Semiconductor Revenues, Space and Employment

The number of employees working in semiconductor cleanrooms is an important statistic for suppliers of consumables such as gloves and garments. The large investment now being made in 300 mm fabs will have the effect of moving people out of Class 1 cleanrooms and into Class 1000. This is due to the use of minienvironments and robotics. Robotics will also reduce the total number of workers per square foot of space and per dollar of revenue.

The U.S. semiconductor industry in mid 2004 employed 240,000 people. Intel alone employed over 70,000 people. McIlvaine estimates that 35 percent of these people work in cleanrooms. This compares with a 60 percent ratio of production workers to total workers in the broader SIC Class 367. For example in 1996 total U.S. employment in SIC 367 was 587,000 people of which 368,000 were production workers. In 1997 there were 64,000 people classified under the job description "electronic semiconductor processor. This includes all the people loading, cleaning, polishing and performing the other related tasks.

Because semiconductor revenue statistics are the most accessible, it is important to determine the relationship between numbers of employees and revenues. It is also important to determine the relationship between revenues and numbers of employees in each class of cleanroom The revenue per employee varies considerably within the industry. At Intel the revenues in 1999 were $29 billion and there were 70,000 employees. Revenue per employee was $414,000. McIlvaine estimates that 32,000 of these employees work in cleanrooms. So revenue per cleanroom employee is $906,000.

National Semiconductor is more typical. It had fiscal 2000 revenues of $2.1 billion. It employees 10,500 people. Sales revenue per employee is $204,000. Sales revenue per cleanroom employee is estimated at $400,000. This number is weighted by the fact that National has assemble plants in Asia where employment per unit of revenue is high.

For purposes of determining usage of consumables and expenditures for cleanroom hardware there is a need to divide cleanroom workers into classes. The newest TSMC 300 mm Fab will have Class 1000 space and Class 0.1 mini environments. This compares to the Class 1 space of traditional ballroom fabs.

In our previous report McIlvaine used the following distribution of cleanroom workers in semiconductor fabs. Sixty percent Class 1; 5 percent Class 100, 15 percent Class 1000; 10 percent Class 10,000 and 10 percent Class 100,000. In this report the following distribution is used. Fifty percent Class 1-10, 5 percent Class 100, 25 percent Class 1000, 10 percent Class 10,000, and 10 percent Class 100,000. This means that there will be lower percentages in Class 1 and higher percentages in Class 1000 and that the others remain the same.

Another useful factor in projecting cleanroom expenditures is the relationship between revenues and wafer starts. TSMC, as an example had sales of NT $65 billion ($2 billion) in the first half of 2000. Annual sales are therefore $4 billion. Wafer annual capacity in 1999 was 1.8 million eight inch equivalent wafers. This is expected to reach 3.4 million by the end of 2000. So this is roughly $1000/wafer. This would mean that a typical 20,000/wafer start per month fab would generate $40 million per month or $480 million per year in revenues.

With the finer line sizes and bigger wafers the revenue per wafer will increase substantially. Previously we used revenues /ft2 of Class 1 space = $22,485. For the next five years we are using an average of $25,000 but with the realization that the umber will increase each year.

Previously the model showed 100 ft2 of cleanroom space per employee. This should increase because of automation. However, we have redefined space to "space in use" and have therefore, adjusted the model to 40 ft²/employee. Since there are often three shifts, the employee is occupying a space of 120 ft/2. The ft2 per employee is lower when capacity is higher and the plants are all operating with maximum number of shifts and workers.