Alibaba’s Qwen AI Will Be Integrated Into Apple Phones In China Amid Push For Local Models

Apple is starting to take cost-cutting (and Chinese supply chains) very seriously.

Just days after reports that the smartphone giant will use China’s DRAM pioneer CXMT (which just priced its IPO) for local memory as a cheaper alternative source to ridiculously overpriced DRAM sourced from the memory cartel triad of Samsung, SK Hynix and Micron, this morning Reuters reported that BABA Qwen AI – much cheaper but just as efficient as most US frontier models – will be integrated into Apple Intelligence in China.

US-listed shares in of Alibaba rose 6% on Wednesday after the company confirmed to CNBC that the Qwen AI model will be integrated into Apple systems in China. 

“Qwen will be integrated into Apple Intelligence experiences within iOS, iPadOS, macOS, and visionOS for users in China,” an Alibaba spokesperson told CNBC. 

The Cyberspace Administration of China included Apple AI services on a list of approved providers, which included products from homegrown companies like Huawei.

The decision follows a long route to a Beijing greenlight for Apple’s AI service since the offering was first announced in 2024. In that time, the technological rivalry between the US and China has intensified as both countries race for dominance in AI.

The Apple-Qwen integration gives users the ability to access the model’s capabilities, “like text and image understanding and generation, without needing to jump between tools,” the Alibaba spokesperson added.

It comes after CNBC reported that Apple is in talks with a small Silicon Valley company that says it can shrink powerful artificial intelligence models enough to run directly on an iPhone, the startup’s CEO told CNBC on Tuesday. PrismML, a Khosla Ventures-backed spinout from the California Institute of Technology, publicly released compressed versions of Alibaba’s open-source Qwen model on Tuesday. The company said it reduced the model from roughly 54 GB to less than 4 GB, allowing all 27 billion of its parameters to run on an iPhone 15 or newer.

Meanwhile, in a stealthy push for local (i.e., “on your cell phone”) models, earlier this week Bloomberg reported that Apple’s planned M7 Ultra chip is being designed to support up to 1.5 TB of unified memory and to push AI performance toward the class of Nvidia’s Blackwell accelerators. Why? To give the company’s upcoming local LLMs access to as much DRAM as possible so the company is not confined to the cloud.

Apple’s M7 Ultra chip coming in 2029 is rumored to support 1.5TB of RAM.

This would make the processor much more capable for on-device AI. pic.twitter.com/7162RWZjpN

— AppleTrack (@appltrack) July 12, 2026

We previously looked at the Chinese LLM scene in two extended articles recently, the first one showing how rapidly China’s open models are catching up to the latest frontier offerings in the US (see “Are Chinese AI models a better value than US models“)…

… and the second one drilling down into each and every AI model in what we called the “definitive Chinese LLM primer.”

We also did an extended overview of the Alibaba offering which increasingly appears primed to take on the leading US frontier models; the schematic is summarized below.

Much more in the full reports (here and here).