TrendForce reported on September 7, 2026, citing South Korea's Economy Tribune, that Apple has signed a long-term supply agreement (LTA) for NAND flash memory. The contract partner has not been disclosed, but market sources named Kioxia as the leading candidate. Some observers have floated a three-to-five-year term with no price ceiling, though neither Apple nor Kioxia has confirmed the deal itself.
Still, the rationale behind such a move is visible in both companies' public disclosures. Apple has warned that supply constraints and rising costs for NAND and DRAM are intensifying, while Kioxia has stated its intention to cover roughly 50% of its 2028 shipment volume through long-term contracts. As AI servers absorb enormous quantities of NAND, Apple's procurement priorities may be shifting from "buying as cheaply as possible" to "securing the volume it needs."
The Kioxia Theory Is Still Not Confirmation of a Contract
What TrendForce's article directly confirms is narrow. Economy Tribune reported that Apple signed a contract, but the counterparty, volume, and pricing terms remain undisclosed. Market sources, the report says, consider Kioxia the frontrunner. No announcement has surfaced confirming a Kioxia deal, nor has any Apple filing named a partner company.
The pricing terms are similarly unconfirmed. The description of a three-to-five-year term with no price cap was presented in the article as "market speculation." Kioxia's Investor Day Q&A session, held in June, declined to disclose contract details citing confidentiality obligations to customers, stating only that it is carefully weighing price and duration.
Cross-referencing public filings and the report against six key contract elements reveals the following boundaries:
| Item | Apple/Kioxia public disclosures | Reported on September 7 |
|---|---|---|
| Contract existence | Neither company has confirmed | Economy Tribune reported a signed deal |
| Contract partner | Undisclosed | Market sources favor Kioxia |
| Subject | Apple disclosed NAND supply constraints | Reported as long-term NAND contract |
| Volume | Undisclosed | Undisclosed |
| Term | Kioxia declines to disclose specific terms | Market speculation of 3–5 years |
| Pricing formula | Kioxia says it is carefully weighing terms | Market speculation of no price ceiling |
Public disclosures from Apple and Kioxia confirm NAND supply constraints and a long-term contracting strategy, but they do not confirm the counterparty, volume, term, or price ceiling of any deal between the two companies. Non-disclosure is not proof that no contract exists—but it also provides no basis for treating a Kioxia order or specific pricing terms as settled fact.
Apple Itself Has Acknowledged a NAND and DRAM Supply Crunch
In its Form 10-Q for fiscal Q3 2026, filed with the U.S. Securities and Exchange Commission on July 31, Apple described its supply environment in unusually candid terms. The company stated that it is experiencing supply constraints and rising costs for advanced semiconductors, storage NAND, and working memory DRAM, driven by industry-wide supply-demand imbalances. It further stated that it expects these trends to intensify.
Apple's scale as a buyer does not eliminate shortages. The company noted that even components it can source from multiple suppliers remain exposed to industrywide shortages and commodity price volatility. It also flagged the risk that it may not be able to renew existing supply agreements on comparable terms, and that suppliers might prioritize producing general-purpose components over parts customized for Apple. Raising prices, it warned, may not fully offset rising costs and could dampen demand for its devices.
As of July, TrendForce estimated that the 2026 NAND market would fall short of demand by 4–5%. Servers account for more than 40% of total bit demand, while smartphones and laptops together make up roughly another 40%. When AI-oriented enterprise SSDs deliver higher margins to suppliers, NAND destined for consumer devices—produced using the same manufacturing capacity—tends to get deprioritized. Apple's reported long-term contract emerges against this backdrop of competition for allocation.
Kioxia's "50% by 2028" Target Started With AI Customers
At its Investor Day on June 2, Kioxia presented an outlook in which NAND market bit demand continues growing strongly through 2028. Under this plan, the company aims to cover approximately 50% of its own 2028 shipment volume through long-term contracts. Locking in long-term orders in advance helps mitigate the boom-and-bust cycles characteristic of NAND, where demand can collapse after capacity investments are made.
