Something that The Memory Guy tries to do is to find other industries that travel down the same path as memory chips, or, better yet, industries that have already traveled down that path and can share the difficulties they have encountered along the way. Perhaps we can learn something from other industries that will make our own job a little easier. Maybe we can avoid certain difficulties that have afflicted those other industries, simply by knowing in advance that we are likely to encounter them.
There’s a phenomenon that economists know well called “The Tragedy of the Commons.” This is something that few in the memory business are aware of, yet it has a high impact on commodities like DRAM and NAND flash.
Some readers will dislike that I call some of the most complex and advanced technology in the world “Commodities”. Let me explain that a commodity is a product that can be sourced from a number of vendors, so the customer doesn’t really care who they purchase it from. If one vendor is unable to supply the product, another vendor’s product can be quickly and easily substituted. This puts all the focus on price and delivery.
That said, we move to the Tragedy of the Commons.
The name “Commons” refers to a common pasture that each English town would have before industrialization. It was open to all the townsfolk for grazing their livestock (sheep, goats, cattle, horses, etc.).
A farmer who used the commons to graze one animal would do well enough, but the more animals he allowed to graze in the commons the better he would do. The same held true for all the townsfolk.
So we can envision a scenario where everyone does well, and keeps adding livestock to their herds. Everything would work well until the number of animals became so large that the commons would be overgrazed, and all of the townsfolk’s livestock would be underfed.
At this point it would make sense to set a limit on the number of animals that could use the commons, but how do you determine rules after the fact? Would you randomly choose one farmer and evict his livestock from the commons? There really is no fair way to manage this.
Now we turn to semiconductors to see how this applies.
The semiconductor business is a very capital intensive business. For some chips, capital depreciation accounts for over half the cost of the chip. When this is the case, it doesn’t pay to operate a manufacturing plant at anything less than full capacity. When the semiconductor cycle is up, and the industry is profitable, most manufacturers will invest their profits building new capacity. In the Tragedy of the Commons, this would be like adding livestock to the herd.
The new production capacity contributes slightly more than is needed and creates excess capacity, leading to an oversupply, and this drives prices down, which destroys profits. This is the semiconductor industry’s version of overgrazing.
If a manufacturer tries reducing production volume to help increase prices, that manufacturer’s costs will skyrocket due to the business’ capital intensity, so there is a very strong incentive not to do that. During oversupplies some manufacturers might permanently close their less profitable (i.e. higher-cost) production capacity. As a general rule, these plants are liquidated because the operation of the plant would result in greater losses than would be caused by its closure.
Is there a way to use this story to prevent the difficulties caused by the semiconductor cycle? Unfortunately, there is no way in a free market (a market without collusion) to prevent this effect.
This problem is quite an intriguing one, and it deserves careful consideration, but there really is no easy way to avoid the predictable oversupplies and shortages that drive the semiconductor cycle.
I have been known to tell investors and market participants that if anyone ever tells them that the semiconductor cycle is dead, they should find someone else to talk to.