ASIC Costs: RE and NRE.

The cost of a chip splits into two parts: non-recurring engineering (NRE) and per-device cost.

Item	Cost (KUSD)	Total (KUSD)
NRE: 6 months : 10 H/W Engineers	250 pa	1250
NRE: 12 months : 20 S/W Engineers	200 pa	4000
NRE: 1 Mask set (45nm)	3000	3000
RE:An 8 inch wafer	5	5n
TOTAL	5	8125 + 5n

For small quantities: share cost of masks with other designs e.g. the MOSIS programme offers multiproject wafer (MPW).

Chip cost versus area

The fraction of wafers where at least some of the die work is the `wafer yield'. Historically yields have been low, but was typically close to 100 percent for mature 90 nm fabrication processes, but has again be a problem with smaller geometries in recent years. The fraction of die which work on a wafer (often simply the `yield') depends on wafer impurity density and die size. Die yield goes down with chip area.

The fraction of devices which pass wafer probe (i.e.~before the wafer is diced) and fail post packaging tests is very low. However, full testing of analog sections or other lengthy operations are typically skipped at the wafer probe stage. Assume processed wafer sale price might be 5000 dollars:

Assume 99.5 percent of square millimetres are defect free.

Area	Wafer dies	Working dies	Cost per working die
2	9000	8910	0.56
3	6000	5910	0.85
4	4500	4411	1.13
6	3000	2911	1.72
9	2000	1912	2.62
13	1385	1297	3.85
19	947	861	5.81
28	643	559	8.95
42	429	347	14.40
63	286	208	24.00
94	191	120	41.83
141	128	63	79.41
211	85	30	168.78
316	57	12	427.85
474	38	4	1416.89

For a chip with regular structure, such as a memory or an FPGA, additional hidden capacity can be deployed by burning fusible straps (aka links) during wafer probe test. This increases yield despite the larger area shipped in defect-free dies. AMD marketed a range of 3-core CPUs where the 4th, present on the die, had been strapped off.