Why Would IBM Sell Its Semi Group?

Rumors are flying, but what’s behind all the hubbub, and more importantly, why now?


Rumors are always just rumors until proven otherwise in business, but in the case of IBM’s semiconductor business, hints about the sale of its semiconductor business are particularly noteworthy.

Much has changed since the days when IBM—as International Business Machines—went head-to-head with AT&T’s quasi-public Bell Labs and Xerox’s Palo Alto Research Center (PARC). The breakup of AT&T in 1982 and the subsequent spinoff of Bell Labs as part of Lucent, ultimately ended up in the sale of Lucent to Alcatel, Bell Labs patents and all. Xerox, meanwhile, spun off PARC in 2002.

Fortunately, Intel took up the slack in the interim, as its PC and server processor fortunes exploded. And since the Millennium, there has been a huge rise in research from around the globe. IMEC, the European powerhouse, has risen from near obscurity to become one of the most prominent sources of advanced semiconductor research. New York’s College of Nanoscale Science and Engineering has likewise emerged as a major research institution for advanced manufacturing.

Add to that the efforts of Sematech, SRC, Leti, as well as research sponsored by the Japanese, Korean and Taiwanese governments, and the picture looks a lot different than it did when a few U.S.-based goliaths were slugging it out. But the stakes also are a lot higher than they used to be. EUV lithography is a multi-billion-dollar research effort with no clear payoff; 450mm wafers are another multi-billion-dollar idea. And setting up a leading edge fab in the future will easily run more than $10 billion in equipment costs alone.

There is also highly advanced research underway inside universities around the globe into areas such as carbon nanotubes, graphene, spintronics, and new transistor and memory types—billions and billions more dollars worth.

IBM alone has a $6 billion research budget, at last count, and it has been at the forefront of every major development ranging from Air Gap to stacked die to deep channel doping. It invented multiprocessing in the 1960s, virtualization, the database, the PC, and it continues to lead in process research. In fact, look at any of the most advanced subject papers in semiconductor research and you’ll probably find at least one reference to IBM.

But unlike in the past, IBM is only one small piece of the growing research pie. Rumors are rumors, of course, and you can never put your trust in anything until there’s a legal contract and government signoff on the deal. But IBM has been at the forefront of all technology and has spotted trends well before the rest of the technology industry even got its first whiff of changes ahead. So if, indeed, IBM is considering charting a different course, the question is what does IBM see that the rest of the industry doesn’t?

Whether it’s a sale or a partnership, the implications of any move by IBM are worth watching carefully. This is a company that makes very deliberate moves with very precise timing, as evidenced by the sale of its PC division when there was still a market for big PC companies. So what, exactly, does IBM see now? And more important, what does it mean for everyone else?