How to Assess Whether HBM Revenue Growth Is Improving Profitability
HBM's high selling prices do not automatically mean high profits. Learn how to assess the quality of HBM revenue by connecting stacking, good-die yields, customer qualification, and production resource allocation.
Assessing HBM profitability requires looking beyond selling prices to the good dies used in each stack, assembly yields, testing costs, and the burden of production facilities. Even if HBM revenue grows, the overall memory margin cannot stand in for the HBM margin if the company has not disclosed a separate figure.
Combining Multiple Dies Changes the Cost Structure
Micron's explanation of HBM technology describes a structure that vertically stacks DRAM and connects it through TSVs. Applying this structure to an economic analysis highlights the need to ensure the quality of stacking and interconnections even after individual dies have been manufactured. Because each finished product uses several good dies, defects late in the process can cause substantial losses of resources already committed. Rather than interpreting a higher stack count as simply an increase in saleable units, consider changes in materials, processing, and testing costs per package.
Revenue Share Differs from Production Resource Share
An increase in HBM's share of revenue can reflect a combination of high prices, increased shipments, and declines in existing products. If commodity DRAM and HBM compete for the same production resources, analyze the possibility that making more HBM reduces shipments of other products. The relevant comparison is then not just profit per package, but profit earned from constrained wafers or equipment time. An announcement of completed customer qualification also needs to be followed through to the scale of volume-production orders, actual deliveries, and cash collection before it can support a revenue estimate.
How to Analyze HBM When Separate Profitability Is Not Disclosed
If public information provides HBM revenue but no costs, there is insufficient basis to back-calculate an exact HBM operating margin. Even an improvement in the company's overall margin may reflect higher commodity product prices or lower costs. In your analysis, use separate columns for confirmed revenue, management's qualitative commentary, and analyst-assumed yields. Examine redesign costs and order terms associated with customers' generation transitions, but do not fill in undisclosed contract prices as though they were facts. Defining what to check next quarter is more useful than producing a single number that merely looks precise.
Worked example
Hypothetical example: If processing 100 packages costs KRW 80 million in total and 80 can be sold after testing, the input cost per good package is KRW 1 million. If 90 good packages are produced at the same total cost, it falls to approximately KRW 889,000. At a fixed selling price of KRW 1.2 million, revenue differs between KRW 96 million and KRW 108 million. These figures illustrate yield sensitivity; they do not represent actual HBM prices or any particular company's yields.
What to check
- Check whether a standalone HBM margin is disclosed
- Separate revenue growth into price and volume effects
- Distinguish customer qualification from revenue recognition on volume production
- Identify the constrained resource: wafers, stacking, or testing
Sources and primary materials
This is educational material about research methods. For an actual company decision, verify the filings and primary sources for the relevant period.