How to Build Revenue Sensitivity Scenarios From Customer Exposure
September 6, 2026
Altsets
Research by Altsets Research
Combine multiple customer revenue shares into explicit demand scenarios so customer-mix risk becomes measurable without turning a linear sensitivity calculation into a forecast.
Data used:Altsets Supply Chain Intelligence: 90k+ entities, 400k+ relationships, 20+ years of history.
Key findings
- With Nvidia at 27.88% of SK Hynix revenue and Apple at 8.75%, a hypothetical 20% Nvidia-linked increase and 10% Apple-linked decline produces a first-order positive 4.701% supplier-revenue sensitivity under linear assumptions.
- Customer-mix scenarios quantify how separate demand shocks can reinforce or offset one another while keeping pricing, capacity, product mix, and contracts outside the simple model.
Customer exposure percentages can be turned into a simple revenue-sensitivity model without pretending they are a forecast. SK Hynix provides a useful two-customer example. Altsets estimates Nvidia at 27.88% of SK Hynix revenue and Apple at 8.75% of SK Hynix revenue.
Those relationships expose SK Hynix to two different demand environments. A first-order scenario model can show how opposite customer-demand moves might combine at the supplier level.
The basic sensitivity rule
Assume sales to a customer change proportionally with that customer's demand shock and all other SK Hynix revenue remains unchanged. Under that simplifying assumption: Supplier revenue sensitivity = customer revenue share multiplied by assumed customer-demand change
For Nvidia, a hypothetical 10% change produces 2.788% of total SK Hynix revenue. For Apple, a hypothetical 10% change produces 0.875% of total SK Hynix revenue. These are scenario contributions, not predicted revenue changes.
Opposing customer shocks can partially offset
Suppose Nvidia-linked demand rises 20% while Apple-linked demand falls 10%. The Nvidia contribution would be positive 5.576%. The Apple contribution would be negative 0.875%.
The combined first-order sensitivity would be approximately positive 4.701% of SK Hynix revenue. The calculation is useful because it shows how customer mix can dampen or amplify a demand scenario.
A small scenario matrix makes the exposure visible
| Nvidia-linked scenario | Apple-linked scenario | Approximate SK Hynix revenue sensitivity |
|---|---|---|
| +10% | +10% | +3.663% |
| +20% | -10% | +4.701% |
| -10% | +10% | -1.913% |
| -20% | +10% | -4.701% |
| 0% | -10% | -0.875% |
The matrix is deliberately mechanical. Real supplier revenue will not necessarily move one-for-one with customer demand.
The two customers represent different product contexts
SK Hynix has a named multi-year technology partnership with Nvidia covering next-generation memory for AI infrastructure. The company also develops mobile DRAM such as LPDDR6 for smartphones and tablets. Apple's historical supplier list identifies SK hynix as an Apple supplier and lists manufacturing locations in China and South Korea.
Those facts help explain why the two mapped customer relationships can participate in different end-market cycles. They do not prove which exact SK Hynix products or current contracts sit behind every percentage.
Customer diversification is not the same as end-market diversification
Two customers can serve the same end market. One customer can also participate in several end markets. A customer-count metric therefore does not tell you whether supplier demand is genuinely diversified. The investor should classify each large relationship by product, end market, geography, channel, contract duration, and demand cycle.
The linear model breaks in several ways
The relationship can have fixed-price contracts. The supplier can be capacity constrained. One customer can gain share from another.
Pricing can change independently from units. Product mix can change. The supplier can redirect capacity.
A customer-demand shock can affect only one product family. All of those effects can make realized supplier revenue differ from the simple weighted scenario. The point of the model is to expose assumptions, not hide them.
Scenario math is better than vague read-through language
Saying SK Hynix is exposed to Nvidia demand is directionally useful. A sensitivity model asks how much the mapped customer share could matter under a stated assumption. It also makes asymmetric relationships obvious.
A 10% Nvidia-linked demand change has more than three times the first-order revenue effect of a 10% Apple-linked change because the displayed Nvidia relationship is much larger. That is a measurable statement.
A repeatable customer-mix workflow
- Identify multiple quantified customers of one supplier.
- Confirm the exposure metric is supplier revenue percentage.
- Define separate demand scenarios for each customer.
- Multiply each customer exposure by its assumed change.
- Add the scenario contributions.
- Keep all unmodeled revenue flat unless another assumption is stated.
- Run several combinations instead of one base case.
- Add product and contract context.
- Label the result as sensitivity, not forecast.
The SK Hynix Nvidia-versus-Apple analysis compares the economic importance of the two relationships directly. This article solves a different problem: how two customer-demand scenarios can combine inside one supplier's revenue base. For relationship definitions and limitations, read the Altsets supply-chain data methodology. Browse Supply-Chain Data Use Cases for other scenario-analysis methods.