Apple’s real memory problem isn’t cost, it’s supply

Apple continues work to get White House go-ahead to source memory from Chinese supplier CXMT, which the US government has placed restrictions on.

For Apple, the issue comes down to simple math. With the cost of making iPhones up 38% because of eye-watering memory price increases — up almost 7-fold since the beginning of 2025 — it makes sense to make pragmatic choices when it comes to sourcing supply, particularly when other computer companies (including HP and Acer) already obtain memory from CXMT.

US resistance to the plan is that because of the way Apple packages memory on chip, the use of that memory might partly contravene the technology transfer restrictions the US has in place. Note: use of off-the-shelf components is fine.

Apple has been lobbying to use memory from the supplier only on devices made and sold in China and would likely argue this is reasonable given that both HP and Acer already use CXMT memory in products sold outside the US. 

Why it’s really about supply

The iPhone maker’s dilemma isn’t just about memory price, it’s also about ensuring it has enough component supply to satisfy demand for its products. This is plausibly a bigger challenge for the company, which is already warning of constrained supply because of lack of available memory. Samsung, Micron, and SK Hynix have allegedly already sold through all their DRAM production for 2027, which only sharpens Apple’s case to secure alternate sources.

With that in mind, it matters that China consumes around 20% of all Apple hardware sold globally. All the company needs is the go-ahead to use RAM from CXMT in those Chinese-made, China-sold devices.

That alone would effectively grow its usable memory supply by the same amount, because the non-restricted memory it currently has to use in Chinese-market products could be freed up for devices sold elsewhere, with CXMT covering the China-sold units instead.

With demand for its products accelerating —even amid a broad industry downturn — Apple really wants to make sure it has enough of the component to meet demand. Those powerful, on-premises 1.5TB RAM-equipped M7 Ultra Mac Studio AI clusters won’t exist unless it’s possible to find memory to put inside them.

The timeline challenge

The proposed arrangement with CXMT might not be a quick fix. Recent reporting indicated the company has already reached its production capacity for 2026, though it is working to double capacity by 2028. Apple has already tested memory chips from the company across products including iPhones and MacBooks.

While Apple this year has applied steep price increases across all its products (except for iPhones), it is now expected to increase the cost of the current iPhone 17 models perhaps as soon as this week. 

Market reports already warn that Apple will raise prices on its soon-to-debut iPhone 18 Pro and iPhone Ultra devices next month, and we expect availability to be constrained, once again as a result of RAM shortages. Even if Apple gets the go-ahead to work with CXMT to close the gap, the positive impact of that arrangement is unlikely to kick in before next year.

Enjoy the pain

Elsewhere, big memory manufacturers, including SK Hynix, have announced plans to invest in new DRAM manufacturing capacity. But this will not be operational until 2029, at the earliest.

This suggests constrained product availability and higher prices for the coming months — and the only way Apple, or anyone else, will be able to limit the impact of those price increases across the entire electronics industry will be if they can obtain additional stocks of memory from vendors outside the big three. If they cannot, you can kiss the age of abundance goodbye.

You can follow me on social media! Join me on BlueSky,  LinkedIn, Mastodon and subscribe to The Core.ComputerworldRead More