Thoughts about the Art Fong narrative memoir,
Chuck House, April 3, 2012

Art's wonderfully droll manner is entirely too modest in this memoir. He omits entirely his years in Corporate Engineering, both with Eb Rechtin (tough years) and with my group. Those years featured Art giving priceless advice for young designers around Thermal Design requirements, RFI and EMI emissions, and other exotica as the "new digital intelligensia" viewed them - things that "old microwave engineers took for granted" but no longer were taught. Things that today affect whether or not your cellphone works, your tablet melts, or your GPS finds the right locale, for example.

But maybe more importantly, in terms of impact on the world, he is entirely self-effacing about the impact of his work at Y/HP, work which ironically resulted in Japan supplanting Silicon Valley in silicon chips as a direct outgrowth of Art's inspired work. It is a story never before told, although I hinted at it in the book that Ray Price and I wrote three years ago. Subsequently, we have gotten detailed sales records as well as corroborating interviews with many key individuals. The excerpt and then the story is herein included:

Excerpt from THE HP PHENOMENON, pp. 135-136

Hewlett was eager for involvement in Japan as well as Europe, especially given his tour there after World War II, when he had the opportunity to meet and work with a number of accomplished scientists. The work of two men in particular would prove propitious for the company in subsequent years - founders Takeshi Mitarai of Canon and Shozo Yokogawa of Yokogawa Electric Works (YEW). In 1961, Shozo Yokogawa sought a partner who could broaden his company’s product line. Yokogawa knew Hewlett from his early visit, and both felt that YEW, a quality producer of electrical appliances and instrumentation, would be an ideal partner for HP upon entering a relatively closed society.

The arrangement, a 51 percent YEW / 49 percent HP partnership because the Japanese Ministry at the time forbade foreign companies from owning a majority interest, was called YHP. The partnership began (analogous to Europe) with kit products being assembled, sold, and supported from Japan rather than from America. It gave HP a valuable entrée into that market. Japan began a similar charter focus on deep parametric measurements, in particular, the classic RLC (resistance, inductance (L), capacitance) electrical passive elements. By 1966, YHP had introduced the first of its RLC test sets, tools that eventually proved vital for the emerging semiconductor manufacturing field because they could dynamically monitor precise doping depths of specific injected chemicals.

Art Fong - Mr. Microwave - had gone back to Stanford for work in digital techniques and software technology in the late 1960s. Transferring to YHP for two years in 1971, he helped the division capture the digitizing wave. The line started slowly, with $3.7 million revenues in 1972, but it burgeoned to a stunning $65 million by 1984 - a very satisfying 27 percent compound average growth rate, the highest for any non-U.S.-based division.

YHP recorded 40 percent of its sales in Japan in 1979, stunningly higher than the typical 5 percent for most of the corporation’s product lines. Some felt this concentration to be the result of parochial divisional perspective; YHP argued that it was serving local needs very well. Indeed, many local divisions exhibited regionalized success that differed widely and somewhat inexplicably from the corporate averages.

The bare bones story here is that William Shockley started semiconductor activity in Mountain View in 1957, but his domineering and paranoid style sent the 'traitorous eight' off to found Fairchild Semiconductor within the year. Over the course of the next 20 years, those eight former employees created 65 new enterprises in the area. Silicon Valley was "named" in 1971, and by 1975 had over forty manufacturers of silicon chips, plus twice as many more building support tools, in the Santa Clara Valley.

As Wikipedia reports it on April 2, 2012:

Intel’s 1103 (1024x1) became the first commercially-available DRAM memory in October 1970, despite initial problems with low yield until the fifth revision of the masks. The first DRAM with multiplexed row and column address lines was the Mostek MK4096 (4096x1), introduced in 1973. The Mostek MK4116 16K DRAM, introduced in 1976, achieved greater than 75% worldwide DRAM market share. However, as density increased to 64K in the early 80s, Mostek was overtaken by Japanese DRAM manufacturers selling higher quality DRAMs at below-cost prices^{[citation needed]}.(The 'missing citation' in the Wikipedia article is interesting in itself. The facts are that Mostek was beaten by 1979, not in the 'early 80’s', by Hitachi design, quality and true cost, despite what the U.S. industry later called 'dumping' in Washington D.C.)

