More about the Silicon Valley legend

My friend Shane Greenstein this week wrote a book review of Moore’s Law, a biography of the famous Silicon Valley chemist, entrepreneur, billionaire, philanthropist and visionary. As Shane opened his Wall Street Journal commentary:

Fifty years ago, Gordon Moore formulated his famous “law,” typically summarized as “the number of transistors that can be placed on an integrated circuit will double every two years.” The accumulation of exponential improvements that he foresaw has indeed ushered in perpetual reductions in the cost of computing and the size of computers. And it’s at the core of the information technology revolution.

Even before his famous 1965 paper, Moore was a Silicon Valley pioneer, there when they put the silicon in Silicon Valley. After attending SJSU, Cal and Caltech, in 1956 Moore went to work for (later Nobelist) Bill Shockley, working on the technical challenges of adapting silicon (rather than germanium) to construct reliable semiconductor devices.

The next year, Moore and others of the Traitorous Eight set the pattern for the Valley when they Shockley Semiconductor to found Fairchild Semiconductor — the Valley’s first (unsanctioned) spinoff. In 1968, two of the eight — Moore and Robert Noyce — took Andy Grove to start Intel. (Four years later, fellow Traitor Gene Kleiner joined Tom Perkins to found the Valley’s legendary VC firm)..

Never fired and married to the same woman for 60+ years, Moore was neither a showman nor celebrity like the late Steve Jobs. Arguably, he and his compatriots did more than anyone else to set the pace of Silicon Valley (at least during the five decades when there was still silicon in Silicon Valley).

The authors of this latest biography — “a chemist, a historian and a journalist” — offer the definitive story of the 86-year-old chemist. As Shane concludes:

The authors have material on just about everything in Mr. Moore’s life—his relationships with his wife and children, the Porsche he drove, even his childhood adventures with nitroglycerin. The detail, on occasion, becomes overwhelming.

Yet the book brings an insider’s perspective into the discussion of Moore’s law. What became the law first emerged in 1965, in an article modestly titled “Cramming More Components Onto Integrated Circuits,” published in the journal Electronics. By the authors’ account, nobody paid much attention to it at the time. Why do we know it today? Mr. Moore revisited the idea in 1975, updated it and devoted public speeches to it. Others began to notice its deep foundations at the boundaries of science and production; Carver Mead, a longtime professor at Caltech, was the one who actually coined the term “Moore’s Law.”

Gordon Moore’s forecast was spectacularly right. Yet, as this compelling biography proves, even if he had never hazarded it, he would remain a legend in Silicon Valley.

Alas, it’s hard to deny the passing of this era. Moore is the only one of the Intel founders still alive, and one of only two (with Jay Last) of the Fairchild founders. Despite its continuing microprocessor monopoly, nearly 15 years later Intel stock is still only half its peak level of August 2000. Internet software — not electronics — is where Kleiner Perkins is making its money nowadays.

"Silicon" Valley gone after 50 years

In 1957, the Traitorous Eight left Shockley Semiconductor to form Fairchild. This is the first Silicon Valley spinoff, as well as the real start of the careers for Robert Noyce and Gordon Moore (Intel) and Gene Kleiner (Kleiner Perkins). The next year, Fairchild sold its first product, the 2N697 silicon transistor, and the rest as they say is history.

Saturday, the Merc reported that Intel is closing its last fab in Q3 2008.

The D2 line has been used to pilot production for fabs elsewhere in the US and the world. This is the last major semiconductor fab in Silicon Valley.

The end of the story has been obvious for a long time, even if the timing has not. Given housing costs, salaries, regulation, water and energy issues, manufacturing in the Bay Area has been heading for the exit for more than a decade.

The question raised by Merc readers (in the comment section) is whether Intel plans to eventually move all the R&D away as well. On the one hand, the Bay Area is the most expensive tech cluster in the country (perhaps second only to Tokyo in the world). There is a lot of unique software expertise here, but semiconductor expertise is more broadly dispersed.

Of course, Intel — like HP, IBM, Toyota, Ford, Sony and Nokia — will have R&D dispersed around the world. Some R&D is going to remain in the US, even if not in the valley. Could there be an Intel without R&D at its headquarters? I don’t know.

Nokia ending vertical integration?

This morning Nokia announced that it’s transferring 3G chipset technology and people to STMicro. The reports are somewhat vague — perhaps deliberately so on Nokia’s part — but it sounds like they are exiting the radio modem chipset business.

From the various accounts

• Nokia is licensing its 3G technology to ST and sending it 200 of its existing workers. That sounds like an exit to me.
• Nokia wants to have dual suppliers, and not be entirely dependent on TI for 3G chips. TI stock fell on the news.
• Nokia plans to license this technology to other firms to gain additional revenues.

On an unrelated note, Nokia is buying obsolete technologies from Broadcom (EDGE) and Infineon (GSM). Perhaps that’s a foot in the door for both firms and they will be able to provide 3G products in the future.

Once upon a time, several mobile phone companies were fully vertically integrated, providing end-to-end solutions of chips, phones and even infrastructure. Firms that notoriously used their own semiconductors were Samsung, Matsushita (Panasonic), Motorola and Nokia. (The articles also imply Ericsson but I’d never heard them mentioned). The move by Nokia — both to license out technology and procure chips from external suppliers — is a classic shift from vertical integration to open innovation, consistent with established principles of open innovation.

HSDPA is the first WCDMA technology that will actually deliver broadband speeds of 1 mbps or more, and was developed with input from Qualcomm and what it learned from developing the earlier EV-DO technology for cdma2000. I was struck by how difficult it is to make HSDPA chipsets, and thus the number of suppliers has been winnowed down tremendously from GSM.

Here are two quotes that capture this effect — classic Five Forces that I teach my strategy students. From the Dow Jones article:

Nokia on Wednesday also announced a move to license out its modem technology to chipset vendors in order to bring in additional revenue and allow new players to enter the market. The industry has traditionally had a high barrier to entry because of the complexity of the technology expertise required. Modems are the interfaces between chips and radio signals and act as the communications center of a chipset.

The Financial Times article was even more specific:

The decision to broaden its range of chipset suppliers and license proprietary technology for high-speed WCDMA/HSDPA mobile phone chips will be an opportunity for chip companies such as STMicroelectronics and Broadcom to enter a new market.

Currently, only Nokia, Ericsson Mobile Platforms and Qualcomm are believed to have technology to make cost-effective, WCDMA/HSDPA chips.

The FT exaggerates a little, in that so far only ST is getting a license to the coveted HSDPA technology. Still, under the rule of ”the enemy of my enemy is my friend,” Nokia has a natural ally in Broadcom, who’s been winning a series of patent fights with Qualcomm.

Is the scarcity of HSDPA vendors a transient issue tied to learning curves, or a permanent shrinking of the supply pool?On the one hand, GSM was once cutting edge and now has a wide range of suppliers. On the other hand, the x86 processor suppliers have certainly been winnowed down over the past 25 years. Adding to the uncertainty are all those WCDMA patents.

Technorati Tags: 3G, Nokia, open innovation, semiconductors, vertical integration