The Information website reported thatIn recent years, Apple has launched a project code-named "ACDC" (Apple Chips in Data Centers). It is currently considering using its own chips in data centers to build its own cloud services, directly providing server rentals to millions of iPhone and Mac developers, and competing head-on with current mainstream cloud service providers.
Since launching its first in-house chip in 2010, Apple has subsequently enhanced device computing performance and AI inference capabilities with its A- and M-series chips. Previous reports have indicated that Apple is already using its in-house chips in its internal data centers for cloud-based collaborative processing of services like Apple Photos and Music, while emphasizing the need to ensure user privacy and security.
According to reports, Apple executives believe that their own chips can significantly reduce costs in AI-intensive tasks. If cloud rental services are opened to the outside world in the future, developers will not only be able to obtain high-performance computing power, but the fees will also be far lower than the current quotes from Amazon, Microsoft and Google.
However, the main driving force behind this concept, former cloud chief Michael Abbott, left Apple in 2023, making the subsequent development of this plan unclear.
Despite this, Apple continues to test server-side applications using its own chips and develops open-source frameworks such as MLX to help developers run AI models on Mac chips. Startups such as Dragonfruit and WebAI have begun using Apple hardware as edge computing platforms to provide AI applications such as image analysis, demonstrating the potential of Apple chips in the cloud market.
Apple previously primarily purchased cloud resources from Amazon and Google, spending approximately $70 billion annually, but only a small portion of that was used for AI training. Apple hasn't made significant investments in NVIDIA GPUs, and while some AI inference has been performed using Google's TPU computing resources, it is now clearly focusing on improving AI inference efficiency with its own chips.
At present, it seems that Apple has adopted a market strategy different from traditional cloud services. It plans to promote cloud services through its existing developer relations team rather than establishing a large enterprise sales team. This means that its target market will focus on the developer community, not enterprise-level users or general small and medium-sized businesses, and it is not even for general consumers.
