© The Financial Times Ltd 2015 FT and 'Financial Times' are trademarks of The Financial Times Ltd.
Last updated: February 7, 2014 7:09 pm
When Lou Gerstner arrived at IBM in the depths of a financial crisis in 1993, the first outsider to be named chief executive in the company’s century-long history faced an immediate choice: to break up an icon of Corporate America or try to save it.
A pragmatic, all-purpose CEO who had worked at RJR Nabisco and American Express, Mr Gerstner was not steeped in the complex technologies that had made IBM king of the mainframe computer era. But he knew a business opportunity when he saw one.
Big Blue, as the company was known, had seen demand for its monolithic mainframe machines eaten from below as cheaper minicomputers and then PCs had made their way into the business world.
For a timeline charting pivotal moments at Big Blue
In turning his back on a break-up, however, Mr Gerstner famously declared that the last thing IBM needed was a grand plan. He had seen the future: by combining hardware, software and services, IBM would be in the strongest position to sell full solutions for the IT headaches that were becoming a feature of complex modern corporations.
That decision has reverberated through IBM ever since. And it has now fallen to Ginni Rometty, who two years ago slipped into the chair once occupied by Mr Gerstner, to take it a step further towards its logical conclusion.
In a move that could see the company sever its ties with a core part of its traditional computer hardware business, the new IBM chief executive has begun to explore a possible sale of IBM’s semiconductor division, say people familiar with its plans.
Even if she stops short of a full sale but finds a buyer for only part of the unit, such as the manufacturing plants, it would amount to the most significant step in IBM’s strategic realignment in years, according to analysts.
Mr Gerstner’s original insight set in train a constant reshuffle of IBM’s portfolio of businesses as it has sought to keep ahead of changing technology trends. That has led to giant acquisitions and disposals, as one of the supertankers of the tech industry has taken years to steer away from computer hardware – a business where profits have become harder to make – and towards software and services.
But while continuing the plan of her predecessors, a sale by Ms Rometty of the semiconductor unit would represent a new departure. Previous disposals, such as last month’s sale of its low-cost server unit to Lenovo for $2.3bn, have involved markets where IBM no longer has a technology edge and profit margins have been squeezed.
The semiconductor business, by contrast, marks part of the high ground of the tech industry and accounts for a large portion of the intellectual property on which the company’s success has always depended. It has been losing money, however: Bernstein Research says it accounted for revenue of $1.75bn in 2013 but made $130m in pre-tax losses.
Silicon Valley dealmakers, traditionally a less frenzied breed than their peers on Wall Street, have had a busy start to the year, writes Ed Hammond.
News that IBM is considering the sale of its semiconductor business – a move that would mark a fundamental shift in direction for the technology group – caps a six week bout of activity.
Ms Rometty appears ready for something radical. She has seen investor confidence bleed away over the past year as IBM’s performance has slumped, ending a long run that had made Big Blue a favourite on Wall Street.
Chips and computer hardware have become a steadily smaller portion of IBM’s business, accounting for only 16 per cent of revenues and 11 per cent of operating profits last year. Also, it was a slide in hardware last year that accounted for much of the company’s earnings disappointment.
“Hardware is flat or declining and services are growing more modestly. Much of the growth is in software,” says Andrew Bartels, an analyst at Forrester Research. Selling the chip division and reinvesting the money in software would be a natural response, he added.
The costs of staying at the leading edge of semiconductor research and manufacturing have also risen inexorably. IBM’s main “fab”, or chip plant, in East Fishkill, New York, pictured right, has absorbed billions of dollars of capital investment as the company has invested in the larger wafer sizes needed to stay competitive.
The technology race to produce ever-smaller features on chips – needed to maintain the exponential advance in cost-efficiency in semiconductors known as Moore’s Law – has also made this too costly a game for all but a small handful of companies.
“It’s harder and harder to achieve that doubling of chip performance every 18 months,” says Mr Bartels.
Yet dropping out of the tech industry’s most demanding race could have far-reaching consequences. IBM’s main chip architecture, known as Power, is credited by some analysts with being superior in servers to Intel’s x86 – though it is the Intel architecture that has steadily invaded data centres in recent years.
Abandoning further development could rob IBM of some its biggest technology advantages, some analysts warn.
The success of Watson – the company’s “cognitive computing” system, which IBM hopes to use to bring a new form of question-and-answer technology to the business world – depends heavily on the company’s in-house chip knowhow, says Rick Doherty, an analyst at Envisioneering who has followed the company’s chip business closely.
As a result, some analysts say that it would be unthinkable for IBM to pull out of Power completely.
One alternative would be for the company to hang on to its research and chip design involving Power, while looking to sell only its “fabs”, or chip plants, and the research associated with the advanced manufacturing operations, says Patrick Moorhead, an independent chip analyst. Even this would be worth “billions of dollars”, he adds.
Such a move might reflect the position IBM now holds in the global semiconductor industry. “It’s not the powerhouse in chip manufacturing and design it once was. But it is still a leader in chip research and IP,” says Mr Moorhead.
The pool of potential buyers, whether for some or all of the division, is extremely small. Intel is considered unlikely given the gulf that exists between the two companies’ technologies. On the other hand, IBM already has close manufacturing ties and licensing arrangements with three of the four other companies involved in semiconductor manufacturing: Samsung, TSMC and Global Foundries.
Timeline: key moments at Big Blue
1964 IBM’s System/360 mainframe makes it the dominant force in the first era of corporate computing
1993 Lou Gerstner, right, arrives as chief executive in the midst of a financial crisis
1995 $3.5bn purchase of Lotus Development Corp lays the foundation for expansion of the software business
1996 In a seminal man versus machine moment, middle right, IBM’s Deep Blue computer beats world chess champion Garry Kasparov
2002 Acquisition of PwC consulting division for $3.5bn rounds out IBM’s move into business services
2002 Group sells hard drive business to Hitachi for $2bn
2005 Group sells PC division to Lenovo for $1.75bn
2008 IBM’s biggest acquisition, the $5bn purchase of business intelligence company Cognos, cements software’s role
Jan 2014 IBM sells commodity server business to Lenovo for $2.3bn; explores sale of semiconductor division
Please don't cut articles from FT.com and redistribute by email or post to the web.
Sign up for email briefings to stay up to date on topics you are interested in