The Impact of IC Shortage on the Display Supply Chain

To understand why we are having a prolonged shortage of integrated circuits, we need to understand something about the economic factors driving the semiconductor industry. Leading-edge semiconductor fabs cost billions of dollars to build and take time to develop and reach satisfactory yields. However, the large, complex chips from the leading edge fabs are sold in high volume and at high prices for servers, laptops, and phones. 

The Semiconductor industry uses a metric of revenue/wafer as a metric to decide on investments and prioritize customers in times of capacity shortage.

Larger diameter wafers are more expensive, but each large wafer can hold many individual chips. The number of potential chips on that wafer, times the number of chips that pass testing (percentage yield), times the selling price of each chip, is the revenue per wafer.

For example:
Potential dye = 500 
Yield = 60%
Selling price = $10
500 * 60% * 10 = $3,000 per wafer 

Leading-edge fabs are expensive, but their large, 300mm diameter wafers yield many costly leading-edge chips. That means there is a good return on that considerable investment, so leading-edge factories are still being built, and there have been problems sourcing these leading-edge chips.

The picture is different for older technology manufacturing. After a few years, technology has moved on, and the wafer fab is no longer the leading edge, but much of the investment has been depreciated. The kind of chips being built can shift to more cost-competitive, lower volume, higher mix, output, but since depreciation is tailing off and yields are very high because of the learning curve effect, it still makes economic sense to run these older fabs.

After another few years, there is no further depreciation, and wafer costs are low. Many 200mm fabs are at this stage. Chip selling prices were low, and revenue per wafer was also low because the wafers were smaller. 

As a result, there was no incentive to invest more capacity in these fabs. As a result, the chips produced in these fabs are relegated to the housekeeping tasks around the systems with leading-edge chips at their heart. These housekeeping tasks include radio interfaces, power management, display driving, human interfacing. These tasks need higher voltage capability that the leading-edge chips can't handle. There is also a problem in building these older, simpler, but tiny chips on big 300mm wafers. One wafer might yield a month's worth of chip demand. Since wafers are built in lots of 24 or so, one wafer lot covers demand for two years, and then the factory has to switch to another device, leading to inefficiency.

The automotive industry has always been very conservative in the semiconductors it uses, often ten or 15-year-old technology. This approach makes sense in modern cars, which are complex systems with over 100 integrated circuits and voltages from 12 volts upwards that would not be kind to leading-edge semis. Therefore any chip in the vehicle that fails could cause a trip to the dealership. Add to that complexity the many years the industry has been refining "just-in-time" supply chain management.

Now we add the perfect storm into the situation. Covid shuts down ports and factories. Millions work from home, driving up demand for home networking, which uses many older "housekeeping" chips. At the same time, demand for autos drops dramatically. Just-in-time logistics shut down demand on the wafer fabs, who shift production to those other "housekeeping" lines. Demand for electronics for the home continues to grow, home automation takes off, so when the Auto Industry sees the need to bring back their "just-in-time" pipeline is empty, and the older wafer fabs are full. The Auto Industry panics, demands capacity, and piles on orders, making the rest of the electronics market see lead times stretch out, panic ensues, and orders keep piling.

The fix is not easy. Nobody has built a 200mm fab for years. Much of the equipment is out of production. Building a fab takes two years, getting it into production takes another year, and since the loading is from many customers, qualifying production takes even more time. Instead, the preferred route is to look at each fab and increase capacity by buying an extra piece of equipment to help with bottlenecks. This approach is much faster and less capital expenditure, as long as that piece of equipment is available.

The bottom line is that we expect the shortages to slowly ease over the next year as many older fabs make incremental capacity improvements. However, we will still see prices increase as the fabs look at the revenue per wafer for each product. We also think there is some double ordering in the market, but fabs are strict about NCNR (No Cancellation, No Return) rules, which will minimize this. USMP will continue to use our Display Industry relationships to mitigate the impact of this crisis on our customers. Please place your orders early to help us to help you.

If you have a project that is considering taking advantage of any display technology, US Micro Products can provide a solution designed for your application. Send us an email at sales@usmicroproducts.com.