What's Behind Apple's AI Strategy in China?
Sign up for ARPU: Stay ahead of the curve on tech business trends.
Alibaba this week launched a new version of its flagship artificial intelligence model, Qwen3, with a key feature: compatibility with Apple’s own MLX architecture. The move, which enables Alibaba’s AI to run efficiently on everything from iPhones to MacBooks, is the clearest signal yet that Apple has found its partner to launch its Apple Intelligence platform in mainland China.
The partnership highlights a dramatic strategic pivot for the world’s most valuable company. After years of lagging rivals in the AI race, Apple is now being forced to rely on outsiders — including a Chinese tech giant — to power the very technology it hopes will define the future of its devices. The move underscores the immense pressure Apple faces not just from competitors, but from the complex realities of global technology regulation.
Why is Apple so far behind on AI?
Apple's slow start in AI has become a major concern for investors, with its stock sliding 20% in the run-up to its 2025 Worldwide Developers Conference, the worst performance ahead of the event since 2010. The company’s AI platform, Apple Intelligence, has had a rocky rollout since being announced in 2024. Its AI-powered news summaries were paused due to errors, and its much-hyped upgrade to the Siri digital assistant has been delayed until at least the spring of 2026 due to significant engineering problems.
To fill the capability gap, Apple has had to make uncomfortable concessions. More complex features within Apple Intelligence, like its Writing Tools, currently rely on OpenAI's ChatGPT. This dependence on a rival was a stark admission of its internal shortcomings. Even Apple's own executives have voiced concerns. Eddy Cue, Apple's chief of services, said in a court testimony that he believes AI devices could replace the iPhone within 10 years, calling AI a “new technology shift” that creates opportunities for new entrants.
How is Apple trying to catch up?
Facing immense pressure, Apple is repeating the strategy that made the iPhone a success in the first place: opening its doors to developers. At its June developer conference, Apple announced plans for a software development kit that will allow third-party developers to build features using its own on-device AI models. The bet is that by giving its vast developer community the tools, a “killer app” for AI will emerge on iOS before it does on competing platforms.
This shift away from its historically closed, “walled-garden” approach is also reflected in its partnerships. While it leans on OpenAI in the West, the deal with Alibaba signals a pragmatic approach to navigating China’s strict regulatory environment, where foreign AI models are heavily scrutinized. Apple reportedly considered partnerships with Baidu and ByteDance before settling on Alibaba, whose Qwen models have ranked competitively against global rivals.
What does the AI competition look like?
While Apple plays catch-up, its rivals are accelerating. Google has aggressively integrated its Gemini models into its core search product, creating an “AI Mode” to offer a more conversational experience. Meta has found surprising success with its Ray-Ban AI glasses, which have sold over 2 million pairs since 2023, proving a market exists for AI-integrated hardware.
The most direct threat, however, may come from one of Apple’s own. Legendary designer Jony Ive, who shaped the look of the iPhone, has teamed up with OpenAI CEO Sam Altman after OpenAI acquired his hardware startup, io, in a $6.5 billion deal. Their stated goal is to build a new family of “AI companions,” with plans to ship 100 million devices, representing a direct challenge to the smartphone’s dominance. While early AI gadgets like the Humane Ai Pin have flopped — Ive himself called them “very poor products” — the combination of OpenAI’s software prowess and Ive’s design genius is a threat Apple can’t ignore.
Why does Apple need a Chinese partner?
Operating in China presents a unique set of challenges that even Apple cannot solve on its own. The Chinese government maintains tight control over online information, and any AI model operating there must comply with strict censorship rules. As reported by Saritha Rai of Bloomberg, Chinese models like DeepSeek are built from the ground up to provide bland, official answers on sensitive topics like Taiwan or President Xi Jinping.
By partnering with Alibaba, Apple gains a powerful, locally-developed model that is already compliant with these regulations. Alibaba’s Qwen models, which use a sophisticated “mixture-of-experts” (MoE) architecture similar to those from Western labs, have consistently ranked near the top of global leaderboards. This allows Apple to offer a competitive AI experience in China without navigating the political minefield of training its own models to meet Beijing’s standards.
This need to partner with local champions is a growing trend for US tech giants. Microsoft, for instance, added DeepSeek’s models to its Azure cloud service after the Chinese startup’s efficient models caused a stir in the tech world. Nvidia CEO Jensen Huang has publicly acknowledged the formidable capabilities of Chinese competitors and lamented the loss of market share due to US export controls, which he argued only spurred on local innovation.
For a company that has long prided itself on controlling every aspect of its hardware and software, the partnerships with OpenAI and now Alibaba represent a profound strategic shift. It's a tacit acknowledgment that in the global race for AI supremacy, even a company as powerful as Apple can no longer afford to go it alone.
Reference Shelf:
Alibaba launches new Qwen3 AI models for Apple's MLX architecture (Reuters)
Apple to Open AI Models to Developers in Bid to Spur New Apps (Bloomberg)
Who Is the Mysterious Founder of China’s DeepSeek? (Bloomberg)