Now that pundits have nervously predicted a modest uptick in global chip sales for 1999, everyone in the business seems to be anxiously watching their pet indicators for early clues to which way industry sales start off the year.
For many folks, those signs will be DRAM pricing and memory production capacity. "Boy, I'll be looking at what everyone else is watching, and that's DRAM pricing," says Dick Green, principal analyst at the Semiconductor and Materials International (SEMI) trade group in Mountain View, Calif.
"We still watch the DRAMs," acknowledges Jeff McCreary, senior vice president of worldwide sales and marketing at Texas Instruments Inc., even though the Dallas chip giant pulled out of memories last year to concentrate on digital signal processing and related mixed-signal ICs.
First indications that the industry was pulling out of its power dive last year came late in the summer when DRAMs' average selling prices began firming up. While fewer chip suppliers than ever are competing in the DRAM markets these days, most of the industry still feels the fallout from excess memory capacity. That's because the memory giants and silicon foundries often shift production capacity to logic chips when DRAMs turn into the big money losers they've been for the past three years.
But other indicators are now being checked closely. People are watching PC unit volumes (particularly those models selling for less than $1,000), Japan's struggling economy, and global economic growth. The signs could come from almost anywhere - sales of new design tools for next-generation processes, cell phones, raw materials for printed-circuit boards, and connectors. Indicators even include the Chinese economy and the Year 2000 software bug.
Industry leaders are watching this crazy bag of indicators closely because many of them have lost confidence in industry forecasts, most of which failed miserably to spot the past three years of industry recession. Some executives, in fact, have promised not to fall for the old "head fake" or false recovery sign that fooled them in 1997.
Remember when the picture started looking pretty good after a very slow start, then the bottom fell out as the Asian flu infected nearly everyone? Chip makers that had started adding new wafer-processing capacity in anticipation of a stronger 1998 were blindsided.
But memory ASPs as an indicator aren't for everyone. Some executives don't want to be suckered by the fickle DRAM business this year. Arthur W. Zafiropoulo, CEO of Ultratech Stepper Inc., says he's not counting at all on a memory-chip recovery in 1999. "I believe we haven't finished seeing an increase of capacity from device geometry shrinks in memories," cautions Zafiropoulo, who expects DRAM prices to tumble 25-to-30% over the next 12 months. While he's not banking on a DRAM recovery until 2000, he does expect to see some pockets of strength in capital spending emerging in the second half of this year.
There is little likelihood of a boom in the chip business this year, however. Most executives expect recovery to be slow and often uncertain. Part of the problem is that new growth depends heavily on general economic factors because there are no new "megaproducts on the immediate horizon," notes veteran Silicon Valley analyst Ed Henderson of Henderson Ventures in Los Altos, Calif.
"Most of us who have tracked the industry for a while know that new technology and products can transcend weak economies, but given the situation today, the semiconductor industry is more at the mercy of the economic factors," he says. "So we'll be spending more time watching mostly economic trends this year, which is something of a change for this industry."