Britain's semiconductor ecosystem, marking a step toward cooperation on future semiconductor manufacturing and potential customer development in the UK, according to Rapidus and reports from Nikkei, Bloomberg, and Reuters.
Physical AI is emerging as a new frontier of model development. Any model, however, is only as good as the data used to train it. Because of this, Japanese startup APTO is launching a physical AI infrastructure lab to help plug the data gap needed to create vision-language-action (VLA) models, with a focus on imitation learning.
