The RFQ process optimizes for the wrong variable. Unit price is easy to compare — it fits neatly into a spreadsheet column, it responds to negotiation, and it produces a clear winner. The problem is that unit price is rarely where procurement decisions actually go wrong.
The real cost shows up later: in delayed shipments when a supplier can't fulfill during a shortage, in deployment friction when firmware doesn't behave as documented, in re-qualification costs when a chipset platform reaches end-of-life. None of these costs appear on the original RFQ.
Supply chain depth is the first non-price variable that matters.
Ask your supplier directly: do you hold chipset-level buffer stock, or do you build to order against spot allocation? A supplier who can't answer this question clearly doesn't have a strategic inventory position.
Technical support quality is the second.
Platform-level support means someone who understands CLI behavior, MIB implementation, and firmware interaction with your specific topology. Product-level support means someone who reads the datasheet back to you. The difference becomes obvious the first time you hit an edge case in production.
Chipset continuity is the third.
A supplier who carries products across multiple unrelated chipset platforms is optimizing for catalogue breadth, not technical depth. Ask which chipset families your supplier is committed to, and what their policy is on lifecycle management when a platform approaches EOL.