Big week ahead. Everyone knows HBM is scarce. But that alone does not tell us what Micron will report on September 30, this Wed.
The more consequential question for this quarter is how much of the broader rise in memory prices has already reached Micron’s revenue.
Micron guided fiscal Q4 revenue to $49 billion to $51 billion. The highest of 22 analyst estimates in our September 22 snapshot was $52.08 billion.
On Tessara, we just published a dated Tessara House View assigning an 80% probability to revenue above $52.08 billion.
Our separate base-case model projects $56.16 billion.
Why we think previous guidance could be low
Micron issued its guide on June 24. Afterward, TrendForce forecast conventional DRAM contract prices would rise 13% to 18% in July to September. Our question is how much of that price movement Micron captured on sales during its fiscal quarter.
There is a second clue. Four Taiwan memory makers’ monthly revenue, measured in Taiwan dollars, averaged 40% higher in June to August than in March to May. Micron’s guidance midpoint implies about 12% growth in average revenue per day after adjusting for the extra week in its quarter.
Those companies sell different mixes of products, and the Taiwan figures are not exchange-rate adjusted. The comparison is a directional signal, not a way to forecast Micron’s revenue directly. But it is a sizeable gap worth explaining.
Our model starts with Micron’s $50 billion guidance midpoint. Applying forecast DRAM prices to our assumed shipments adds $7.83 billion; carrying forward a NAND pricing premium adds $0.72 billion. We then deduct $1.37 billion for assumed price caps in long-term agreements and $1.02 billion for pricing we estimate was already captured in March. That brings us to $56.16 billion.
The $7.83 billion is our calculation, not a TrendForce revenue forecast or a selling price Micron has reported. We add nothing for the extra week because it is already included in Micron’s guidance.
The strongest reason we could be wrong
Micron’s June guide may already have incorporated more of the subsequent pricing improvement than our model assumes. If so, our bridge counts some of the same upside twice. Long-term contract caps, weaker price capture and softer shipments could compound that error.
This matters more than a generic “memory prices might fall” objection. In our model, flat weekly shipments alone still produce $54.73 billion of revenue. But combine flat shipments with weaker price capture comparable to SK hynix’s measured rates, and the scenario falls to $51.10 billion, below our published threshold.
What we’ll learn on September 30
Reported fiscal Q4 revenue above $52.08 billion would validate the specific call we put on the record. It would not, by itself, prove that our explanation was right. Reported DRAM pricing can also move with HBM and product mix, while Micron may not disclose enough to isolate what its June guide assumed.
We’ll publish what held, what failed and what remains unresolved after the results, regardless of the outcome.
If you think we have misjudged how much pricing was already in the June guide, reply. That is the assumption I most want challenged before the print.
Teng