TSMC reportedly has $1 billion of Apple chips awaiting delivery of DRAM — will this be what powers the iPhone 18 Pro?
A billion dollars of Apple's 2nm chips are stuck at TSMC because the memory has not arrived, with Apple's own packaging choice the reason why they cannot just be stockpiled and finished later.
TSMC reportedly possesses around $1 billion worth of finished A20 Pro wafers that have been halted in the production process due to a lack of DRAM to integrate with them. Apple has decided to use TSMC's Wafer-Level Multi-Chip Module (WMCM) process, which combines the processor and memory at the wafer level, preventing the option of storing finished dies and adding memory later.
This design choice means these chips cannot be stockpiled for later use. Analyst Tim Culpan stated that the production of the A20 Pro on TSMC's N2 node is progressing smoothly, with good output and yields. However, the issue lies in the fact that the wafers are accumulating, waiting for the DRAM to arrive before they can be packaged and shipped.
This delay could result in thin inventory and extended delivery estimates for Apple's upcoming devices, including the iPhone 18 Pro. The issue stems from Apple's decision to switch to WMCM from the previous Integrated Fan-Out (InFO) process, which allowed for easier integration of memory after the logic die was fabricated. By integrating processor and memory at the wafer level, Apple aimed to increase DRAM capacity and lower latency for on-device AI features.
However, this design change has inadvertently created a memory shortage, as the AI infrastructure buildout has taken up the manufacturing capacity that would otherwise serve the smartphone market. Apple remains confident in meeting initial demand, but the delay could lead to thinner retail inventory and longer delivery estimates.
The upcoming iPhone 18 Pro is expected to ship around 200 million units, so the thin inventory scenario could pose a significant challenge for Apple.
Written by urgent.news from TechRadar's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.