what is a multi project wafer
A multi-project wafer, or MPW, is a single wafer run shared by several different chip designs from different customers, splitting the fixed cost of a fabrication run across multiple parties so each individual customer pays only a fraction of what a dedicated wafer run would cost — a common way for smaller teams and prototypes to access advanced manufacturing processes affordably.
```flowchart
{
"rows": [
{ "type": "nodes", "items": [
{ "title": "Several customers each need a small chip prototype run", "sub": "none needs a full dedicated wafer's worth of chips", "tone": "neutral" }
]},
{ "type": "arrow" },
{ "type": "group", "title": "Multiple designs share one wafer", "items": [
{ "title": "Each design occupies a section of the wafer", "sub": "fabrication cost split across all participating customers", "tone": "green" }
]},
{ "type": "arrow" },
{ "type": "nodes", "items": [
{ "title": "Each customer gets a small batch of their own chips", "sub": "at a fraction of a dedicated run's cost", "tone": "orange" }
]}
]
}
```
**An MPW solves the problem that a full dedicated wafer run is often far more chips than a single small design actually needs, especially for prototyping.** Fabricating an entire wafer dedicated to just one design incurs the full mask and fabrication cost regardless of how many chips that customer actually needs, which can be wasteful for prototypes or small production runs — an MPW instead divides a single wafer's area among several unrelated designs, so each participating customer pays a proportional share of the shared cost rather than the full cost of an entire dedicated run.
```svg
```
| Aspect | Dedicated wafer run | Multi-project wafer (MPW) |
|---|---|---|
| Cost per customer | Full fabrication cost | Shared, proportional cost |
| Typical use | Full production volume | Prototyping, small runs, education |
| Number of designs per wafer | One | Several, from different customers |
| Manufacturing schedule flexibility | Customer controls the schedule | Tied to the shared MPW run's schedule |
**Because an MPW run is shared, participating customers have to work within a fixed schedule set by whoever organizes the run, rather than choosing their own timing.** An MPW run typically has a set submission deadline that all participating designs must meet, and a shared fabrication schedule that no individual customer controls — this loss of scheduling flexibility is one of the tradeoffs that comes with sharing a wafer's cost across several unrelated designs.
**MPW programs have historically played an important role in making advanced process nodes accessible to universities, startups, and researchers who couldn't otherwise afford dedicated wafer runs.** By pooling many smaller-scale designs together, MPW programs let organizations without the budget for a full dedicated production run still get real, working silicon prototypes manufactured on genuinely advanced processes — this accessibility has meaningfully lowered the barrier to entry for chip design experimentation and research.
**Moving from a successful MPW prototype to full production volume requires transitioning to a dedicated wafer run, since MPW capacity is inherently limited per customer.** An MPW run only allocates a small section of the wafer to each participating design, so it's well suited for validating a design's basic functionality but not for producing chips at meaningful commercial volume — a design that succeeds on an MPW run and needs to scale up typically has to move to a dedicated wafer run once its needed volume exceeds what a shared prototyping run can provide.
Read the multi-project wafer through a cost-sharing lens: rather than every small or early-stage chip design bearing the full cost of a dedicated wafer run, an MPW pools several unrelated designs onto one shared wafer, splitting fabrication cost proportionally and making advanced manufacturing genuinely accessible to teams that couldn't otherwise afford it.