Apple has signed a rare 3-5 year NAND contract, possibly without a price cap, as AI's impact on production capacity shifts the bargaining power in the storage market.

Apple has signed a rare 3-5 year NAND contract, possibly without a price cap, as AI's impact on production capacity shifts the bargaining power in the storage market.

As the world's most powerful consumer electronics buyers begin to abandon their efforts to drive down prices and instead sign long-term contracts that "may not have a price ceiling," the pricing power in the memory chip market is undergoing a fundamental shift.

On September 8, South Korea's Economic Tribune reported that Apple has negotiated a long-term NAND flash memory supply agreement with Japan's Kioxia, and its procurement logic is shifting from the traditional "price pressure + multiple sources" to "locking in quantity + long-term contracts".

This long-term contract is for 3 to 5 years and may not have a fixed price cap. Apple has historically relied on its massive purchasing power to gain an advantage in component negotiations, and this move is seen as a rare 180-degree turn in its procurement strategy.

Apple is facing fierce competition for capacity from data center customers who are willing to sign long-term contracts and pay high premiums. For Apple, securing NAND supply years in advance is becoming just as important as driving prices down to the lowest possible level.

Abandoning the obsession with price reduction, reversing the procurement logic

Apple has historically adopted a short-term procurement model that combines price pressure with multi-source competition with NAND suppliers.

This 3- to 5-year long-term contract with Kioxia, potentially without a price cap, completely breaks with convention. Long-term supply agreements are more common in the field of high-bandwidth AI memory, and Apple's move indicates a significant shift in the balance of power in the memory market.

The identity of the counterparty, purchase volume, and price of the agreement between Apple and Kioxia have not been disclosed, and Apple has not confirmed it. However, market observers believe Kioxia is the most likely supplier. Kioxia, along with SK Hynix, is reportedly one of the few major suppliers that have adopted this type of agreement.

AI's impact on production capacity is putting cost pressures directly on end-user pricing.

According to a TrendForce report, the memory cost of a 256GB iPhone 18 Pro could be nearly four times higher by the third quarter of 2026 than it was a year ago; the memory cost as a percentage of the bill of materials (BOM) is expected to have risen from about 10% in earlier 256GB Pro models to about 34%, and could exceed 40% in the first half of 2027.

According to supply chain reports, Apple's first foldable iPhone is expected to feature a large amount of high-quality memory, with a starting price of approximately $2,099 to $2,299, and a high-capacity configuration potentially exceeding $3,000. This further explains why Apple locks in NAND flash memory supplies years in advance, even if it means accepting higher component costs.

Bargaining power shifts; long-term contracts may become the industry's pricing anchor.

Apple's shift is not an isolated incident.

TrendForce indicates that Samsung Electronics, SK Hynix, and Micron are potential candidates for long-term contracts covering LPDDR5X and LPDDR6 memory used in upcoming iPhones. Samsung reportedly plans to allocate 60% to 70% of its total production capacity to long-term contracts.

The NAND supply is more dispersed, with SK Hynix expected to account for about 30% of the NAND supply for foldable devices, Samsung for 15%, and the two South Korean companies accounting for a combined 45%. The remaining share is left to other suppliers, which is precisely where Kioxia and Apple can expand their cooperation.

For investors, there is an expectation gap: the market still prices NAND according to the traditional "silicon cycle" framework, believing that overcapacity and a correction will inevitably follow the price increase. However, the capacity squeeze driven by AI has long-term and structural characteristics, which may mean that the peak and duration of this memory cycle may exceed the market's traditional expectations.

Risk Warning and DisclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.