Nikkei News quoted sources as sayingApple is in talks with Foxconn to upgrade the cloud server behind the "Apple Intelligence" service next year, and it is expected to be replaced byNew M4 processor architectureconfiguration, thereby corresponding to higher computing performance.
Currently, the cloud servers behind Apple's "Apple Intelligence" service use the M2 Ultra processor specification, so some believe that Apple should upgrade the servers to the yet-to-be-announced "M4 Ultra".
The main operating mode of the "Apple Intelligence" service is still to emphasize the operation of the device end, but the part that exceeds the computing power of the device end will be transferred to the cloud collaborative computing mode, and this process will be based on thePrivate Cloud Compute cloud computing environment architectureThe execution runs through the Apple Silicon processor hardware environment, and the calculation results are executed using a larger natural language model. The execution results are finally transmitted back to the user device through a secure network connection for presentation.
In addition, Apple also explained that the overall operating structure of "Apple Intelligence" will first focus on device-side computing, and will only seek cloud-based collaborative computing when computing power is insufficient. The content calculated through Apple's own independent cloud servers will not be backed up or permanently retained by user-uploaded data. Apple will not use this data for subsequent training or know the details of the data content.
Currently, the "Apple Intelligence" service is only available in the United States. It is expected to expand to more countries and regions by the end of the year, and plans to add support for more languages and open it to more countries and regions in 2025. Therefore, it will be necessary to strengthen the server computing architecture behind the "Apple Intelligence" service to meet the huge demand for usage.
