Ming-Chi Kuo refutes claims of massive Apple chip backlog amid DRAM shortage

Analyst Ming-Chi Kuo dismisses reports of a $1 billion pile-up of unfinished Apple processors at TSMC, citing disciplined supply chains and early planning over supply chain disruptions caused by memory constraints.

Ming-Chi Kuo has pushed back against claims that TSMC is holding roughly $1 billion of unfinished Apple processors because of a global DRAM shortage, arguing that the supply chain does not work that way. Writing on X, the veteran TF International Securities analyst said Apple does face genuine memory constraints this year, but he added that the company plans chip production at TSMC months ahead, based on the memory it expects to secure, rather than allowing large quantities of wafers to pile up unfinished.

The dispute follows reports that Apple silicon built on TSMC’s advanced 2-nanometre process is sitting in work-in-progress form because there is not enough LPDDR5X memory to complete packaging. TechRadar said the chips are likely intended for the iPhone 18 Pro and that Apple’s move to a newer wafer-level multi-chip module packaging method means memory must be available earlier in the process. That is a material change from previous packaging approaches, because the memory can no longer be added later without disrupting the flow.

Tom’s Hardware reported that the alleged shortage is tied to heavy demand from artificial intelligence infrastructure, which is tightening supply across the DRAM market. It said Apple is still expected to launch the iPhone 18 Pro on schedule, although customers may see thinner stock and longer waits. The publication also said Apple has been talking to suppliers including Micron, SK Hynix and Samsung, while exploring alternatives such as China’s CXMT, though that route carries political and regulatory complications.

Kuo’s argument is that the scale of the reported bottleneck is overstated. He said Apple and TSMC both operate with highly disciplined supply chains and would have little reason to produce and store large volumes of logic chips that cannot be finished. That view does not deny the memory shortage, but it does suggest the problem is more likely to affect allocation and delivery timing than to leave vast piles of semi-finished Apple silicon stranded inside TSMC factories.