Size matters at dawn of 'lethal recovery'

SAN MATEO, Calif. — Peter Drucker, the management guru, maintains that a primary task of managers is to find the right size for a business. Never has that issue been more crucial to survival than in the electronics industry as it moves from catastrophe toward tentative recovery in the closing months of 2002.

But rather than a robust return of double-digit annual growth fueled by a market driver — the next PC, cell phone or Web — the immediate future holds a single-digit recovery with no clear driver. Instead, expect a gradual strengthening across a wide range of unglamorous existing markets.

At the same time, fundamental structural changes in the global industry are pulling the ground rules out from under habit-bound managers. The new landscape is treacherous, extracting the full price of missteps made during the recession and offering no bailouts. Survival will come only to those companies that understand their position and can make themselves the right size to serve their customers profitably.

This scenario foretells a plethora of mergers, acquisitions, layoffs and outright failures, as individual companies sense their positions and fight to achieve proper size and expense levels. It also foreshadows a landscape vastly changed from the explosive, U.S.-dominated electronics industry that this generation of investors had assumed was a birthright.

Costs are down and margins up across the industry, inventories are low and demand seems to be stable in many markets. But in several ways this recovery may be different from any in recent history.

Traditionally, the industry recovers from recession on the back of a market-driving technology. Once it was aerospace, then the PC. The buildout of the Internet and the global cell phone network teamed up to create the mid-'90s recovery that grew into the Web bubble. Each of these comebacks was marked by explosive growth of a few small companies that "got it," followed by a heavy wave of venture investment and then, finally, the gradual repositioning of the industry giants.

This time there is no obvious candidate for the next big thing. No existing technology has the combination of novel utility, wide appeal and lack of infrastructure or financial barriers that, taken together, would spark rapid growth. And with public, corporate and individual debt all at epic levels in the industrial world, the financial freedom to pay for the next big thing appears to be lacking, even if it should suddenly appear.

Instead, growth for a time may come solely from the incursion of electronics into other, slowly growing markets: automotive, energy, industrial and medical, for instance. Such a model would lead to growth rates in the single-digit range, with opportunities spread widely across the global economy and, consequently, with higher costs of sales.

Add to that scenario an accelerating shift in the geography of the industry. As the United States wallows in doldrums and Europe slowly grows, Japan is in crisis. "Sometimes Japan appears to have completely lost its way," Dataquest analyst Gary Smith said in a recent talk. South Korea, Singapore and to some extent Taiwan appear to be struggling with their deep dependence on North America.

But the People's Republic of China paints an entirely different picture. Deployment of cell phones and wireless data networks on the mainland is booming. China is buying — and building — consumer electronics at a prodigious rate. And it is pressing ahead with the construction of a large number of IC fabs. But the PRC has no intention of remaining just the world's low-cost manufacturer. It is also moving aggressively into design, setting up engineering shops, repatriating U.S.-trained engineers and graduating an estimated 10,000 EEs a year from its own campuses.

"China is ramping faster into complex design activity than either Japan did in the 1960s or Korea did in the 1980s," observed Wilf Corrigan, chairman and CEO of LSI Logic Corp. As recovery comes, U.S. companies will have plenty of competition for the new opportunities. They may lack any access whatsoever to the growing end markets in China.

The industry is also going through a technology change, a passage that's not unusual for the trough of a recession. Processes are moving through the 130-nanometer node and toward 90 nm. At this point, design issues like signal integrity, power concerns and process variations are forcing fundamental changes to the design flow. Just bolting on point tools won't be enough to make 90-nm systems-on-chip accessible to the average design team. As always in such a transition, the passage is marked by the wreckage of failed projects, the furor of intense work in the EDA community and speculation that the process engineers may have finally gone a node too far.

Meanwhile, the nonrecurring-engineering cost per gate in SoC design continues to decline. But today, virtually all the gates in a system may come from one design: a system-level chip. So all of the NRE and all of the risk come upfront to the organization that undertakes the SoC design. This is an expense load and a level of risk — at least several million dollars per chip design — that few organizations can handle alone.

The initial impact of this shift was masked by the Web bubble. With vast quantities of venture money pouring into anything vaguely network-related, even startups could put an SoC design in their business plan, spin it once or twice and still have money in the bank.Today, managers are becoming more circumspect. When a design fails, it can take a company with it.

These, then, are the parameters that delimit the necessary size for a venture in this strange recovery. On the one hand, if a company is serving a market that requires advanced SoC devices, the vendors simply must be large enough to accept the NRE and the financial risk. They must also have the channel clout to drive volumes great enough to recover their costs. A company not big enough to meet these criteria in major markets will be either a tech acquisition or road kill.

On the other hand, a company that cannot be a primary OEM in such a market is constrained by the same lack of market growth and absence of venture investment. Such companies must absolutely minimize their fixed-expense levels so that they reach break-even early and achieve positive return on investment at relatively low revenue growth. That means minimal front-end investment, minimal staff and no frills. Increasingly, every salary in small and moderate-size electronics companies must be traceable directly to revenue. And increasingly, exit strategies involve not initial public offerings at science-fiction multiples or even pre-emptive acquisitions by well-heeled competitors, but gradual growth to sustainability through a network of joint-development relationships, one of which finally leads to acquisition.

No middle ground

In this bleak scenario there is no middle ground for a big company with small revenues and great prospects. Intellectual-property vendors with royalty-weighted business models are at serious risk. Medium-size semiconductor companies — especially those with the fixed expense of an internal fab — must reform. But even moderate-size fabless companies must seriously ask if they are big enough to incur the expense and risk of continued new designs by themselves. If the answer is no, then they are not fabless semiconductor companies — they are soon-to-be IP vendors.

And as the avalanche of reorganizations in Japan indicate, even huge captive IC operations must face the same scrutiny. With low-cost manufacturing already having fled to Taiwan and soon to China — and with powerful alternative design centers existing in Europe, and emerging in China and India — the captive semiconductor operation must deliver a proprietary advantage measurable in market share and margins to survive. Alternatively, it must fall into the maelstrom that at its center has a few giant foundries producing the vast majority of the world's CMOS ICs.

See related chart