AIBRIEF
Apple is reportedly skipping M6 Pro, Max, and Ultra chips and designing the M7 Ultra for up to 1.5TB of unified memory, roughly double the M5 Ultra's target. The catch: memory chip supply, not chip design, decides whether it ships.
Apple's 2028 high-end Mac is being designed around up to 1.5 terabytes of unified memory, roughly double the previous generation's target, and that capacity doubling is the structural reason the next Mac Studio is likely to extend the $500 base-price hike Apple just pushed through.
The roadmap comes from reporting that Apple is skipping the M6 Pro, Max, and Ultra variants entirely, jumping from the base M6 (later 2026) to the M7 generation in 2027 and 2028 (Gizmodo, citing Bloomberg's July 12 newsletter; Bloomberg's June 25 article). The M7 design was finalized only about six months after M6, a faster turnaround attributed to AI workloads crowding out the traditional constraints of CPU speed, graphics, battery life, and thinness (Gizmodo, citing Bloomberg's July 12 newsletter; Bloomberg's June 25 article).
Unified memory is the load-bearing number: the pool shared by the Mac's CPU, GPU, and AI accelerators. The new claim is that the M7 Ultra will scale it to a level typically reserved for data-center GPUs from NVIDIA's Blackwell generation.
The 1.5TB configuration is not committed. The current memory-chip shortage is the binding constraint: if supply stays tight, the top configuration may not ship at all (MacRumors; Tom's Hardware).
The pricing baseline is already set: Apple raised the base Mac Studio by $500 to $2,500 last month (Bloomberg's June 25 article). A 2028 machine designed to double the prior memory target has a clear path above that number; whether the top configuration reaches buyers depends on chip supply, not Apple's roadmap.