Apple Intelligence in China: Qwen comes to Siri on November 4

Share
Apple Intelligence in China: Qwen comes to Siri on November 4

Two years after the rest of the world got it, Apple's AI has a launch date in the company's biggest market outside the United States — with Alibaba's model riding along, and the rebuilt Siri left at the border.

Apple will bring the first batch of Apple Intelligence features to China on November 4 as a developer beta, with the public release in December alongside iOS 27.3, and Alibaba's Qwen will be available inside Siri, the writing tools and Visual Intelligence. Apple announced it in a press release posted to its Chinese newsroom on Saturday morning Beijing time, ending a delay that has run since Apple Intelligence first shipped in the United States in late 2024 — long enough that Chinese iPhone buyers have watched Huawei, Xiaomi and Honor put their own assistants on the home screen first.

The first wave is the 2024 feature set, not the 2026 one. Apple's release lists live translation across Messages, FaceTime and Phone (and through AirPods), the Writing Tools rewrite-proofread-summarize trio, Visual Intelligence, Clean Up in Photos, Genmoji, natural-language photo search, memory movies, a Priority Mail section, email summaries, Smart Reply, notification summaries, priority notifications, a Reduce Interruptions focus mode, audio transcription summaries in Notes and call transcription summaries, plus Type to Siri.

Device support is narrower than the global list: iPhone 16 and later, iPhone Duo, iPhone 15 Pro, iPhone 15 Pro Max and iPhone Air, with Siri and the device language set to English or Simplified Chinese. A plain iPhone 15 or an iPhone 14 does not qualify. That is the same hardware cut-off Apple applied everywhere else, but it lands harder here, where the older models are still a large share of the installed base and where the phrase "AI phone" has been a retail talking point for two years.

Then there is the sentence Apple put near the bottom: this update does not include Siri AI or the other next-generation Apple Intelligence features announced at WWDC 2026. So China gets the assistant the rest of the world has had since 2024 — the one that understands messy speech and keeps context, but not the conversational Siri rebuild that was supposed to define the iPhone 18. The Verge reported the same gap, noting that the launch "doesn't bring the new Siri AI to China."

The Qwen part is already being read as "Qwen powers Apple Intelligence in China," which is not quite what Apple announced. The system features are driven by generative models Apple built itself; Alibaba's model is a partner integration in specific experiences, the same shape as the ChatGPT hookup that arrived with iOS 18. Ask Siri something that needs broader knowledge, point Visual Intelligence at a scene, or hand it a photo, a PDF or a slide deck, and the request can be routed to Qwen's text and image understanding. Writing Tools also gets a "Compose with Qwen" option that drafts from a prompt inside any app.

It is opt-in and account-linked. Users sign in with a Qwen account or Sign in with Apple, and Siri asks for permission before a question, document or photo is sent. Alibaba's own announcement, as carried by 智东西 Zhidx, says requests and responses are not used to train its models — the privacy promise that made the ChatGPT integration palatable in 2024, now localized.

The regulatory path explains the shape of the deal. China's cyberspace regulator registered Apple's own generative model in July, Reuters reported in August that Apple had trained a model specifically for China, and Tim Cook confirmed in Shanghai on October 15 that Apple Intelligence and the Alibaba model had been approved. Apple's release says support for iPad, Mac and the rest will follow "after the necessary approvals" — so the phone ships first, and the multi-device story is still waiting on paperwork.

Which is the part worth keeping: this is a template, not just a launch. A Western AI product entering China now keeps its own model in the core, registers it domestically, and bolts on a Chinese partner as an opt-in extra for the queries a foreign model should not answer. If that arrangement survives its first year, it is the answer every US lab will eventually copy — and it is a more interesting precedent than the feature list, which Chinese reviewers have already seen on rival phones.

The fair counter-argument is that this is a parity story, not a capability story. A Chinese buyer choosing between an iPhone 18 and a Huawei flagship is not short of AI features; what Apple sells is integration, on-device processing and a privacy posture, and none of those show up in a spec comparison. December also puts the launch after the holiday shopping peak, in a market where Apple's revenue has been under pressure and where "the iPhone has no AI here" has been a live objection at the counter. Fixing that objection is worth more than winning a benchmark — but it will not be visible in a launch-week demo.

We covered the state of the rebuilt assistant in September — Apple ships iOS 27. Its new Siri AI is still a beta — and that is exactly the piece China is not getting yet.


