On September 21, 2026, TrendForce announced that major U.S. cloud providers had raised their enterprise SSD demand forecasts over the past two weeks. Orders in the fourth quarter of 2026 could exceed those of the third quarter, which were already at a high level, and enterprise SSD prices are expected to keep rising. However, this is an outlook from a research firm. Order volumes, the size of price increases, and the names of individual providers have not been disclosed. At the end of July, the company had expected price increases to slow. Why is it now taking a more bullish view? The key lies in North American procurement plans and demand for high-capacity QLC products.
From a July "Slowdown" to a September "Continued Rise"
The summary of TrendForce's enterprise SSD contract price report, published on July 29, explained that supply-side capacity expansion was calming urgent orders and that buyers were resisting price hikes. Shipment delays for next-generation platforms were also cited, and the view was that the pace of contract price increases would slow. This did not mean demand had vanished. Server orders would continue, but once supply began to catch up, sellers' ability to raise prices would weaken.
In the new announcement on September 21, the factor driving supply and demand had changed. Major U.S. cloud providers had raised their enterprise SSD demand forecasts over the past two weeks. TrendForce analyzes that the background is a move to bring forward equipment deployment and to increase the uses to which SSDs are put. Based on current procurement conditions, it indicated that total fourth-quarter orders could exceed those of the third quarter.
| Date published | Factors cited by TrendForce | View on prices | Scope and limits |
|---|---|---|---|
| July 29 | Improved supply, buyer resistance to prices, delayed platform shipments | Pace of increases to slow | Public summary on Q3 enterprise SSD contract prices |
| September 21 | Upward revision of demand outlook by U.S. cloud providers | Forecasts continued increases | Q4 enterprise SSD prices. No rate of increase given |
TrendForce's public summary at the end of July anticipated a slowdown in enterprise SSD contract price increases due to improved supply and buyer resistance to prices, while its September 21 announcement predicted that enterprise SSD prices would keep rising in the fourth quarter following an upward revision of North American demand.
The two sources cannot be read as a numerical revision of the price forecast for the same quarter. The July summary deals with contract prices for the third quarter, while the September announcement discusses enterprise SSD prices for the fourth quarter in qualitative terms. What the table shows is how the assessment of supply and demand shifted between different publication dates. How much actual contract unit prices will change, in percentage terms, cannot be calculated from public materials.
High-Capacity QLC Orders Join High-Performance TLC
Explaining the current demand increase solely through orders for high-speed SSDs would be a mistake. According to TrendForce, the products sought by U.S. cloud providers are expanding from high-performance TLC to QLC, which prioritizes capacity and cost efficiency. The company cites as background the processing of AI agents that search for and temporarily hold large amounts of information, as well as unstructured data stored in vector databases. The view is that what is placed in high-speed memory and what is placed in high-capacity SSDs will affect the cost of AI services.
There are also regional differences in how demand is explained. For China, TrendForce cites a move to offload KV cache to high-capacity QLC SSDs in AI configurations centered on DeepSeek. The KV cache is the precomputed information a model retains so it can refer back to past inputs. While it may reduce the amount that must stay in high-speed memory, it requires designing for SSD reads and writes and for response times. This Chinese example should not be added on as a breakdown of North American order volumes.
On the product side, preparations for high-capacity QLC have also advanced. In May, Micron announced that it had begun shipping the 245TB "6600 ION", which uses QLC NAND. TrendForce's September 1 tally also cited expanded shipments of Samsung's 176-layer QLC products, an increase in Micron's 232-layer QLC products, and expanded mass production of SanDisk's high-capacity QLC products. It can be confirmed that shipments and mass production of high-capacity QLC products are progressing. However, these announcements do not reveal the volumes adopted by specific cloud providers or the order value for the fourth quarter.
In its September 1 announcement, TrendForce estimated combined revenue for the top five enterprise SSD brands in the second quarter of 2026 at $37.59 billion, up 103.6% from the previous quarter. The company believes higher shipment volumes and contract prices contributed, but the doubling of revenue cannot be used as evidence that the number of units ordered or the storage capacity shipped also doubled. Revenue moves even when product capacities and configurations change.
Not Directly Linked to PC SSD Prices
While enterprise procurement is strong, the outlook for consumers is different. The summary of the 2027 NAND market report that TrendForce published on September 16 predicts that while AI-driven enterprise SSDs will grow, prices for PC and smartphone products will be pressured by weak demand and increased supply capacity in China. What could be confirmed was limited to the public summary and table of contents; the detailed price tables and forecast figures in the paid report were not verified.
Therefore, the forecast that "enterprise SSD prices will rise" cannot be transferred as is to a forecast for the prices of SSDs sold in retail stores in Japan. Suppliers, contract terms, required capacities, and inventory buildup all differ. What TrendForce showed on September 21 was the strength of North American enterprise orders, while what the September 16 public summary showed was weak consumer demand in 2027. Forecasts that differ in region, use, and timing cannot be combined into a single retail price trend. Even though NAND production capacity is shared, prices diverge depending on how much is allocated to which use. On September 21, TrendForce also analyzed that stronger enterprise demand works in suppliers' favor in allocating production capacity and setting prices.
However, sellers' bargaining power has limits. The summary of the contract price report at the end of July touched on the possibility that a widening per-capacity price gap between high-capacity SSDs and HDDs could erode the total-cost-of-ownership advantage of choosing SSDs in data centers. When companies expand capacity, they look not only at power and installation space but also at procurement cost per unit of storage. Even if the uses for QLC expand, it has not been decided that all storage will be replaced with SSDs.
What to Check in the Fourth-Quarter Numbers
What is confirmed in the September 21 announcement is that TrendForce learned that U.S. cloud providers had raised their demand forecasts and, from that, expects fourth-quarter orders and prices to rise. Order volumes by provider, the QLC share of the mix, and the rate of price increases have not been disclosed. These blanks cannot be filled with past revenue or product capacities.
To verify the outlook later, one needs to check actual fourth-quarter orders and shipments, enterprise SSD contract prices, and the QLC share of products. It will also be worth watching how much of the NAND supply expected to grow in 2027 goes to enterprise use, and whether buyer resistance to prices strengthens. Even if orders increase, price increases can slow if supply grows faster. TrendForce's July explanation and September explanation differ in which side of this supply-demand balance they weighted more heavily. High order volume and sellers being able to sign contracts at their desired unit prices are separate matters.
Conversely, even if products exist, if customer qualification and data center deployment do not progress, the current outlook will not necessarily turn into shipment results. Even where high-capacity QLC spreads, the speed and capacity required will vary with the type of data stored. It will be important to judge whether procurement proceeds use case by use case, rather than as a uniform switch from TLC to QLC.