Apple reorganizes CPU plan skipping M6 Pro and Max to accelerate M7 generations for on device AI and leadership changes under Johny Srouji
Apple is making a sweeping reorganization of its in house CPU generational plan in order to hasten the availability of on device AI functionality. Bloomberg has broken down the 1st steps of this "revolution" in a detailed report 2 indicating that the company will launch a more modest entry level M6 base processor for the lower end systems while completely omitting the intermediate Pro & Max alternatives for that generation. It will be the 1st time since transitioning to Apple Silicon that the consumer line will not be sold with all of the available chip variants for an active generation.
Rather than following Apple's classic rollout cycle, it appears they plan to leap straight to the M7 gens for their high end build options, skipping more powerful M6 variants used previously, so as to accelerate delivery of advanced computing and graphics hardware architectures needed for the huge computational overheads of on device AI and graphics rendering software.
The stock M6 chip tested in a refreshed entry level Mac Book Pro available under the label J804 is due to launch before the premium range. Codenamed Komodo or H18G the M6 is very much about increasing memory bandwidth in order to speed up more demanding tasks. Our M6 is targeted to provide about 200 giga bytes a second worth of memory bandwidth which is a huge jump over the 153 gigabytes a second of current M5 architecture.
Aside from just the bandwidth the M6 will include an improved memory structure as well as a whole new neural engine optimized for machine learning applications. Many of the processing cores within the silicon package have been improved in general performance and include hardware acceleration for both encoding and decoding video. The GPU has been completely redesigned with Apple testing configurations with a maximum of 12 GPU cores whereas the M5 was tested with up to 10 cores.
Once the base M6 is launched, Apple will ink the launching of the M7 family quite rapidly. The base M7 processor to be called Delos or H19G is expected to arrive as soon as early next year with an increased memory bandwidth of about 240 gigabytes per second. This will be followed by the high end Andros range that comprises the M7 Pro and M7 Max processors in Q4 2027.
The top level chip M7 Ultra is expected to launch in 2028. The Ultra chip which takes the architecture of 2 Max chips doubling overall system performance is given for only the top level desktop hardware such as Mac Studio. The whole M7 family has been built from ground up focusing on hardware accelerated artificial intelligence operations.
Prior to completely shifting over to the more recent chip designs Apple intends to launch 1 last chip for the current generation. The M5 Ultra codenamed Sotra D is expected to serve as a upgrade to the Mac Studio desktop. This chip should contain in the neighborhood of 36 central processing unit cores and 80 graphics cores making it a very capable consumer grade processor. Apple has reportedly tested these (desktop) units running with as much as 768 gigabytes of memory however, the current industry wide shortage of components may actually limit available configurations in the retail space.
These we monumental technological changes come amidst significant changes to Apple's ledership. Johny Srouji now directly oversees all hardware engineering work conducted in the Mac iPhone iPad and Apple Watch groups. This consolidated structure in led by Johny Srouji in the wake of John Srouji's tight relationship with CEO John Ternus. The company's increased responsibilities will have to address a severe component shortage while a surge in material prices squeezes profit margins and impacts roadmaps.
