Thursday, November 12, 2009

connecting is such fun

This week, with Ray Price in town, has been wonderful for us. We've met a number of old colleagues around the circumstance of the book being published. Alan Bickell shared a cup of coffee, and we reminisced about Y-HP, and he helped put more "flesh on the bone" about the semiconductor test set done there, as well as the emergent Samsung memory chips a decade later. Dean Morton added his perspective the next day, and Art Fong contributed more that night. The story is impressive -- it has to do with how a small remote division ignored Palo Alto leadership for years, finally prevailing on a very difficult project, only to find they couldn't sell it, and yet eventually would reshape the entire semiconductor world and the companies in it.

When Y-HP sold one unit, eventually, to Hitachi, that company found that it could control doping levels for impurities much more precisely -- a fact that relatively soon led to a massive adoption by other Japanese firms, and gave the world better memory chips much cheaper than U.S. vendors, notably Intel and Texas Instruments which were two of the leading worldwide Dynamic RAM vendors. There were three repercussions in relatively short order:

1. The Japanese "took over" the 16K and then the 64K RAM chip business, going from less than 10% of the business in 8K chips (1977) to more than 80% of worldwide chips in the 64K chips in six years. Midway through, U.S. vendors got very concerned, even going to Washington DC to seek "government relief" --> leading eventually to MCC (1982) and Sematech (1986) national initiatives. Bob Noyce, co-founder of Intel, left that company to head Sematech.
2. Dick Anderson, HP's Computer Systems VP, delivered the 'Anderson Bombshell' in Washington DC in March 1980, describing the twin facts that Japanese 16K memory chips were not only cheaper (and thought to be "dumping" in America), but 1000% more likely to pass incoming inspection, and lasted 500% longer in operation. Intel and TI both viewed this as "sour grapes" from a competitor even though HP was the largest memory chip purchaser in the world at the time. HP's BPC microprocessor (16-bit micro) had shipped 77,000 units to customers before Intel could source 8086 production units to anyone; TI and HP had been locked in a nonsensical price war initiated by TI over handheld calculators. TI, infringing heavily on HP patents, threatened to withhold memory chips if HP sued. Bad blood flowed.
3. Intel's Andy Grove capitulates to the Japanese in 1986, taking Intel out of memories and into microprocessors. He still describes this as "an epiphany." All other U.S. manufacturers bail out as well. The memory chip market today (2008) is an $18 Billion worldwide market, with no U.S. manufacturers.

No one, to our knowledge, has until now made the connection between the HP semiconductor test sets and the radically better processing technology that Hitachi and others then pioneered. The incredible 'rest of the story' is that Y-HP tried for years to sell its testers to U.S. vendors, with no success; consequently, it was virtually HP's only division with radically higher sales in Japan than anywhere else in the world. Palo Alto management for years associated this with bad marketing skills at Y-HP in contrast to NIH within U.S. semiconductor firms.


David Pitkin said...

Sorry for my ignorance but what is Y-HP and NIH?

chuck said...

Y-HP is Yokagawa Hewlett-Packard, a joint venture that existed for thirty years. NIH is "Not Invented Here". Sorry for the acronyms...