Strangely, all of the U.S. companies in 1974 ignored the new parametric test set that Y/HP invented with Art’s leadership; in desperation, Y/HP turned to Japanese companies, virtually none of whom were worthy competitors on the world stage. Because of the fine-grain quality control that this test set allowed on diffusion depths, semiconductor yields and consistency were dramatically improved. Within three years, Japanese companies matched the early lead of Mostek (a Texas company) in 16K DRAM, and beat all U.S.companies to producible 64K DRAM in 1981, 128K in 1983, and 256K in 1985. By 1985, when Japanese 256K DRAMs almost bankrupted Intel Corporation, it was abundantly clear - only a handful of Silicon Valley companies still built chips in the Valley.

What was more remarkable was that HP had a lot to do with all of this, and in fact 'knew' it, but none of the players recognized how interconnected this all was to Fong’s invention. The key testimony was that of Richard Anderson, HP’s VP of Computing, in Washington D.C. in February 1980 - words that the later Trade Commissions chose to ignore, as did most U.S. chip vendors for another five years.

Here is the Excerpt from THE HP PHENOMENON, pp. 220-222

More importantly, HP’s quality emphasis was highlighted in 1980 in a speech by Dick Anderson, by then a Computer Systems vice president in charge of the HP 1000 and 2000 lines of computers.

At a forum in Washington, D .C., his topic was the stunning difference in quality between Japanese and American vendors of integrated circuits. American vendors, notably Intel, were at first enraged, then sheepishly apologetic, and then energized into reaction and repair.

By 1980, Japan had totally dispelled the notion that Japanese goods were cheap and shoddy. Instead, in a remarkably short time, "Made in Japan" came to mean the world standard of quality for products ranging from steel, automobiles, and heavy machinery to cameras, scientific instruments, and electronic gear. For American and European firms that used to control world markets in these fields, the transformation was calamitous. In semiconductors, where virtually all of the basic technology was created in America for transistors, integrated circuits, and microprocessors, Japanese competitors had moved up so far and so fast that U.S. firms went to Washington for help. The Semiconductor Industry Association, a leading U.S. electronics trade group, asked the federal government to help combat unfair Japanese competition. They argued that predatory trade and pricing practices, together with massive Japanese government assistance, were responsible for the rapid growth of the semiconductor industry in Japan. T. R. Reid, in an American semiconductor industry exposé, wrote an article titled "Meet Dr. Deming, Corporate America’s Newest Guru," in which he cited Anderson’s speech:

"The Japanese moved ahead in semiconductor sales at least in part because they moved ahead in semiconductor quality. Evidence was set forth in an infamous - or famous, depending on your point of view - meeting of electronics industry executives in Washington, D.C., in March 1980. Among the papers presented was one known today as "The Anderson Bombshell." Richard W. Anderson, an executive at Hewlett-Packard, spoke on his company’s experience with Japanese electronic devices. He explained that HP began in 1977 to produce a computer that used Random-Access Memory (RAM) chips which store 16,000 digits of information. Such memory chips - known in computerese as "16K RAMs" - were invented in the USA. But American suppliers could not produce enough acceptable 16K RAM chips to meet HP’s needs, so the firm somewhat reluctantly turned to Japanese manufacturers as an additional source of supply. Hewlett-Packard gradually began to realize that there was a significant difference in the Japanese memory chips. They were far more likely than the American product to pass HP’s standard factory inspection. And the computers in which they were installed went much longer without a memory failure than did comparable machines using American memory chips. In sum, Anderson said Japanese firms were producing higher-quality chips at about the same price as the American product.

Reid’s story referenced Dr. Edward Deming, but it didn’t do justice to the emotional situation. First, for a decade HP had been the top purchaserof memory chips from American semiconductor manufacturers - Mostek, Texas Instruments, Motorola, and Intel - along with microprocessor chips for the handheld and desktop calculator lines. Second, the higher-quality Japanese chips were more than 1,000 percent better at initial inspection, and 500 percent better over time - not just a little bit, but a whole lot better! Given the raw TI patent infringements and the angry feelings between the two companies in the handheld calculator business, and the head-to-head competition between HP’s Fort Collins NMOS II shop and Intel to build more capable microprocessors, Anderson’s criticism had all the earmarks of a vendetta rather than an objective, disinterested, third-party report. The difficulty was the data. And the data traced to that fundamental work in corporate quality assurance - work not done with nearly the same rigor at any other U.S. computer manufacturer. But then, virtually every other computer manufacturer was building equipment that, at best, met Class C environmental specifications - meaning that they worked only in a restricted temperature and controlled humidity environment. HP built its controllers and all succeeding computing equipment for Class B."

Click here to go back