However, this 50% figure cannot be directly linked to any contract with Apple. The negotiating partners Kioxia officially cited are primarily data center and enterprise customers driving strong AI inference demand, with several companies reportedly seeking even longer contract terms. While Kioxia intends to maintain its existing smartphone and PC business as a foundation, it aims to raise the share of revenue from data center and enterprise customers to over 60% in the medium to long term.
Expanding production capacity also takes time. On August 27, Kioxia and Sandisk announced plans—contingent on government support—to invest roughly ¥5 trillion in Japan through 2032. This multi-year plan involves upgrading equipment, related infrastructure, and technology at the Yokkaichi and Kitakami plants. While this points to future supply capacity, it has no bearing on boosting NAND supply for this autumn's iPhone lineup.
For Kioxia, long-term contracts are not merely a tool for locking in prices and stabilizing revenue. The company has explained that it aims to retain flexibility to accommodate changing product specifications, while weighing price and duration to reduce investment risk from fiscal year 2028 onward. Buyers commit to volume; suppliers gain justification for capital investment. If the reported Apple deal is accurate, it would represent the extension of a contracting practice that has taken hold with AI companies into the smartphone sector.
Without a Price Ceiling, Apple Would Be Buying Volume, Not a Discount
If a contract without a price ceiling is indeed real, the primary benefit to Apple would not be a guarantee of low prices. Rather, it would be easier access to the agreed-upon volume even as supply-demand conditions tighten further. Because the supplier can pass rising market prices through to the contract price, it can accept a long-term volume commitment without sacrificing upside during price surges.
This condition stands in contrast to Apple's traditional image. The company has long been known for using its massive order volumes and multi-supplier sourcing to drive down component prices. But if sellers can redirect production toward AI servers instead, a large order volume becomes not just leverage for discounts but also a long-term burden the supplier must be willing to bear. The suggestion that Apple accepted price volatility in exchange for securing volume upfront doesn't mean its negotiating power has disappeared—rather, it suggests a shift in what Apple prioritizes at the negotiating table.
That said, the absence of a price ceiling doesn't necessarily mean Apple will simply pay spot prices. How frequently prices are reviewed and which benchmarks are referenced remain unknown. Minimum pricing, purchase obligations, and how specification changes would be handled are also unclear. Kioxia's official materials state that its contracts retain flexibility to accommodate evolving customer specifications. Without the contract itself being disclosed, it's impossible to calculate what Apple would actually pay or what margins Kioxia would earn based solely on the phrase "no price ceiling."
The Remaining 55% for the Foldable iPhone Can't Be Worked Out by Elimination
TrendForce also cited a SeDaily report projecting that, for Apple's yet-unannounced foldable iPhone, Samsung and SK hynix would together account for 45% of NAND supply—30% for SK hynix and 15% for Samsung. However, the same article does not break down the remaining 55% by company.
This gap cannot be assumed to represent Kioxia's share. Multiple companies could plausibly be involved in Apple's NAND supply chain, and Kioxia and Sandisk continue to operate a joint venture for flash memory development and front-end production. Treating manufacturer, distributor, and markings identified through teardown analysis as interchangeable risks misrepresenting actual supply ratios. It's also possible that different chips are used depending on storage capacity.
The same reconstructed report also cited figures for 12GB LPDDR5 DRAM: Samsung at 37%, SK hynix at 33%, and Micron at 30%. However, these product names and memory specifications predate any official announcement from Apple. Neither the release date nor the supply ratios have been finalized. TrendForce explicitly noted in its September 3 forecast that it is based on current market information and analyst assessments, and that final specifications and pricing could change once Apple makes its official announcement.
The real-world impact of any such contract cannot be fully assessed from Apple's announcements alone. Only after post-launch teardown investigations reveal NAND suppliers by storage capacity, and Kioxia's earnings disclosures show shipment ratios and shifts in customer composition attributable to long-term contracts, will it be possible to verify whether this reported deal actually translated into secured volume. Device pricing and initial inventory levels will also offer clues as to what such a contract ultimately delivered for consumers.