Capital budgeting decision: Manufacture wireless chip in house or farm out
Home-on-the-Range Chips, Inc. (HORCI)
Home-on-the-Range Chips, Inc. is a high-tech engineering firm that specializes in designing chips for intelligent wireless applications. They recently completed work on an easily customizable chip with networking capabilities that allows common household appliances and electronic devices to communicate with a central computer or smart phone as well as with each other. The decision HORCI faces, and for which they seek your advice, is whether to produce the chip themselves or outsource production to a company with chip manufacturing expertise.
HORCI estimates the market for this chip will last 5 years, after which the chip will be obsolete. They expect total sales of 3.5 million chips (700,000 chips per year) at $9.50 per chip. The new chip will neither cannibalize nor enhance the sales of any other HORCI chips.
Before you can set up this capital budgeting problem, you need a brief primer on the chip manufacturing process. The fundamental unit in semiconductor production is the wafer, a disk of pure silicon polished to mirror-like brilliance on one side. Each wafer costs $32.00. Individual chips are etched onto this wafer, with one wafer capable of producing many chips. The chips are referred to, singularly and collectively, as “die.” The average number of die obtained from one wafer is referred to as the “gross die per wafer.” Not every die produced from a wafer is useable. Each must be tested. The percentage of the gross die that is of high enough quality to move on to the next phase of production is called the “probe yield.” HORCI believes its gross die per wafer will be 150 with a probe yield of 82%. This means that the average number of die per wafer that advances to the next phase of production will be 123 (150 * 0.82).
During the next production phase, each die is programmed and tested. The percentage of die that pass this final test and become salable chips is called the “final test yield.” HORCI believes that 96% of the die tested will pass. The final test cost per die is expected to be $0.25. Those that pass will be packaged at a cost of $1.38 per chip. The total costs per wafer and the calculations for the yielded part cost are shown in Exhibit 1.
Yielded part cost
To meet expected demand, HORCI plans to build a plant capable of processing 5,930 wafers per year. The plant, along with the equipment in it, will cost $8 million and be put in place over the first year of the project (year 1). Beginning in year 2, plant and equipment will be depreciated to 0 using a 5-year straight-line schedule.  Salvage value is expected to be $1.2 million. In addition, HORCI expects overhead charges related to producing the new chip will be $920,000 per year, beginning in the first year of the project (plant construction) and continuing through the entire 5-year production run. Finally, HORCI will need to purchase the silicon wafers required during the first year (as working capital) to produce the chips it plans to sell beginning in year 2. This investment in working capital (inventory) will be recovered when the project terminates.
This is a 6-year project, with one year to put the plant into place and begin production, and 5 years of sales. Your model should recognize cash flows at the end of each year. For example, the $8 million for the new plant, $200,000 or so for the wafer inventory, and the $920,000 for overhead will be recognized at the end of year 1. Note that the $920,000 overhead in year 1 is a charge against revenue (which is, or course, zero). This will produce earnings before tax of a negative $920,000. Because HORCI can write this loss off against other operating income, the resulting negative income tax of $276,000 in effect becomes a tax deduction, reducing the operating loss in year 1 to $644,000. The entire project will be financed with equity only. The costs or values for these and all other input parameters are summarized in Exhibit 2 below. (HINT: This is a copy the input sector for my model.)
Part I: Certainty
Early in the design phase It was clear that HORCI’s new chip was going to be a winner. Preliminary analysis of the chip’s potential value assumed that HORCI would have the chip produced by a firm that specialized in chip manufacturing. After all, HORCI’s engineers were experts in chip design, not chip manufacturing. The lowest bid they received was $5.50 per chip, excluding packaging costs, from Production Clean Up Corp. (PCUC). A cursory analysis for this project, assuming PCUC produces the chip, suggested that the project would increase HORCI’s equity value by approximately $1.0 million and have an internal rate of return of around 30%. However, a recent informal internal study found that HORCI could manufacture the chip themselves for a yielded or final part cost of just $1.91, inclusive of packaging costs (see Exhibit 1). If so, the value of the chip project would almost double.
problem is loosely based on a paper delivered at the 2007 Crystal Ball User
Conference by Alan E. Gorlick of the
 I make this assumption to keep things as simple as possible. I realize that plant and equipment should be depreciated at different rates.