Apple is said to be using TSMC’s SoIC-X process to stack the dies in its upcoming M5 chip, just as it has so far used in the Ryzen X3D. But the big, unknown question in the rumor mill is: What is Apple stacking on top of each other?
SoICs are a strong driver for TSMC, but they are expected to grow even more in the coming years. In the spring, the company explained that stacking solutions, such as those used in the Ryzen chip’s 3D cache on substrates, are also gaining momentum in this area. This will be taken into account accordingly when expanding capacity.
Ultimately, it’s no surprise that Apple will move to this technology at some point. The process has now been tried and tested for several years and is constantly being refined; new production stages should soon be stacked on top of each other much faster than they have in the past. Apple’s chips are also getting bigger and bigger, and stepping up to offer additional options would be a good option. Rumors now suggest that Apple is planning an M5 chip (Pro, Ultra, and Max have been mentioned as possibilities), which is intended for cloud and AI functions, on a core chip from the N2, which is essentially another CPU stack, and possibly a GPU. In turn, both would keep talking to each other.
Much remains unclear.
What will ultimately be stacked on top of each other and at what stage of production remains a vague rumor. If TSMC’s N2 technology is indeed the core, we shouldn’t expect it until late 2025. TSMC recently announced that the core chip technology and the chip stacked on it will continue to converge in the future, and so far there has been an average of one node between them. Some chips have also been stacked much later than the core technology, as it simply takes extra development time to get these features ready for mass production without any issues. However, the proposed N2 + N3 chip is likely to be a realistic 2026 product.
