You are here

Investor's Business Daily Viewpoint: Modest Recovery Setting Up Next Semiconductor Boom | Cypress Semiconductor

Investor's Business Daily Viewpoint: Modest Recovery Setting Up Next Semiconductor Boom

Last Updated: 
Sep 03, 2009

Investor's Business Daily Viewpoint: Modest Recovery Setting Up Next Semiconductor Boom

T.J. Rodgers' Viewpoint on the Semiconductor Boom published in the September 24, 2003 issue of Investor's Business Daily.
Technology investors are apprehensive about the 60% run-up in the Nasdaq index since October 2002.
Some analysts forecast a correction, citing examples of stock appreciation at datacom companies that still are losing money and mixed high-tech corporate performance. But one must look through the anecdotes -- the continued datacom slump, the resurgence of the PC, the erratic growth of cell phones or the emergence of electronic consumer products -- to assess the health of the fundamental silicon technology that supports those disparate end markets.
From that perspective, semiconductors have clearly been in an upturn for all of 2003.
The silicon business has been the only common thread in the otherwise chaotic technology end markets since my arrival in Silicon Valley in 1970.
In those days, the military consumed 20% of semiconductors. Today, military sales are under 3%. By the mid-1970s, ups and downs in selling mainframe computer components to IBM and the BUNCH (Burroughs, Univac, NCR, Control Data and Honeywell) created the business cycles.
When we launched Cypress Semiconductor in 1982, our first big order came from Digital Equipment Corp. (DEC), the prime mover of the minicomputer, which took share from the mainframe business and the BUNCH.
(Today, the BUNCH is either out of the computer business, or plain out of business.)
Cisco And Cell Phones
Later, in 1989, we manufactured the processor used for network computing in Sun Microsystems workstations, which, along with the PC, killed DEC's minicomputer business --and DEC. The boom of 1995 was fueled by PCs, which, in turn, became powerful enough to cannibalize Sun's workstation business.
The boom of 2000 was fueled not by personal computers, but by communications: routers from Cisco Systems and cell phones from Nokia and Motorola.
Obviously, various electronic end markets are up and down, sometimes up and out, but silicon, the "crude oil of electronics," as Jerry Sanders, founder of chip maker AMD, put it, enjoys every up cycle.
Indeed, silicon doesn't just enable new electronic technologies; it eventually gobbles them up. Today, any competitive silicon company can make a wire for 4 microcents and a transistor for 2 microcents.
With a cost structure like that, our industry can dominate any market in which it has the end-market knowledge to compete. Intel started as a supplier of components for computers; it's now the world's most important computer maker.
In the plenary address at the 50th International Solid State Circuits Conference, Gordon Moore restated what most of us in Silicon Valley know: Moore's law has ahead of it at least a decade of progress -- that's four technology generations -- and another factor of 20 in cost. If our competitors don't like the 2 microcent transistor, they really won't like the 0.1-microcent transistor.
The last two semiconductor cycles, which included two terrible downturns, had amazingly similar phases that followed this pattern:
** A brief unit-volume crash, precipitated by end-market demand (PCs in 1995 and datacom/Internet in 2000).
**  A one-year period of low volume and severe price erosion.
**  A one-year pseudo-recovery with rising unit volumes and low prices.
**  A second crash inflicted on a weak semiconductor sector by some external event (the 1998 Asian currency crisis and the Enron scandal plus the 2002 recession).
**  A mild recovery with high unit volumes and low prices.
**  A boom with very high unit volume, product shortages and slightly rising prices.
Our company is now in phase five: We will ship 150 million units this quarter, approaching our all-time 2000 peak unit volume of 157 million units, albeit at only 60% of our peak revenue due to low prices. The aggregate semiconductor industry unit volume is currently down only 8% from its all-time peak.
During the 1996 down cycle, our statisticians analyzed the cycles of our share price, and the results were published to shareholders. ("Thinking About Cypress Stock" is available at under "Investors.")
In that analysis, we found the price-sales ratio of our stock, and that of many semiconductor firms, was the best indicator of stock performance over a long period (10-plus years) and multiple business cycles.
Over the cycles, the P-S ratio has a floor that reflects share price when a company is losing money or its markets are in a downturn, or both.
There is also a peak P-S ratio that characterizes good times." In December 2002, we updated the 1996 analysis, which accurately predicted the floor and peak P-S ratios for six subsequent years: 1997-2002.
Since 1986, we have had four share price peaks that were separated by 1,214, 1,363, and 1,491 days. We are currently 1,204 days into the fourth cycle.
The timing and magnitude of the share price peaks fall almost perfectly on a straight line. Other semiconductor companies and our industry also show similar, highly nonrandom cyclicality.

After the PC-based semiconductor boom of 1995-96, one of our directors speculated that the 1995 boom was so big, our industry would settle into a mature period of slower growth.
He was wrong.

None of us knew in 1996 that two new voracious silicon consumers -- the Internet and the cell phone -- were about to add their demand to the flat, but still significant PC market demand for silicon. In the current upturn, we probably won't need a new "killer application" like cell phones for a robust recovery.

The modest recovery of 2003 so far is not a result of a rebound in the still dormant datacom market. Over the last two years, consumer electronics has filled in much of the hole left by the collapse.

Focus On Coming Upturn
For example, cameras now have electronic retinas, and even "film" is made from silicon. Home video games have computers and graphic displays that rival those of the supercomputer of only a few years ago.
I predict the modest recovery of 2003 will turn into the boom of 2004-05 after the datacom market returns to even modest consumption.
The era promises of Internet growth were exaggerated, not wrong. Internet traffic, which drives a plethora of silicon-based equipment, has increased every quarter, but invisibly, because the new equipment installed has come from warehouses, not factories.
Sometime in the first half of 2004, the slack in that rope will play out, tightening the connection between semiconductor supply and demand. In this technology upturn, we stand at the middle of the beginning, not at the beginning of the end.
T. J. Rodgers is president and CEO of Cypress Semiconductor.