NextFin News -- Apple is reportedly testing DRAM memory chips from recently-listed China-based ChangXin Memory Technologies (CXMT) for its iPhone and MacBook product lines. Back in July, there were reports that Apple had initiated tests of CXMT DRAM for devices sold in the China market, according to reports.
Testing is not the same as procurement, nor does it mean a partnership is a done deal. Follow-up reports said the news could not be independently verified, and neither Apple nor CXMT responded to requests for comment.
Even so, the report sends a clear signal to the market: when the world’s largest consumer-electronics brand starts qualifying a memory supplier that previously wasn’t on its radar, it suggests the memory market’s strain has reached a new inflection point.
"Exponential" Memory Price Hikes
The immediate driver behind Apple’s move was a historic surge in memory prices.
On Apple’s FY2026 third-quarter earnings call on July 30—also Tim Cook’s final one as CEO—he used a phrase he said he had never used in more than four decades in the consumer-electronics industry -- “a 100-year flood on memory pricing with exponential increases in memory prices.”
During the same call conference, CFO Kevan Parekh disclosed that rising memory costs accounted for all of the quarter-over-quarter decline in adjusted gross margin—more than 100% of it. Excluding the memory factor, Apple’s gross margin would have been higher. Cook also said quoted prices for memory purchases had risen for three consecutive quarters and would continue to climb in the next quarter.
Pressure has already been passed through to end-user pricing. This June, Apple raised prices across multiple product lines, including the Mac and iPad. In China, the starting price of the MacBook Air rose from RMB 8,499 to RMB 9,999, and the iPad Air jumped by more than 20%. Although Q3 revenue of US$109.4 billion set a new high for the June quarter, the revenue growth guidance Apple provided for the September quarter (9%–11%) was clearly below Wall Street’s expectation of 12.1%. The midpoint of the gross margin guidance, 46.5%, was also below the quarter’s 48.1% (excluding tariff refunds). Memory costs are the key factor weighing on both of these forward-looking metrics.
The source of the price increases is not on the consumer electronics side. The wave of AI data center buildouts is absorbing memory capacity at scale. The three major DRAM suppliers—Samsung, SK hynix, and Micron—have already shifted a large portion of their advanced-node capacity to higher-margin HBM (high-bandwidth memory) and server DRAM, leaving an increasingly tight supply of consumer-grade DRAM for phones and PCs. TrendForce data shows that in Q1 2026, contract prices for commodity DRAM rose 93%–98% quarter-on-quarter, followed by another 58%–63% QoQ increase in Q2.
Apple’s situation is straightforward. Memory is more expensive, and supply is tighter.
Testing Itself Sends a Signal
If you look only at the transaction level, Apple is still a long way from actual purchases from ChangXin Memory Technologies (CXMT); the two sides have not even reached a price consensus. Reports say Apple had hoped to use CXMT to lower procurement costs, but CXMT has insisted its quoted prices be no lower than those of Samsung and SK hynix.
This also shows that Apple’s primary objective in testing CXMT is not simply to push prices down.
To understand this, you first need to see the supply landscape Apple is operating within. The global DRAM market has long been dominated by three suppliers: Samsung (about 40.5% share in Q1 2026), SK hynix (29.6%), and Micron (19.9%). Together they account for 90% of the market. On the same earnings call, when an analyst asked whether Apple was seeking a more diversified memory supply, Cook made a rare comment: “The DRAM market is basically three suppliers. Clearly, if there were a fourth, that would be a good thing. It would help us on the supply side, and perhaps on the pricing side as well. We’re evaluating all options.”
Apple has always been an exceptionally conservative buyer. Given its scale and quality standards, it typically does not add suppliers for core components lightly; every additional source means extra qualification costs, engineering adaptation work, and quality-control risk. When it does move, it usually signals one thing: the existing options are no longer enough.
CXMT itself is not a taker of idle capacity, either. As the world’s fourth-largest DRAM manufacturer (with an estimated global share of about 7.7%–8% in Q1 2026), it is already operating close to full utilization, prioritizing supply to domestic customers such as ByteDance, Tencent, Xiaomi, and Huawei. In Q1 this year, Changxin posted revenue of RMB 50.8 billion, up 719% year on year; first-half revenue was expected to reach RMB 110–120 billion. HP and Acer have already adopted small volumes of its chips in devices sold into non-U.S. markets.
