Apple may skip M6 Pro, Max and Ultra chips as memory crisis reshapes Mac roadmap: Report

Apple's response to the ongoing global memory shortage may extend well beyond higher product prices. After recently increasing the cost of several Macs, iPads and smart home devices, the company is now reportedly preparing to alter its Apple Silicon roadmap, delaying some high-end processors while prioritising future chips designed for on-device artificial intelligence.

According to Bloomberg, Apple still intends to introduce the base M6 chip later this year, but the company is expected to bypass the usual Pro, Max and Ultra versions of the processor. Instead, those premium chips are likely to arrive as part of the M7 generation, marking one of the biggest shifts to Apple's in-house silicon strategy since it abandoned Intel processors.

More from Tech

The reported roadmap change comes as Apple continues to deal with an unprecedented increase in memory costs, fuelled by soaring demand from AI infrastructure and data centres.

Apple's chip roadmap could take an unexpected turn

Bloomberg reports that Apple is testing the entry-level M6 processor for a refreshed MacBook Pro expected later this year. The chip is said to introduce an updated memory architecture, a redesigned Neural Engine and improved graphics performance aimed at accelerating AI workloads.

The M6 is also expected to offer memory bandwidth of up to 200GB per second, a notable increase over the M5's reported 153GB per second. Apple is reportedly testing versions with up to 12 GPU cores, giving the chip additional graphics performance for demanding creative applications and AI processing.

However, unlike previous Apple Silicon generations, the M6 may not receive Pro, Max or Ultra variants.

Instead, Bloomberg says Apple plans to move directly to the M7 family for its high-performance Macs. The base M7 processor could arrive as early as the first half of next year, followed by the M7 Pro and M7 Max before the end of the year. The M7 Ultra is reportedly scheduled for 2028.

The next-generation chips are expected to place an even greater emphasis on AI, with the base M7 reportedly supporting memory bandwidth of up to 240GB per second to better handle on-device machine learning tasks.

Memory shortages continue to shape Apple's premium Macs

The reported strategy shift comes against the backdrop of an industry-wide memory crunch that has already begun affecting Apple's existing hardware.

Earlier this week, the company increased prices across several Mac, iPad and home entertainment products, attributing the decision to a dramatic rise in memory and storage costs. Apple said the rapid expansion of AI data centres has created unprecedented demand for high-performance memory, forcing it to pass some of those higher costs on to consumers.

The same pressures have also altered the configuration of Apple's flagship desktop computer.

When the current Mac Studio launched with the M3 Ultra chip, customers could configure it with up to 512GB of unified memory. Apple has since quietly removed several high-memory options, leaving 96GB as the only readily available configuration for buyers selecting the M3 Ultra model. At the same time, the starting price of the M3 Ultra Mac Studio has climbed from US$3,999 to US$5,299.

Despite the reported overhaul of the M6 roadmap, Apple has not abandoned its highest-end silicon altogether.

According to Bloomberg, the company is still working on the M5 Ultra, which is expected to debut later this year in an updated Mac Studio. The processor could offer up to 36 CPU cores while Apple has reportedly tested support for as much as 768GB of unified memory. Whether those top-end configurations ultimately reach customers may depend on when the memory market stabilises.

Taken together, the reported roadmap changes suggest Apple is adapting its silicon strategy to cope with an industry undergoing profound change. Rather than following its established release cadence, the company appears to be balancing AI ambitions with the realities of soaring component costs and constrained memory supplies.