Download the authoritative guide: Cloud Computing 2018: Using the Cloud to Transform Your Business
Under the leadership of CEO Sam Palmisano, IBM has pared down many segments it considered to be commodity businesses, including printers, hard drives and—most famously—its PC business.
Yet IBM remains a chipmaker, albeit a small one, and one analyst thinks it's time for that business to go as well.
Boris Petrov, managing partner of the Petrov Group and a former director of strategic marketing with chip fabricator Chartered Semiconductors, put forward the theory in a report originally for the Asian publication DigiTimes.
In his essay (available in PDF format) here, Petrov puts forward the theory that IBM wants to grow to become a $200 billion company and focus on its systems and "Smarter Planet" initiatives.
As it is now, the hardware business at IBM is the smallest part of its revenue, outside of its financial services arm, and accounts for about 10 percent of the company's total income.
IBM's services arms, on the other hand, represent more than two-thirds of its total revenue.
IBM would unload its fabs to Globalfoundries, the spinoff from AMD (NYSE: AMD) that includes AMD's old fabs in Germany. Globalfoundries is only a year old, but it has financial resources thanks to Advanced Technology Investment Company, the Abu Dhabi firm that basically bought the fabs off AMD to spin them off into a separate company.
ATIC also purchased Chartered Semiconductorand has merged its operations with Globalfoundries. Both Chartered and Globalfoundries are members of IBM's semiconductor technology alliance, which gives them access to advanced semiconductor manufacturing processes.
"With its fabrication facilities worldwide, and a foundation in complex processor design and manufacturing, Globalfoundries should be able to incorporate state-of-the-art support for IBM, drive business economies and ensure growth," Petrov wrote.
"Globalfoundries could be an ideal outlet for IBM's IC fabrication business, enabling it to sell a business that has not met financial performance requirements for years, and still providing it the depth, scope, and resources needed to not only provide manufacturing security for IBM but also further ensuring success of its foundry business," he added.
IBM would not exit the business of chip fabrication entirely, however.
"IBM will continue to do research—research and engineering services are IBM's strengths—not low cost manufacturing," Petrov wrote. He said IBM has invested in silicon-on-insulator process technology, eDRAM, through-silicon-via technology (TSV), and 3D silicon
"In my scenario, IBM will provide process and materials research [and the potential buyer for the fab] will provide common platform wafer foundry services,” he wrote.
An IBM spokesperson declined to comment, as did a spokesperson for Globalfoundries, who said "We don't comment on rumors and speculation on any of this because we get this all the time."
The idea sure makes sense to Nathan Brookwood, research fellow with Insight 64.
"I think he's absolutely right about how this plays out but it will be several years before you see anything happen here. IBM doesn't really want to be in the semiconductor business, but they never had a partner they could trust to give them the advanced semiconductor processes they need to maintain their lead in the systems business," Brookwood told InternetNews.com.
IBM really had no alternatives, Brookwood points out. The only firm to match IBM's process skills is Intel (NASDAQ: INTC), and it's not about to make chips for IBM, even if IBM is a huge Xeon customer. Texas Instruments is out of that business and TSMC isn't known for advanced process design. As it is, TSMC has earned a reputation for hobbling both AMD and Nvidia in their attempts to move to 40 nanometer parts.
But Global needs time and a track record before IBM will trust it with its mainframe and Power processors.
"Globalfoundries is well on its way to establishing itself as that partner," Brookwood said. "If that continues to play out, and in particular they need to get the fab in upstate New York running and demonstrate they can do 22 nanometer, then I could see IBM saying 'Now we have somebody whom we can trust in that regard, and here, take our fabs, please.'"