What’s Behind IBM’s Dutch-Treat Strategy

Think of the newly announced alliances among IBM/Siemens/Toshiba and Advanced Micro Devices/Fujitsu as proof that we have entered the Bob Hope era of semiconductors: “Thanks for the memories.” And, hey, it’s also a bit of a joke.

Just a few years ago, a single computer memory chip could hold slightly more than a quarter of a million pieces of information; today’s memory chips--formally known as dynamic random access memories, or DRAMs--can hold millions. But these technical improvements are expensive.

So IBM President Jack Kuehler and Siemens Chairman Karlheinz Kaske herald their joint venture with Toshiba to produce the next generation of advanced memory chips as essential to “global competitiveness” in this capital-intensive marketplace.

Even AMD Chairman Jerry Sanders--long one of the most insulting and vociferous critics of the Japanese electronics industry--now says, “I believe the world has changed and that globalism is the future.” So now his company will be co-producing different types of memory devices with Fujitsu.

Industry “experts” insist that the cost of producing these advanced memory chips is so high that not even giant multinationals such as Toshiba, IBM and Siemens can afford to foot the tab. Dutch-treat “global alliances” such as these will assure that U.S. companies can get access to these “critical components.”

What a bunch of hooey! The real issue here isn’t “globalism” or these companies’ inability to afford to build semiconductor plants; they’re simply tired of being in a technology race that none of them can profitably win.

Japan’s semiconductor manufacturers have lost billions in desperate memory-chip price wars to capture global market share. Consequently, the IBM trio is less concerned about competitiveness than defensiveness.

So while the new DRAM trinity estimates that its venture will consume at least $1 billion, it involves more than just pooling costs. IBM, Siemens and Toshiba can each comfortably afford that large an investment. The point is that they each now believe that the game isn’t worth the candle.

This move sends a clear signal to the marketplace that there is no competitive advantage for an individual company to massively invest in memory chips. This alliance effectively asserts that even as DRAM innovation grows more expensive, it has become more routinized and predictable.

Every 18 months, a new generation of memory chips emerges from the silicon chrysalis--supplanting its forbears. No wonder these companies have decided that cooperation is less expensive than competition. Proprietary technology has become meaningless here.

To be sure, this cooperation is built on the assumption that no technical discontinuities will disrupt the innovation cycle in the chip business. If some Tsukuba Science City start-up or CalTech graduate student devises a cost-effective way to make a personal computer terabit chip--a chip that could hold more than a trillion pieces of information--then all bets are off on these global joint ventures.

In this context, “globalism” and “competitiveness” are words full of sound and fury but signifying nothing. What’s happened is that the very process of high-tech innovation has become commoditized.

There seems to be no way for a company to keep a competitive advantage long enough to assure an adequate return on investment. When that happens, no intelligent company that cares about profit margins is prepared to pay a premium for incremental innovation.

You can bet that when it comes to semi-custom computer chips and microprocessors, there will be no gushy paeons from IBM or Toshiba to “globalism” and “cooperation,” because that’s where the competitive edge and the money is going to be. This strategic alliance is designed to staunch a wounded cash flow, not to boldly invent the future.

Please note that another “strategic alliance” was announced amid the shadows of the DRAM couplings and triplings. America’s other industry with “global competitiveness” problems--car manufacturers--revealed that they would cooperate on crash-test dummy and safety research.

“By joining efforts, we will eliminate redundancies in testing, save money and share the knowledge and experience of three companies,” one Chrysler executive said.

Let’s hear it for economies of scale. In fact, the auto companies understand that it would be difficult for any one of them to gain a proprietary advantage in auto safety without a public hue and cry to share their inventions with the competition. In other words, auto safety--like DRAMs--is not a domain where proprietary innovation can endure for long.

Increasingly, we’re going to see business alliances--some global, some not--in areas where companies simply don’t believe that they will get an adequate ROI--return on innovation.

Conversely, we will begin to see companies withdraw from partnerships in fields such as computer-aided engineering, biotechnology and materials sciences, when they sense the opportunity to capture both a proprietary position and profits.

In other words, whenever you see big companies announcing an alliance, recognize that they are paying far more attention to managing their costs than to creating their opportunities.