Why Company-Wide Supply-Chain Exposure Is Not Segment Exposure
July 5, 2026
Altsets
Research by Altsets Research
Keep company-wide relationship percentages separate from reportable-segment and end-market exposure unless a direct source supports the internal allocation.
Data used:Altsets Supply Chain Intelligence: 90k+ entities, 400k+ relationships, 20+ years of history.
Key findings
- Nvidia's Compute & Networking segment represented about 89.6% of fiscal 2026 revenue and Data Center represented about 89.7% of total revenue by end market, but neither table allocates company-wide supplier relationships to a segment.
- SK Hynix and Micron relationship percentages should remain company-wide unless customer-specific segment or product evidence supports a narrower denominator.
A company-wide supply-chain percentage should not be silently converted into a segment-level exposure. That mistake becomes especially tempting when one business segment dominates the company. Nvidia is a good example.
For fiscal 2026, Nvidia reported 193.479B USD of Compute & Networking segment revenue and 22.459B USD of Graphics segment revenue, for 215.938B USD total revenue. Compute & Networking therefore represented about 89.6% of total company revenue. Nvidia also reported 193.737B USD of Data Center revenue by end market.
Those numbers make AI and data-center activity economically dominant. They still do not mean every supplier relationship percentage can be assigned directly to the Data Center segment.
Relationship percentages use company-level denominators
Altsets estimates the SK Hynix to Nvidia relationship at 27.33% of Nvidia's cost base. Altsets estimates the Micron to Nvidia relationship at 14.00% of Nvidia's cost base. Those are company-level customer-cost percentages in the displayed relationship data.
They do not say 27.33% of Data Center COGS comes from SK Hynix, 14.00% of Data Center COGS comes from Micron, the relationships belong only to one Nvidia segment, or the percentages can be applied directly to Data Center revenue. The denominator has to remain the denominator actually measured.
Segment dominance does not solve allocation
If Compute & Networking is roughly 89.6% of total revenue, an analyst might be tempted to assume most supplier exposure belongs there. That may be directionally plausible for certain AI-memory relationships. It is still not a measured allocation.
Segment revenue and supplier cost exposure are different accounting dimensions. The company may purchase components centrally. Inventory can serve multiple products.
One supplier can support several systems. Different products can have different memory content. The relationship data does not automatically reveal the internal segment split.
End-market tables add another classification layer
Nvidia's fiscal 2026 end-market table reported:
- Data Center: 193.737B USD
- Gaming: 16.042B USD
- Professional Visualization: 3.191B USD
- Automotive: 2.349B USD
- OEM and Other: 619M USD
Data Center represented about 89.7% of total revenue. That percentage is strikingly close to the Compute & Networking segment share. But the tables answer different questions.
One is reportable segment revenue. The other is specialized end-market revenue. Neither is a supplier cost-allocation table.
Why this matters for investment models
Suppose an investor wants to estimate how a memory-price change affects Nvidia's Data Center gross margin. A company-wide relationship percentage is useful as a starting constraint. It is not enough to calculate the Data Center margin effect.
The model would still need which Nvidia products use the supplier's components, component content per system, segment mix, pricing, inventory timing, purchase commitments, alternative suppliers, and product-specific cost structure. Without those inputs, the segment model contains a hidden allocation assumption.
The same rule applies to supplier revenue percentages
Altsets estimates Nvidia at 27.88% of SK Hynix revenue and 17.62% of Micron revenue. Those figures describe each supplier's company-wide revenue concentration on the Nvidia relationship. They should not be rewritten as 27.88% of SK Hynix HBM revenue, 17.62% of Micron HBM revenue, or a direct share of either supplier's AI-memory business.
Public product commentary can show that HBM is strategically important. It cannot retroactively change the denominator in the proprietary relationship metric.
Segment evidence can still make the relationship more useful
The correct approach is not to ignore segment information. It is to use segment information as context. Nvidia's Data Center scale tells us the company is overwhelmingly driven by accelerated computing and AI infrastructure in the current period.
SK Hynix and Micron both publicly discuss HBM demand. That makes the relationship economically interesting. But the strongest defensible statement remains at the company-relationship level unless more granular evidence exists.
A practical evidence ladder
An investor can separate four levels. A company-wide relationship is quantified against total supplier revenue or total customer costs. Segment relevance comes from public filings indicating that a segment is economically important, while product relevance comes from disclosures showing that the supplier participates in the relevant product category.
A segment-specific relationship requires a direct source quantifying the commercial relationship inside that segment. Only this fourth level supports a true segment-level exposure percentage.
A repeatable segment-allocation workflow
- Record the original relationship denominator.
- Retrieve the company's segment and end-market mix.
- Identify product categories plausibly connected to the relationship.
- Do not rebase the relationship percentage onto a segment without evidence.
- Search customer-specific product disclosures.
- Build a range when allocation remains uncertain.
- Label company-wide and segment-level conclusions separately.
- Revisit the model when more granular disclosures appear.
The supplier-product validation guide explains how to establish product relevance without inventing contract detail. The margin-sensitivity guide shows why preserving the correct denominator matters when relationship data is fed into a financial model. For metric definitions, read the Altsets supply-chain data methodology.