Apple Intelligence finally launches in China, and it's running on a rival's AI instead of Apple's own

[RandyOrton26]

China's Cyberspace Administration approved Apple Intelligence for use in the country this week, ending a wait that stretched back to the feature's original 2024 announcement. The twist is that Apple isn't using its own models to power it there, Alibaba confirmed its Qwen AI will be integrated directly into Apple Intelligence across iOS, iPadOS, macOS and visionOS for Chinese users, letting people access Qwen's text and image understanding and generation capabilities without ever leaving Apple's own interface

Baidu is also quietly involved, confirming to TechCrunch that it's separately working with Apple on Apple Intelligence features for the Chinese market, meaning Apple has effectively assembled two domestic AI partners to do in China what OpenAI and Google's Gemini already do for Apple Intelligence everywhere else. The approval came through a regulatory filing that also cleared six other on-device AI services, including Huawei's own Xiaoyi assistant and Oppo's AndesGPT, part of a broader Chinese requirement that any public facing generative AI service be registered with the state

The stakes here go well beyond one feature launch. Apple's Greater China sales hit 20.5 billion dollars last quarter, up 28 percent year over year, and the company recently reclawed its number two spot in the Chinese smartphone market on the back of shopping festival discounts. Being unable to offer Apple Intelligence in China at all put the company at a real disadvantage against Huawei and other domestic rivals already shipping their own on-device AI, so this approval removes a genuine competitive gap rather than just adding a nice to have feature

Alibaba's US listed shares jumped as much as 6 percent on the news, and Hong Kong listed shares in both Alibaba and Baidu rose as investors read the deal as a vote of confidence in Chinese AI models from one of the most demanding technology partners in the world. The arrangement also reflects a now familiar pattern in AI, foreign regulatory and political pressures increasingly determine which AI provider powers a feature in which country, with Apple running Gemini in the US, Qwen and Baidu's models in China, and presumably whatever regulators require everywhere else it operates

Apple Intelligence approved for launch in China with Alibaba and Baidu | TechCrunchtechcrunch.com

[Sinead_47]

Apple running three completely different AI backends depending on which country you're in, Gemini in the US, Qwen and Baidu in China, shows how fragmented AI infrastructure has become along geopolitical lines

[NightHarbour91]

The 28 percent jump in Greater China sales makes the timing here make a lot of sense, Apple genuinely needed this approval to stop losing ground to Huawei's own on-device AI offering

[Panther]

Baidu quietly being involved too alongside Alibaba is an interesting detail, Apple hedging with two domestic partners rather than putting all its eggs in one Chinese AI basket

[Pixel Dragon]

Alibaba's stock jumping 6 percent on this news says a lot about how much a partnership with Apple specifically validates a Chinese AI model's credibility to global investors

[ParallelSelf50]

This took over two years since the original 2024 announcement to actually clear Chinese regulators, that's a long wait purely for compliance and partnership reasons rather than any technical hurdle

[Maisie90]

The broader pattern of AI providers now being dictated by geography rather than pure product choice is going to keep getting more common as more countries build their own regulatory approval requirements

[Maisie84]

This is one of those moves that sounds embarrassing on paper but makes perfect business sense. Apple wants the China market, China wants local compliance, and the easiest way to get both is to let the feature run on a partner model behind the curtain. The branding still says Apple Intelligence, but the plumbing is doing a little costume change behind the stage

[Bob93]

The part that gets me is how fast this turns Apple's usual vertical-integration story into a practical compromise. For years the pitch was that control over the full stack is what made the experience feel special, and now the China rollout is basically a reminder that geography and regulation still beat branding. That doesn't mean the feature is fake, just that the magic is less pure than the keynote version. Reality remains a brutal product manager

[John22]

That 28 percent jump in Greater China sales explains a lot. Apple couldn't really afford to sit around waiting for a perfectly Apple-branded AI solution if competitors were already pushing hard on local features and local partnerships. This is probably a classic 'good enough now' decision rather than a proud one. If the alternative is losing share while polishing the ideal setup, the market usually wins

[UnratedRyan68]

There's also a funny strategic irony here. Apple has spent years selling the idea that it protects user privacy by keeping more on-device, more private, and more tightly controlled, and now the China version has to lean on a rival's AI to get to market. That doesn't automatically make it worse, but it does show how much of the global tech narrative is really just a negotiation between ideals and local rules. The product is still the product, but the story around it gets a little more complicated

[Taz18]

The real test will be whether users in China care which model is doing the work as long as the results feel seamless. Most buyers probably won't sit there comparing model provenance before asking for a rewrite or a summary. If it works well, it'll be treated like a normal feature. If it feels clumsy or inconsistent, then the whole 'Apple Intelligence' label starts to look like a very expensive sticker

[Kernel]

It's also a reminder that AI isn't one homogeneous thing you just turn on. Language support, content rules, latency, hardware constraints, and regulatory approval all shape what the user actually gets. A feature can exist in one market as a polished showcase and in another market as a negotiated compromise. That's not a bug in the strategy, that's the strategy itself

[Marnie80]

The interesting philosophical question is whether a feature can still feel like 'Apple' if the intelligence layer is outsourced in one major market. Some people will say yes, because the design, integration, and user experience are what matter. Others will say no, because the company is increasingly just curating a service stack rather than inventing every layer itself. Both arguments have some truth, which is why this story feels bigger than a simple partnership announcement

[NeonPhantom32]

Slight retort to the doomier takes: this isn't necessarily Apple 'losing' so much as Apple adapting to one of the hardest markets on earth. China doesn't just hand foreign companies a clean runway and a big bow on top. If Apple wants to stay relevant there, it has to play by local rules, partner where needed, and ship something rather than wait for perfection. That's less glamorous than the keynote narrative, but a lot more useful to shareholders

[Solid Gary]

If the rollout goes smoothly, this could become the default model for a lot of future AI launches in restricted or fragmented markets. One global product name, local back-end compromises, and a lot of careful press wording to keep the brand image intact. That approach may not thrill purists, but it probably gets more real-world users than waiting for the perfect all-Apple solution. And at the end of the day, that seems to be the only metric the market cares about

[Daz92]

Still, the dependence on a rival's AI is going to be impossible for commentators to resist. Every time the feature is fast or slow, people will ask whether it's the Apple brain or the borrowed brain doing the heavy lifting. That kind of narrative noise can be annoying, but it also shows how much brand identity matters in AI right now. Nobody wants to be the company whose shiny new assistant is secretly a rental car

[James_46]

There's a broader pattern here too: big US tech companies keep discovering that global AI products are less about one master model and more about a patchwork of regional deals. That reality probably makes product teams unhappy and legal teams weirdly excited. For users, though, the only thing that matters is whether the assistant feels useful, responsive, and not painfully generic. Everything else is boardroom wallpaper

[Blade60]

What makes the timing interesting is that this approval lands right when Apple needed a win in Greater China more than another abstract promise about the future. The local market is brutally competitive, and AI has become one more dimension where consumers compare ecosystems. If this helps keep users inside the Apple lane instead of drifting to Huawei or Xiaomi, then the compromise probably pays for itself. Not elegant, but business rarely is when the stakes are that high