72-Why Overcapacity Often Gets Worse

Here is an industry with an overwhelmingly dominant leader. No competitor can touch this leader on either Function or Reliability. It looks like the future is secure for this leader. But is it? Some examples of dominant leaders argue yes. Others, though, suggest a warning that dominance may not last for many years.

Posted 1/15/09

The global semiconductor industry is in severe overcapacity today. There are two causes for this current overcapacity: competitor expansion and a fall-off in demand. Competitors expanded rapidly over the last few years when demand was relatively high. New semiconductor capacity comes on stream in big chunks, produced in factories costing more than a billion dollars. Now, many of those factories are running at half their rated capacity as demand has fallen off in the last year. The situation is bad enough that, today, no company can make a profit on standard semiconductor memory chips.

So, why don’t semiconductor manufacturers reduce industry overcapacity by closing plants? The answer lies in the cost structure of these factories. Seventy percent of these factory costs are fixed. They neither rise nor fall with short term changes in demand. As a result, these factories continue operating as long as their operators can make a cash contribution to fixed costs.

These high fixed costs explain why semiconductor prices fall so low in overcapacity. Prices have to fall low enough to discourage someone from producing. That discouragement has to include pricing through the level of any fixed cost. In a commodity-like market, such as semiconductor memory chips, the industry price is equivalent to the cash costs of the next person into the market. As a market expands, the cash cost of the next addition to capacity sets the price. In a market that is shrinking, the cash cost of the last person to leave the market is a pretty good estimate for industry prices. The last person to leave the market is usually the high cost producer. So, a declining demand industry sees industry prices fall below the cash costs of the high cost producer until enough high cost production is taken off-line to balance demand.

Even once a plant reaches the stage where it can not make a cash contribution on sales in its present structure, there are still instances where that capacity does not close permanently. Instead, the capacity recycles in the industry as another competitor, often with a lower cost structure, acquires the productive facility and keeps it operating at a lower cash cost. (See the Symptom and Implication, “Industry profits are low but downsized capacity remains” on StrategyStreet.com.)

Once productive capacity exists, it goes away with difficulty, even in an industry downturn. High cost capacity may not go away permanently until it is unable to compete with the cost structure of newer, more technologically advanced plants, even in a rising price environment. (See the Symptom and Implication, “The industry is adding new more efficient capacity in the effort to reduce cost” on StrategyStreet.com.)

***

Update 2022:

In 2021 global semiconductor industry sales surpassed $550 billion on a record shipment of 1.15 trillion semiconductor units.  The current industry capacity is not sufficient to meet current industry demand. The industry is in a significant up cycle.

Today, 75% of production of semiconductors takes place in East Asia.  90% of the most advanced chips are made in Taiwan.  Taiwan Semiconductor Manufacturing Company is, by far, the industry leader.  It holds 54% of the foundry market, followed at some distance by Samsung at 17%.  The industry is highly cyclical.  Over a 20 year average the industry grows on the order of 13% a year.  Because of the cyclicality leading companies have had to develop significant flexibility to adjust to the rapid pace of change in pricing in the market.  The foundry part of the industry has coalesced around only 3 firms who have the capability to manufacture the most advanced semiconductors: TSMC of Taiwan, Samsung of South Korea and Intel.  Intel, which in 2021 can produce only 10 nm semiconductors, is considering outsourcing some production to TSMC who can produce 5 nm semiconductors and who has just developed the capability of fabricating 3 nm semiconductors In a new factory costing close to $20 billion.

It is likely that TSMC and the other two leaders of the foundry market have control of pricing due to their substantial economies of scale compared to their smaller competitors. See HERE for considerations in establishing a company pricing policy.

***

Update 2/26

The foundry business is at the core modern electronics. It supports many of the features modern society has come to rely on in mobile phones and other electronics. In 2025, the foundry market is rapidly expanding due to the demands of artificial intelligence, 5G phone systems and electronic vehicles. Most semiconductor manufacturers outsource the foundry value added work to just five competitors, who own more than 90% of the market. These competitors include: TSMC (70% share), Samsung (8%), SMIC (6%), UMC (5%), and Global Foundries (4%). Demand is so high in 2025, that all competitors are making attractive profits.

The overwhelming leader in profitability is TSMC, followed by Samsung. TSMC is also the overwhelming dominant leader in both Function and Reliability. The company has the capability to produce both 3 nm and 5 nm nodes (Function). No one else has the range of their capability. The clearest sign of the company’s superb Reliability is its strong relationship with the industries most sophisticated Very Large customers, including Apple, Nvidia, AMD and Qualcomm among others. No competitor can touch them today. Samsung is trying but continues to lose share to TSMC.

The other three of the top five competitors in the foundry business vie with one another for the more mature (larger node) products, including the 28 nm products required for automotive electronics. Still, the market is nowhere near overcapacity so all of these competitors can remain profitable. Their utilization rates and returns are good as you would expect when a market is short of capacity. Eventually, this market will turn and then these lower-ranked competitors must studiously avoid falling into a Leaders Trap and then ensure that each have industry-best relationships with their largest customers.

The future belongs to TSMC to lose. No one can touch them on either Functions or Reliability for its Very Large and Large customers. We have seen such dominance before. Some companies are able to maintain their dominance over many years. Among these very capable competitors are Caterpillar in construction equipment, Visa in credit card networks, Gillette in cartridge razors, Microsoft in operating systems, Google in search, John Deere in agricultural equipment and Coca-Cola. I am sure TSMC would like to be in the same competitive position as any one of those for the next 25 years.

But the history of market dominance does not promise TSMC an easy future. Consider the companies who faced overcapacity that continued to worsen until they were weakened in a major way or out of business entirely. These companies include AOL, Blockbuster, DeBeers, Xerox, Kodak, Blackberry and Nokia. In our work in hostile markets, we have learned that failure, not success, is the market rule.

***