Nvidia is negotiating to guarantee billions of dollars of debt for Nscale's West Virginia AI campus, the site of Anthropic's $44.6 billion first building. The Information reported the talks on Thursday and Bloomberg corroborated them; under a residual value guarantee, Nvidia would step in if Nscale defaulted on construction payments, take temporary control of the facility and help find a buyer and a new tenant, per TechCrunch.

The project is the one Nscale is staking its public listing on. Anthropic's contract covers the first 460-megawatt building, which is meant to run Nvidia's next-generation Vera Rubin chips, and the two companies have discussed a second building worth roughly another $45 billion. Nscale is seeking an IPO at a valuation of up to $35 billion while running about 25,000 advanced chips in August against a plan for nearly 500,000, and Microsoft pulled out of a planned Nscale site in England this month after grid power slipped into the 2030s.

Nvidia's support for neoclouds has until now covered leases and chip purchases rather than construction, and the total is already large: roughly $170 billion in investments and credit support across its customers, including $1 billion in convertible financing committed to Nscale in September. Co-signing a builder's loans is a different trade. It converts Nvidia's revenue into balance-sheet risk on a building whose anchor tenant has been warned by Blackstone that it may receive a below-investment-grade rating after its own IPO — and it is the clearest sign yet that the buildout now needs the chip vendor's credit to clear, not just its silicon.

What to watch: whether the guarantee shows up in Nscale's IPO disclosures, and whether Apple names a firm December date for iOS 27.3 in China.

Would you trade the new Siri for a working assistant today, or wait for the rebuild? Write to us via our contact page.

Get SiliconNoon in your inbox

One email a day with the AI stories that matter — what happened, the numbers, and why it matters. No spam, unsubscribe in one click.

Read more

Vibe-coded game ports put Halo and GTA in the browser

Vibe-coded game ports put Halo and GTA in the browser

Classic console games are collapsing into the web at a pace nobody asked for: in the past two weeks, hundreds of AI-decompiled ports of PS2- and PS3-era titles have appeared online, and several of them play like the real thing. Hundreds of AI-decompiled games — Halo: CE, GTA: Vice City, Call of Duty: Black Ops, Skate 3, The Simpsons: Hit and Run — now run in a browser tab, and the ones we've seen reports of don't look like bootleg garbage.

Goncourt drops the prize novel accused of being AI-written

Goncourt drops the prize novel accused of being AI-written

France's literary season just produced its first casualty of AI, and nobody involved can actually prove authorship either way. The Académie Goncourt withdrew Thélyson Orélien's novel C'était ça ou mourir from its selection on Friday 25 September, four days after the book won the Prix du Roman Fnac. The academy said the decision was driven by "the plagiarism of which its author has been guilty over time" and by "the convergent result of several analyses by credible researchers and journalists"…

China's AI video boom: 221,900 shows, 1,055 hits

China's AI video boom: 221,900 shows, 1,055 hits

Chinese cities are paying studios to make AI video at industrial scale, and the audience has not kept up. Also: a Chinese startup opens a crowdsourced robot-training market, and the FT prices the neolab wave. Chinese local governments are competing to subsidise AI video production, and the output has already outrun the audience: 221,900 new AI shows launched on Douyin in the first half of 2026, and only 1,055 of them passed 100 million views, the benchmark the industry uses for a hit.

Meta's hologram avatars put a photoreal you on video calls

Meta's hologram avatars put a photoreal you on video calls

Meta spent Connect this week selling hardware. The announcement that will outlast the hardware is the face it puts on your calls: at Wednesday's keynote the company said its holographic avatars — the productized version of the Codec Avatars research it has demoed since the Orion prototype — are coming to the Meta VR Glasses and to the Ray-Ban Display frames. The avatars are photoreal, they run on a model Meta calls Muse Realtime Avatar, and the first place you will meet one is a WhatsApp call.

XPRIZE's wildfire AI found fires in 10 minutes — couldn't stop them

XPRIZE's wildfire AI found fires in 10 minutes — couldn't stop them

The $11 million XPRIZE Wildfire handed out its awards on Wednesday, and the honest headline is the one the organizers didn't write: the detection half worked, the firefighting half didn't. Meanwhile Black Forest Labs quietly put a robot model at the top of Nvidia's leaderboard. XPRIZE awarded $4.55 million in prizes at a ceremony in New York on Wednesday, and both $3.5 million grand prizes went unclaimed because no team could finish the job it was actually set. The competition ran two tracks.