TECH NEWS – According to an Apple employee responsible for platform architecture, this has enabled the Cupertino-based tech giant to achieve unparalleled performance.

Apple first implemented its new Fusion Architecture with the release of the M5 Pro and M5 Max. This enabled the two system-on-a-chip (SoC) designs to feature a more advanced chiplet design than the monolithic architecture used in previous Apple Silicon chips. The new chipsets generally feature a 2.5D design, in which individual blocks are located in separate areas of the chip. However, an Apple employee responsible for platform architecture mentioned in a recent interview that the M5 Pro and M5 Max boast vertically stacked chips. This change comes as a big surprise, as it offers numerous advantages.

Anand Shimpi, former head of AnandTech and current Apple employee, works in the company’s hardware technology division. He spoke with the German publication Heise Online to share further details about the M5 Pro and M5 Max. He noted that one of the biggest changes in the latest Apple Silicon family is the introduction of stacked chips. Thanks to experience gained with the UltraFusion architecture of the M2 Ultra and M3 Ultra, the new Fusion architecture has been successfully applied to the M5 Pro and M5 Max. Stacked chips typically mean that individual blocks are stacked on top of one another, creating a high-bandwidth, low-latency interface that is also extremely energy-efficient, although we haven’t yet seen the chip’s architecture to confirm this.

Essentially, this is just another version of the same concept. With previous Ultra chips, Apple combined two identical system on a chip (SoC) units to create a larger SoC. This time, however, they have distributed numerous functions across two different chips. These chips are not mirror images of each other; separate IP blocks have been integrated into both. The stacked chips mean that the M5 Pro and M5 Max mimic a design similar to 3D packaging, in which different components are stacked on top of each other like a sandwich for both the CPU and GPU blocks. This enables significantly faster communication between the components. Depending on which components are stacked vertically, however, this performance gain comes at a cost: the combined heat generation of the stacked components causes temperatures to rise.

He’s in competitive analysis and optimisation.

— 𝐷𝑟. 𝐼𝑎𝑛 𝐶𝑢𝑡𝑟𝑒𝑠𝑠 (@IanCutress) March 21, 2026

In a previous multi-core stress test, though, the M5 Max ultimately achieved lower temperatures than the M4 Max. Interestingly, while Heise Online reports that Shimpi is an Apple employee responsible for platform architecture, Ian Cutress tweets that he works in performance analysis and optimization. In short, the site may be mistaken in its comment regarding stacked chips.

Source: WCCFTech, Heise