China kept Apple Intelligence waiting for almost two years. This week the country's internet regulator finally said yes — but on its own terms. The Cyberspace Administration of China registered the service for local release, with one catch: instead of OpenAI, the system runs on Alibaba's Qwen model for text and image generation.
Apple worked through nearly every major Chinese AI player before landing on Alibaba. The company held talks with Baidu, tested DeepSeek, and looked at ByteDance, but each deal stalled over the technical work of adapting outside models to Apple's requirements. Qwen ended up as the primary partner, and according to people familiar with the matter, Baidu will still play a smaller supporting role.
A local model isn't optional in China — it's the price of entry. Regulators require that generative AI aimed at the mass market run on domestic infrastructure and answer to domestic data laws. That means the Chinese version of Apple Intelligence will work differently under the hood than the one sold everywhere else.
Apple hasn't given a launch date. Going by the company's usual pattern, approval to release typically takes a few months, so Chinese users will likely see the features land alongside the fall updates to iOS, iPadOS, macOS, and visionOS.
Markets reacted immediately: Alibaba's US-listed shares jumped 4 to 6 percent on the news. The stakes are high for Apple too. iPhone shipments in China rose 24.4 percent year over year in the second quarter, and without on-device AI, Apple risked losing ground to Huawei and Xiaomi, both of which built AI assistants into their phones years ago. Whether the Chinese version keeps pace with the global one is still an open question — Apple's talks with Baidu and DeepSeek never made it past the pilot stage.
