TAIPEI (Taiwan News) — Apple has been working with Taiwan Semiconductor Manufacturing Co. (TSMC) to develop its own chip to run AI software on data center servers. 

The project, internally code-named Project ACDC (Apple Chips in Data Center), has been in development for several years and it is uncertain when, if ever, the chip will be introduced, the Wall Street Journal (WSJ) reported on Monday (May 6).  

Apple and TSMC have been working to design and start production of these AI chips, but sources told WSJ it’s unclear if their efforts have produced any concrete results. The server chips will likely run AI models, known as inference, instead of training AI models like Nvidia chips, sources told WSJ. 

Compared to its rivals, Apple has had the most experience and most success in designing its own chips, according to WSJ. It started using its own processing chips in iPhones and iPads in 2010 and AI-focused processors in its mobile silicon in 2017.  

In 2020, the company started replacing Intel chips in its Macs with its own custom silicon made by TSMC, the report noted. Advanced versions of Apple’s M-series chips can perform certain AI functions, including inference in servers, one source told WSJ.

Apple prefers most use of AI to occur on an iPhone or Apple Watch, but it still has to run some functions on remote servers accessed over the internet. This is where the new Apple server chip would come into play, according to WSJ. 

Meanwhile, Apple on Tuesday (May 7) also unveiled its new iPad Pro, equipped with the new M4 chip. The M4 chip is made using TSMC’s second-generation 3 nm process technology.