What Is NRE Cost
Non-recurring engineering cost, or NRE cost, refers to the one-time upfront expense of designing and preparing a chip for manufacturing — covering things like design work, software tools, and photomasks — which is paid once regardless of whether a company ultimately manufactures one chip or ten million.
NRE cost is fundamentally different from per-unit manufacturing cost because it doesn't scale with how many chips are eventually made. Manufacturing a chip involves both a one-time NRE cost — covering design, verification, and mask creation — and a separate recurring per-unit cost for actually fabricating and packaging each chip; because the NRE portion is paid exactly once, spreading it across a larger production volume makes each individual chip's total effective cost lower, which is why NRE cost is a central factor in deciding whether a custom chip design makes economic sense at all.
| Cost type | Paid | Scales with volume? |
|---|---|---|
| NRE cost | Once, upfront | No — fixed regardless of units produced |
| Per-unit manufacturing cost | For every chip made | Yes — total cost grows with volume |
| Effective cost per chip | NRE ÷ volume + per-unit cost | Decreases as volume increases |
High NRE cost is one of the main reasons custom chip designs only make sense at sufficient production volume. Because NRE cost is fixed no matter how many chips get made, a company planning to produce only a small number of units may find that the NRE cost dominates the total cost per chip so heavily that a custom design isn't worthwhile — this dynamic is a major reason many products use off-the-shelf or semi-custom chips rather than a fully custom design unless their expected volume is large enough to justify it.
NRE cost has grown substantially at each new leading-edge process node, making the volume threshold for justifying a custom design climb accordingly. The design tools, engineering effort, and mask sets required at the most advanced process nodes have become significantly more expensive with each generation, meaning the production volume needed to make a custom design at the leading edge economically worthwhile has grown right along with it — a major factor behind the industry's continued interest in techniques like chiplets that can reduce the NRE burden of moving to the newest nodes.
Startups and smaller design teams often specifically evaluate NRE cost against a multi-project wafer option before committing to a fully custom mask set. Because a full custom mask set represents a large fixed NRE expense, sharing that expense across many designs via a multi-project wafer arrangement is a common way for lower-volume or early-stage projects to get a chip fabricated without paying the full NRE cost of a dedicated mask set alone.
Read NRE cost through a fixed-cost-divided-by-volume lens: it's a one-time expense that doesn't change no matter how many chips eventually get made, meaning the more units a design ultimately ships, the smaller that fixed cost's share becomes in each individual chip's total economics.