06-reference/research

micron dram asp independent series vs trendforce

2026-11-15·research-brief·source: deep-research·by Ray Data Co (deep-research synthesis)
memory-cycledramasp-seriestripwiremicrondata-provenance

Micron's own DRAM price disclosures make a usable second series, but only for the direction of the turn, not the size of the numbers

The question

Verbatim: "Can Micron's sequential DRAM ASP disclosures be assembled into a second, genuinely independent ASP series to cross-check TrendForce - and does the resulting series agree with TrendForce's conventional-DRAM contract QoQ prints for 1Q26-4Q26?"

Context: [[2026-10-03-dram-asp-series-basket-consistency-tripwire-1]] made one series binding for Tripwire #1: TrendForce's conventional-DRAM contract-price QoQ range from the free press centre. That leaves one free source as the only input to the scoreboard's earliest tripwire. That brief's open follow-up asked whether Micron's own disclosures could serve as a check. This brief builds the series, measures the three confounders (HBM inclusion, the fiscal-quarter offset and contract structure), and scores agreement.

Headline answer: Yes. The series can be assembled from free primary sources, and it is genuinely independent: it is realized revenue divided by shipped bits, not a survey of contract quotes. It agrees on direction. Both series decelerate for two consecutive quarters into 3Q26, and Micron's version is built on actuals rather than forecasts. It does not agree on level. After aligning the calendars, Micron's realized gains run roughly 2-12 points below TrendForce, with a midpoint gap of about 6-8 points. That gap fits HBM dilution plus contract price ceilings, but this brief cannot fully attribute it. 4Q26 cannot be checked yet. Micron has reported nothing after FQ4 FY2026, and the matching quarter (FQ1 FY2027) is due around mid-December. Verdict: keep the Micron series as a sign-of-the-turn check, not a magnitude check. It also becomes unusable for 1Q27 unless HBM is stripped out.

What we already know (from the vault)

What the web says

Micron's DRAM figures by fiscal quarter. Micron reports DRAM revenue, a bit-shipment band and a price band each quarter, and its DRAM line includes HBM.

Micron FQ (FY2026) Approx. calendar span DRAM revenue Seq. revenue Bit shipments Prices (Micron's words) Source tier
FQ1 Sep-Nov 2025 $10.8B +20% "slightly" up ~20% B (transcript summaries via search, not fetched)
FQ2 Dec 2025-Feb 26, 2026 $18.8B +74% "up mid-single digits" "increased in the mid-sixties percentage range... and included favorable mix" A- (Motley Fool transcript of the call, fetched)
FQ3 Feb 27-May 28, 2026 $31.3B +67% "up low-single-digit percentage range" "increased in the low-60s percentage range, driven by tight industry conditions and favorable mix" A (Micron prepared remarks PDF, fetched)
FQ4 May 29-Sep 3, 2026 (14 wk) $39.8B +27% "up mid-single-digit percentage range" "increased high-teens percentage range" A (Micron prepared remarks PDF, fetched)

Convergences and contradictions

Synthesis for RDCO

The series is real, free and independent, and on the one thing Tripwire #1 measures, it agrees. From four quarters of prepared remarks and transcripts, Micron's blended realized DRAM price change reads +~20% (FQ1, tier B), +mid-60s (FQ2), +low-60s (FQ3) and +high-teens (FQ4). As a cumulative index on band midpoints (my arithmetic, FQ4 FY2025 = 100) it runs 120 → 198 → 320 → 378. It shares no inputs with TrendForce. TrendForce surveys negotiated contract quotes for conventional DRAM. Micron reports what one supplier actually billed across all its DRAM, HBM included, divided by bits shipped. Both series say the QoQ gain peaked and then decelerated twice into the June-August window. The single-source risk the parent brief flagged is therefore partly retired: if TrendForce's direction were wrong, Micron's actuals would show it. They do not.

The level gap is real and should be read as structural, not as error in either source. To compare like with like, I placed TrendForce on Micron's calendar. Each Micron fiscal quarter is weighted as one-third of the prior calendar quarter's level plus two-thirds of the current one (TrendForce index 4Q25 = 100; midpoints give 195.5, 313.8, 362.4, 407.7 for 1Q-4Q26). This is a crude step model, so I also ran a one-month realization-lag variant. The lag variant makes TrendForce's implied gains larger (FQ3 +78%, FQ4 +40% at midpoints), so the gap's sign survives the timing assumption. Three confounders, quantified where possible:

On top of these sits the ±3-point resolution of the verbal bands. Together, HBM dilution plausibly explains most of the FQ3 gap (~6 points at h ≈ 9%). The FQ4 gap of ~8 points at midpoints (2-5 on the revenue back-solve) is roughly covered by HBM plus SCA ceilings plus band resolution. That is consistent, but it is not a proof. The residual is the size of the confounders, so this series cannot check TrendForce's magnitudes.

For the scoreboard, the Micron series should be a corroboration row with a sign test, never a substitute reading. This keeps the parent brief's binding rule intact. Its value is that it answers the sign-of-the-second-derivative question from actuals, with a lag of one fiscal quarter. Its known failure mode is coming. From January 2027, HBM reprices with "significant price increases year over year" (TrendForce: blended +121% YoY). With HBM at 10% of DRAM revenue, a step of that size adds roughly 10-12 points to Micron's blended ASP in the quarters containing January, regardless of conventional DRAM (my arithmetic). That alone could make FQ2 and FQ3 FY2027 read as re-acceleration while conventional contract prices keep moderating. SCA floors also mean Micron's realized ASP will fall later and less than contract prices in a rollover. So this series is a damped, lagging check on Tripwire #4 (the confirm), not an early warning. Neither point is a position action.

Pre-registration for the next print (FQ1 FY2027, Sep 4-early Dec 2026, reporting ~2026-12-16, unconfirmed). Calendar-aligned TrendForce implies an FQ1 FY2027 gain of +10.9% to +15.9%, a sharp deceleration from the aligned FQ4's +23.8-28.6%. Netting the structural gap of 2-6 points gives an expected Micron reading of roughly +7% to +12%. AGREE if Micron states the DRAM price gain at "mid-teens" or lower but still positive, and the revenue/bit back-solve's upper bound is below 20%. DISAGREE (no deceleration) if the stated band is "high-teens" or higher. DISAGREE (ahead of TrendForce) if prices decline. The FQ1 revenue guide does not settle this in advance. The guide is $61.5B against $54.2B, +13.4% total, with one fewer week. Depending on bit assumptions, that implies a blended DRAM-plus-NAND price change anywhere from about +8% (bits +5% in total) to +22% (flat per-week output, no inventory drawdown) (my arithmetic). Given that the guide is consistent with a high-teens print, the DISAGREE branch is a live possibility, not a formality.

A small correction to carry. [[2026-10-01-micron-fq4-fy2026-dram-hbm-tripwire-rescan]] cites "a more moderate rate of price increases" as primary corroboration of TrendForce's 4Q26 deceleration. The sentence applies to fiscal 2027 "beyond fiscal Q1." It corroborates the 2027 trajectory, not 4Q26. Not applied here; flagged for the parent run.

Why this is in the vault

It answers the 2026-10-03 follow-up on whether Tripwire #1's single-source TrendForce dependency can be checked without paying for a subscription. The answer, a sign-only Micron corroboration row with a pre-registered FQ1 FY2027 band, is what the memory-cycle-v1.1 quarterly review scores in mid-December. The brief also flags the 1Q27 HBM repricing step that will make the Micron row unreadable unless HBM is stripped out.

Open follow-ups

Related

Sources