Commentary: It's Time to Cash In Some Chips, Big Blue

By Spencer E. Ante

If IBM's (IBM ) new chief executive, Samuel J. Palmisano, had any remaining hopes of a honeymoon, they were dashed at a May 15 meeting with financial analysts. Palmisano told Wall Street his two biggest problems were the PC and hard-disk drive businesses. But the analysts ignored all that and grilled the CEO on what they consider IBM's biggest headache: semiconductors.

Wall Street is right: The global chip slump is dragging down IBM's bottom line. On Apr. 8, Big Blue admitted to investors that earnings would fall far short of expectations. The biggest culprit was the chip operation, known as the microelectronics group. Even though semiconductors account for only 5% of IBM's $86 billion in revenue, the high fixed costs of chip manufacturing mean any revenue slide pummels profits. In the first quarter, this division posted a $138 million pretax loss, according to Goldman, Sachs & Co.--on top of a $184 million loss for the second half of last year.

So as the chip slump drags on, with recovery unlikely before next year, what does IBM do? It pushes ahead with its biggest investment plan ever. This August, the company plans to reopen a state-of-the-art chip plant in East Fishkill, N.Y., that it has refurbished to the tune of $2.5 billion. That's on top of another $2.5 billion investment to expand and upgrade chip plant and equipment that was announced in October, 2000, at the height of the boom.

IBM execs have said that Fishkill will double to triple Big Blue's chip capacity over the next year as the plant ramps up to full production. When the CEO was asked at the May meeting what IBM was thinking, even Palmisano conceded: "We need to make some adjustments."

Just what those adjustments are is the subject of intense speculation. But while many experts believe he should slash chip capacity in old plants, Palmisano seems reluctant to make any big changes. Like Chairman Louis V. Gerstner Jr. before him, the new CEO believes IBM's innovative silicon technologies give its products an edge. What's more, any retrenchment would be a confession that the company's decades-long, multibillion-dollar semiconductor research investment isn't paying off. Palmisano's point about cutting-edge technology is well-taken. His mistake is using that to justify hanging on to unproductive plants.

Consider Big Blue's main chip factory, in Burlington, Vt. Its aging production lines are running at 30% to 50% utilization rates. Many analysts assumed IBM would sell the plant. But on May 21, IBM Senior Vice-President John E. Kelly told BusinessWeek that Burlington stays. "I'm going up Thursday to meet with my 700 top managers to tell them [the rumors] are not true," he confided.

Instead of slashing, IBM's plan is to make a series of minor cutbacks, while trying to lower its cost of capital through partnerships with chipmakers such as Sony, Samsung Group, and Infineon Technologies. And Palmisano hopes to turn Fishkill into a high-volume chip foundry, using new technologies to design and build advanced chips for its customers. "In the future, years ahead, this can be a tremendous opportunity," says Kelly.

But IBM may not have the luxury of waiting. Indeed, Palmisano's whole plan seems out of sync with industry realities. Cross-border partnerships aren't a new idea. And promising technologies IBM pioneered, such as copper wiring on chips, are no longer unique to IBM. As for foundries, the world is awash in contract-manufacturers--in Taiwan, Korea, and soon, China. "With current demand, there's no way they can fill this [Fishkill] plant," declares Mario Morales, vice-president for semiconductor research at International Data Corp.

Which brings us back to Burlington. Even though the company laid off 501 of 8,500 workers there, total head count in microelectronics is still at 20,000, unchanged from the end of 2000. Kelly admits that more cuts are necessary--but he insists that "it's not massive portions of the place."

Whether it's Burlington or other plants with aging technology, IBM needs to downsize its least profitable chip assets. And Palmisano and Gerstner already have a model for that. In April, they announced plans to sell IBM's hard-disk drive business to a joint-venture with Hitachi Ltd. By any reasonable calculation, Big Blue's old chip lines should be next.

Ante covers IBM from New York.

    Before it's here, it's on the Bloomberg Terminal.