Why Supplier Revenue Growth Does Not Equal Customer Unit Demand
July 16, 2026
Altsets
Research by Altsets Research
Separate price, shipment volume, and product mix before turning supplier revenue growth into a customer-demand read-through, especially in cyclical markets such as memory.
Data used:Altsets Supply Chain Intelligence: 90k+ entities, 400k+ relationships, 20+ years of history.
Key findings
- Micron fiscal Q3 2026 DRAM revenue rose 67% sequentially while bit shipments increased only in the low-single digits and prices increased in the low-60s percentage range, showing why revenue growth cannot be read as unit growth.
- Large mapped Nvidia exposure at Micron and SK Hynix identifies economically relevant customer relationships, while public earnings data is still needed to decompose pricing, volume, and product mix.
A supplier can report explosive revenue growth even when unit shipments move only slightly. That matters when investors use supplier earnings as a read-through to a major customer. Altsets associates Nvidia with 17.62% of Micron revenue. Micron's fiscal Q3 2026 results show why that relationship percentage cannot be combined with headline supplier growth and treated as customer unit demand.
Micron's DRAM growth was mostly price and mix, not bit growth
Micron reported fiscal Q3 DRAM revenue of 31.3B USD, up 67% sequentially. But DRAM bit shipments increased only in the low-single-digit percentage range. Prices increased in the low-60s percentage range, helped by tight industry conditions and favorable mix.
So the revenue growth was not remotely equivalent to a 67% increase in physical DRAM shipments. That distinction changes the customer read-through.
NAND showed the same problem even more clearly
Micron said fiscal Q3 NAND revenue increased 99% sequentially. NAND bit shipments rose only in the mid-single-digit percentage range. Prices increased in the mid-80s percentage range.
A researcher looking only at revenue could infer an enormous demand-volume acceleration. The company's own operating data shows pricing and mix did most of the work.
The relationship percentage does not separate units from price
Nvidia at 17.62% of Micron revenue tells the investor that Nvidia is an economically important mapped customer. It does not say whether the associated revenue is changing because of more units, higher memory prices, richer product mix, a different memory generation, contract pricing, and capacity allocation. The relationship metric identifies the importance of the customer. It does not decompose the supplier's revenue bridge.
This matters for AI demand read-through
Micron said its record quarter reflected the strategic value of memory in the AI era. The company also said HBM4 was in high-volume shipments for its lead customer's platform. Those statements support strong AI-memory demand. But the same quarter also had extraordinary DRAM and NAND price increases.
A valid Nvidia read-through needs to distinguish AI demand evidence from industry pricing evidence and company-specific product-mix evidence. Collapsing all three into one "customer demand" signal overstates what the earnings report proves.
SK Hynix shows the issue is not unique to Micron
SK Hynix also reported record Q2 2026 results and said both DRAM and NAND prices rose significantly quarter over quarter. The company highlighted high-value products including HBM and AI-server DRAM. SK Hynix is associated with 27.88% of revenue from Nvidia in the displayed Altsets data.
Again, strong supplier revenue and a large customer relationship do not tell us how much of the revenue change came from customer units. The same decomposition problem appears at another supplier.
Price can create a false demand signal
Imagine a supplier with flat shipments and a 50% price increase. Revenue can rise roughly 50% before mix effects. An investor who interprets that revenue growth as a 50% rise in customer unit demand will build an incorrect downstream forecast.
The error becomes larger in cyclical industries where average selling prices move dramatically. Memory is an obvious example.
Volume can also be hidden by falling prices
The reverse can happen. A supplier can ship more units while revenue grows slowly because prices fall. That can make end demand look weaker than it actually is.
So the decomposition works both ways. Price and volume should be separated whenever the public data allows it.
Product mix creates a third variable
Even units and average price can be insufficient. Selling a larger share of HBM, high-capacity server memory, or another premium product can raise revenue and margin without a proportional increase in total bits. Mix can therefore create supplier growth that is economically relevant to AI demand but still not equivalent to unit growth. This is why product context matters.
A better read-through hierarchy
When using supplier earnings to infer something about a customer, ask in this order:
- Did supplier revenue change?
- Did shipment volume change?
- Did average selling prices change?
- Did product mix change?
- Is the relevant customer relationship economically material?
- Is there customer-specific product evidence?
- Do the reporting periods overlap?
- Are contracts or capacity constraints affecting the result?
Only after those questions should the investor make a customer-demand inference.
Relationship data still adds something important
Public earnings materials can decompose price and volume. They often do not quantify how economically important one named customer is to the supplier. That is where the relationship data matters.
The public disclosure explains why the supplier's revenue changed. The relationship data helps answer which customer relationships are important enough for that change to matter in another company's research. The two sources are complementary.
A hypothetical example
Suppose the customer-linked relationship represents 17.62% of supplier revenue. Suppose total supplier revenue rises 50%. Without more information, it would be wrong to say the customer's demand rose 50%.
It would also be wrong to multiply 17.62% by 50% and call the result customer unit growth. That multiplication can at most form a highly simplified revenue scenario under explicit assumptions. It cannot identify price, volume, or mix.
A repeatable price-volume read-through workflow
- Retrieve the supplier-customer exposure.
- Read the supplier's revenue growth.
- Find shipment or unit growth if disclosed.
- Find average selling-price changes.
- Identify product-mix changes.
- Separate industry pricing from customer-specific evidence.
- Preserve reporting-period differences.
- Build scenarios only after the decomposition.
- Do not convert supplier revenue growth directly into customer unit demand.
The Micron-Nvidia demand analysis shows why the relationship is economically important. The earnings lead-lag guide explains how supplier disclosures can be sequenced before a customer's report. For relationship definitions and limitations, read the Altsets methodology. Browse Supply-Chain Data Use Cases for other demand and earnings research workflows.