Three things used to be certain: death, taxes, and the announcement of a newer, faster, and cheaper computer two days after you buy one. Innovation continues to produce faster CPUs and less expensive hard drives, but the market for memory seems to have stalled. For several years now, 1-MB DRAM SIMMs have cost between $30 and $40. What happened to the steady drop in price per megabyte that we became so accustomed to?
The problem isn't in the labs. RAM manufacturers are making 16-, 64-, and 256-Mb chips. Both Hitachi and NEC recently announced success in developing a 1-Gb RAM chip. But those chips won't be available in quantities until the year 2002, and their rivals call these announcements a public relations move. Neverthe
less, the research scientists continue to produce a new generation of RAM chips about every three years. But we're not seeing this research success translate into commensurate price drops.
The problem seems to lie with the CPU and software industries, where relentless competition has created an unexpectedly large and continuing demand for RAM. This demand continues to exceed supply, as people clamor for new machines running software that can't function without 8 MB of memory.
Multitasking OSes like OS/2, video games, and multimedia all need as much RAM as possible to function. Newer RISC chips such as the IBM/Apple/Motorola PowerPC add to the demand for more RAM, because the native object code for these RISC chips can be 30 percent to 40 percent larger than the corresponding code for the older 680x0 Macintoshes. One Intel engineer took credit for the problem, saying that his company's relentless price-cutting and innovation made high-end CPUs with high-end memory requirements available to the ma
sses. As long as the demand continues to soar beyond capacity, the RAM companies keep the price high and pocket the profits.
This high demand slows the adoption of new technology in two ways. First, RAM manufacturers have no need to push the newer 16-, 64-, and 256-Mb chips because they're making enough money on the smaller chips.
Second, systems designers often skimp on memory to bring the price of complete systems down to bargain rates. 64-Mb chips produce 16-MB SIMMs, and their use forces many systems manufacturers to produce computers with a minimum of 32 MB of memory. The lower demand for these large configurations slows demand for the bigger chips, and that, in turn, slows the evolution. At the moment, computer manufacturers are more likely to push 4-MB machines with a lower initial cost, even though people may be frustrated when they try to run new memory-hungry applications on them.
The only solution will be to equalize the supply. The high profits give the industry plenty of mone
y to invest in new fabrication facilities, and there's plenty of evidence that new capacity will emerge. When this happens, increased competition will again drive manufacturers to greater innovation in their efforts to gain market share. Then, and probably only then, will we see prices start dropping again. Maybe we should try to talk Microsoft into investing in a RAM fabrication plant?