In a timeline when Moonshot AI’s advancements with the Kimi-K3 model as a match for Anthropic’s Fable 5 and OpenAI’s GPT-5.6 Sol is defining artificial intelligence (AI) progress from China, another important development is the fact that Apple Intelligence has been given the green light by regulators in China. They aren’t the only ones though, which makes for a rather interesting plot. The contours of Apple Intelligence are becoming clear, and industry analysts believe this removes Apple’s biggest competitive disadvantage in the market.
Event Context
Chinese cyberspace regulators have approved seven on-device AI service applications. Alongside Apple, phone makers Xiaomi, Huawei, Oppo, Vivo, Samsung and ZTE (Nubia) also now have permission for on-device AI rollouts. The approvals by the Cyberspace Administration of China (CAC) mark a landmark regulatory clearance under China’s Interim Measures for the Management of Generative AI Services guidelines, which allows major global and domestic smartphone brands to finally deploy native AI capabilities directly onto devices sold in the country.
Player Focus
Foreign firms must comply with China’s strict data localisation rules by partnering with domestic tech giants. Apple Intelligence, Samsung’s Galaxy AI, Xiaomi PengPai AI / MiMo, Huawei Xiaoyi, Oppo AndesGPT, Vivo BlueOnDevice and Doubao on Nubia made in partnership with Bytedance, have now marked necessary localisation and often heavy compression of models for on-device and local compute. Rather than opening separate third-party apps, Chinese smartphone users will be able to access AI features such as text summarisation, writing assistance, custom emoji creation, and system-level smart assistants running completely locally.
Tarun Pathak, Research Director at Counterpoint Research points out there are “close to 95 million Apple Intelligence-capable iPhones in China, which have been waiting for AI services.”
A significant realignment for Apple
To gain regulatory approval from the Cyberspace Administration of China (CAC), Apple overhauled its standard AI architecture. The Chinese version of Apple Intelligence features a unique dual-partner structure, as it entirely swaps Western models such as OpenAI’s GPT, and shifts cloud infrastructure to domestic Chinese AI companies.
“That’s why China is the one market where Apple has had to fundamentally redesign its AI strategy, which required partnering with local players not just for models and search, but likely across the stack, including data storage and cloud infrastructure. If Apple is able to execute its AI in China, it will be a big challenge for Huawei, which does not have access to the advanced chips to support heavy AI workloads,” Pathak points out.
There are three server-based models as well—the AFM 3 Cloud which Apple calls a server-side workhorse, optimized for speed, efficiency, and performance, the AFM 3 Cloud Pro for demanding use cases like agentic tool use and complex reasoning, and the ADM 3 Cloud (Image) for image generation and editing, which also unlocks advanced photo-editing tools in the iPhone.
Also read: Apple gives developers new keys to its AI future
Team Analysis
Also read: Inside Apple’s AI architecture: Custom Gemini, sparse models and divergence
Alibaba’s Qwen LLM (also called Alibaba Tongyi Qianwen) will serve as the primary system engine. This means it will underline typical on-device and server side AI tasks including text processing, writing tools, summarisation, and generative image creations. It is not clear at this time whether Alibaba has provided Apple with a customised model or if it is the 27 billion parameter Qwen 3.6, which unlike Apple’s own 20 billion parameter sparse model, keeps all 27 billion parameters active.
Also read: Apple is suing OpenAI for allegedly stealing trade secrets
Baidu will handle a separate feature track focused on AI-powered search, advanced web lookups, and visual search recognition. However, this may not be the end of the array, with Apple believed to be considering more models including those from DeepSeek and Bytedance.
During a technical deep dive session immediately after the Worldwide Developer Conference (WWDC) keynote for 2026, Apple’s senior vice president of Software Engineering Craig Federighi detailed Apple’s AI structure is built atop the company’s own Apple Foundation models. This is the third generation for these models, with the first generation from 2024 and the second generation arriving last year.
There are two on-device models—the AFM 3 Core which has a 3-billion parameter dense model, and the AFM 3 Core Advanced which is a 20-billion parameter natively multimodal model that uses a sparse architecture to activate anywhere between 1-4 billion parameters at a time depending on the request type.
The two on-device models and the three server-side models in particular are where Gemini plays a role. “These models are specifically designed for Apple Intelligence experiences,” explains Federighi. Apple neither uses any of the Gemini models that Google deploys for its customers, nor does Apple use the infrastructure which is used by Google to deploy models for their customers. These are models that are custom made for Apple, by Google.
Pathak points out an interesting dynamic in China. “The approval brings Apple into the AI race in China, a market with arguably one of the world’s highest consumer AI awareness and preference. It’s also important to understand that China’s AI race is being fought on use cases, not model size. Competitors like Huawei are pursuing hybrid, full-stack AI strategies spanning hashtag#silicon, OS, and on-device models,” he says.
Match Outlook
“Strong consumer enthusiasm toward AI applications and China’s rapidly evolving AI ecosystem provides a highly favourable environment for Apple Intelligence, helping Apple sustain strong momentum and outperform the broader smartphone market in China during the second half of 2026,” says Linda Sui, Founder & Principal Analyst at Smart Analytics Global (SAG).
Sui says SAG expects Apple to introduce localised Apple Intelligence in China alongside the commercial rollout of the iPhone 18 series later this year. “Depending on software readiness and deployment schedules, selected Apple Intelligence features could become available as early as September 2026, likely together with the commercial release of iOS 27, or shortly thereafter through subsequent software updates,” she says.
The AI approval comes at a critical point in Apple’s product cycle in China, and trends from this market tend to have a bearing on the broader region too. “On one hand, the upcoming iPhone 18 Pro series is expected to face headwinds from higher pricing and relatively limited hardware upgrades compared with previous generations. On the other hand, Apple’s first foldable iPhone is expected to remain a niche product during its initial launch phase because of limited production capacity and premium pricing,” Sui explains.
The introduction of a localised and fully Apple Intelligence suite, could prove to be one of the stronger anchors defining upgrades for existing iPhone users in China. Meanwhile, Apple and the European Union are waiting to see who blinks first, with Apple restricting the suite in iOS 27 and iPadOS 27 in the EU markets, for now. The other question is whether more countries will now demand greater participation from local AI companies in the Apple Intelligence suite?