So Apple’s testing of Changxin Memory can be understood as a kind of pressure gauge. What a stress test cares about is where a system bends first before it breaks. The fact that Apple is willing to initiate qualification for a non-traditional supplier, in itself, suggests that today’s supply tightness has gone beyond what can be managed merely by paying up to existing suppliers.
The memory market has long been known for violent swings, and this round of price increases also carries clear cyclical characteristics.
DRAM supply and demand have long operated in a pattern of mutual overshooting. Shortages push up prices, and higher prices in turn drive capacity investment. New capacity typically takes two to three years to come online, and only then does the shortage ease. Once prices rise to a certain point, buyers begin diversifying suppliers; and once supply returns to normal, buyers often revert to the few vendors with the best performance and highest reliability. By that logic, Apple’s testing of Changxin Memory may be nothing more than a temporary hedge at this cycle’s peak.
This counterargument is reasonable. If DRAM lead times shorten noticeably, contract prices fall for multiple consecutive quarters, and Apple stops qualifying alternative suppliers for devices destined for specific markets, then Changxin Memory would start to look like a temporary patch. If any two of the above signals appear, the thesis of a structural shift would need to be reassessed.
But a few factors are worth examining separately.
First is the nature of AI demand. AI data-center investment is not a short-lived craze; it is a sustained wave of capital spending. New capacity from Samsung and SK Hynix (such as Pyeongtaek P4 and the Yongin cluster) was unlikely to start production until after 2027 at the earliest, and a sizable share of that added capacity will still be allocated first to server-grade products. This means the tight balance in consumer DRAM supply will be difficult to fundamentally relieve over the medium term.
Second, supplier concentration itself is becoming a risk that is being re-examined. Cook’s public call on the earnings call for a “fourth supplier” already showed that this is an issue Apple is proactively addressing, not merely a one-sided concern from analysts. When a company that historically has not been sensitive to supply-chain concentration now discusses it openly, the very act of discussing it becomes a signal.
Over the longer term, the definition of “supply resilience” among large hardware manufacturers is changing. In the past, resilience primarily meant redundancy in suppliers’ production capacity. Today, it is increasingly understood as “switchability between different sources.” This shift is not about whether any one supplier is good or bad; rather, when a single category has only three suppliers, output fluctuations from any one of them are directly transmitted to the procurement side in the form of costs and scheduling.
A Signal Beyond Apple
The second-order effect of this story is that when the largest buyer in a market begins widening its supplier search, it changes not only its own procurement strategy, but also the pricing and bargaining dynamics of the entire industry.
Apple has a scale, cash flow, and ability to make procurement commitments that other hardware makers simply do not. If even Apple was feeling supply constraints in the current memory market, the situation was even harsher for smaller device manufacturers: they neither had Apple’s bargaining leverage nor could they absorb the qualification and adaptation costs that come with switching suppliers.
In fact, this pass-through had already begun. When Microsoft announced Xbox price increases, it said the prices of storage and memory used in game consoles had risen by more than 2.5x; in April, BYD raised the price of certain intelligent-driving add-on packages from RMB 9,900 to RMB 12,000 due to higher costs for automotive-grade storage. Memory price hikes had spread from chipmakers to device brands, from phones and PCs to cars.
This also explains why news that Apple was testing CXMT drew attention far beyond the usual scope of supply-chain reporting. It is a thermometer for gauging how tight the market is, and procurement is just one tick mark. If this testing activity was merely a footnote in the current cycle, it would fade as shortages eased; if more hardware manufacturers began following with similar moves, that would mean the procurement logic of the memory market is undergoing a deeper shift.
What can be said with certainty at this point is that the memory supply was tight, prices were holding at elevated levels, and large buyers were feeling pressure and widening their search scope. The uncertainty is how long these realities will last and how they will evolve.
Several indicators worth tracking on an ongoing basis: the quarterly direction of DRAM contract prices (regular data from firms such as TrendForce); the share of capacity the three major suppliers allocate to consumer DRAM and how it changes; whether more leading consumer-electronics brands begin qualifying non-traditional memory suppliers; and the progress of newcomers such as CXMT in ramping up technology and capacity.
The line Cook left on his last CEO earnings call—“we are evaluating all options”—is it merely a stress response at the top of the cycle, or the beginning of a long-term adjustment to the company’s procurement strategy? The market will provide a clearer answer over the next few quarters. Apple’s testing of ChangXin Memory is one early signal that deserves to be taken seriously.
热门跟贴