Nick: Keep in mind that costs consist of those that are fixed and those that are variable. The fixed costs of a semi plant are huge, while the cost of turning out one more chip, given that you already have the plant up and running, are tiny. As long as a producer can sell the marginal chip for more than the marginal cost, it makes sense to do so, even if that extra profit, multiplied by the factory's total capacity, is insufficient to cover the fixed costs as well.
That is how some industries continue with everyone losing money for a while - because they would lose even more money if they idled their plants, which wouldn't make the interest bill on the money they borrowed to build their factories go away. That condition, however, doesn't give anyone an incentive to build any more plants, so if all competitors are looking at things from purely an economic view, and demand for the product continues to rise, sooner or later demand will soak up the extra supply and everyone will move back into the black. The question in DRAM is whether there might be players who are building plants for reasons of national prestige (the way every two bit third world country had to have its own steel plant in the 1950's) and whether some players are willing to lose big money for the next few years, knocking out weak competitors, to be in a position to make really big money in 2000+. |