Published on: August 19, 2026
NAND Supplier Revenue Jumped 77% in Q2 2026 — The Price Increase Is Already Booked as Profit
Combined Q2 2026 revenue at the five largest NAND Flash brands reached US$68.87bn, up 77% quarter-on-quarter, with Micron rising 99.2% into third place and SK hynix Group posting a record operating margin. The growth was ASP-driven rather than bit-driven, which matters more than the headline number. For channel buyers and spot traders, this is harder evidence than any supply-demand forecast: suppliers have converted the price cycle into booked profit, and the commercial case for loosening in Q4 has effectively disappeared. At the same time, YMTC's 8-plane QLC reached retail shelves, giving the domestic second-source conversation its first real quotable data point.
TrendForce published the Q2 numbers on 18 August, and they settle an argument.
Top-five NAND Flash brands: US$68.87bn combined, up 77% QoQ.
The breakdown:
- Samsung — US$23.06bn, +70.7%, share easing slightly to 29.3%
- SK hynix Group (incl. Solidigm) — US$14.27bn, +89.5%, record operating margin
- Micron — US$11.85bn, +99.2%, moving past Kioxia into third
- Kioxia — ~US$10.72bn, +79.9%
- SanDisk — ~US$8.97bn, +50.7%
The number that matters is not the growth. It is where the growth came from.
TrendForce is explicit: this was ASP-driven, not bit-shipment-driven.
Suppliers did not double what they shipped. They doubled what they charged.
That distinction changes how you should read the rest of the year.
Volume-driven growth means the market is expanding and suppliers are rewarded for adding output. Price-driven growth means restraint is the profitable strategy.
Put plainly: the industry just spent a quarter proving that shipping less earns more.
So what exactly is supposed to motivate a Q4 loosening?
TrendForce answered that too — ASP support is expected to carry revenue through Q3 2026 even as smartphone and PC demand weakens.
Consumer softness and memory relief have decoupled. That link broke sometime in 2025, and a lot of sourcing models never got updated.
Is there a second leg
There is one, and it landed this week.
On 19 August, ZHITAI launched the Ti600s built on YMTC's Xtacking 4.0 X4-6080 — described as the industry's first 2Tb 8-plane QLC die.
The specs:
- 147% higher throughput than prior-generation QLC
- 3,600 MT/s interface, 50% faster generation-on-generation
- 7,400 MB/s sequential read, 6,900 MB/s write
- 4K random read +44%, random write +114%
- DRAM-less controller architecture
- RMB 1,189 / 1TB, RMB 1,889 / 2TB, RMB 3,899 / 4TB
Why this one deserves separate attention.
Global QLC capacity has been locked under hyperscaler long-term agreements for two years. Mid-size buyers have not been in the queue at all — they have been outside the building.
Now there is an 8-plane QLC line with finished product on retail shelves at published prices.
That is a rare thing in 2026: a domestic alternative you can actually put an RFQ against.
But state the caveat clearly.
Retail SSD is not enterprise eSSD. DWPD, power-loss protection and customer qualification are not steps you skip. Quoting a 7,400 MB/s retail review figure to an enterprise account is how you inherit a warranty problem.
What to change this week
Delete "Q4 relief" from the build plan.
Not reduce it. Delete it. The income statement has already answered that question.
Reprice every outbound quote off current spot.
Referencing June-July lows today means eating the spread yourself. This increase has been converted to profit; it is not giving that back.
Place restocking orders now.
Enterprise SSD, client SSD, raw NAND — Q3 ASPs are still climbing. Waiting costs money on a known slope.
Rewrite the LTA language.
Older agreements were drafted assuming mean reversion. Write continued Q3-Q4 ASP escalation into the terms instead of leaving an open "at market" clause pointed at your own margin.
Start pricing the domestic second source, but do not switch yet.
YMTC-based die and finished drives are worth an RFQ now, particularly for client and commercial SSD, module and channel spot work. Run the qualification anyway. Specify to the official datasheet.
The read
The most useful thing about this quarter is not US$68.87bn.
It is the proof that supplier and buyer interests are fully opposed in this cycle, and the supplier side just won a quarter decisively.
Memory cycles used to be read off capacity, inventory and demand forecasts. Those still work. But none of them is as direct as looking at the operating margin.
SK hynix just posted a record one.
Until that number turns, every "it is about to loosen" call is missing its evidence.