TREE NEWS update: Tencent President Martin Lau said advanced chip supply restrictions will delay the next version of the company’s AI model, as fewer high-end chips are available for training whether rented overseas or built domestically. Chief Strategy Officer James Mitchell said Tencent has not yet seen a negative impact on its cloud business, adding that domestic chips already run inference efficiently for models such as Hy4 flash, with token throughput on domestic silicon far higher than imagined a year ago.
Tencent says advanced chip curbs have not hurt cloud business
The split between training and inference is the real story here: restrictions bite on frontier model development while domestic silicon proves adequate for serving existing models at scale. That distinction matters for how the constraint is priced — a delayed model version is a competitive setback, but an intact cloud business suggests the revenue-generating layer is insulated. Whether domestic inference efficiency keeps pace as models grow more demanding, and how much the training delay compounds over successive model generations, are the open questions worth tracking.
Generated by AI for reference only.
Share on WeChat
Open WeChat → Scan → then tap "…" to send to a chat or Moments.
Tap "…" in the top-right corner to send to a chat or share to Moments.