IBM, GlobalFoundries and Creative Divestitures

By Charles King, Pund-IT®  June 23, 2021

Divesting or selling business assets happens pretty regularly in the technology industry. Why so? For tactical and strategic reasons bolstered by the evolutionary nature of tech products and markets. Simply put, things change quickly in IT, and vendors that fail to adapt place themselves and their businesses in jeopardy.

For over two decades, IBM has pursued creative divestitures and spinoffs more often than most of its peers, usually with positive results. But the recent notice of a legal action it has filed against GlobalFoundries which took over IBM’s Microelectronics assets and chip making processes in 2015 is a reminder of how even well-intentioned and constructed plans can fail. Let’s consider IBM’s history of business divestitures, its deal with GlobalFoundries, what appears to have gone wrong and its impact on both companies.

IBM’s creative divestitures and spinoffs

It is arguable that the seeds of IBM’s modern divestiture strategy were planted in 1991 when the company spun out its printer manufacturing organization into a new business, Lexmark. But it fully matured a decade or so later when, in 2002, IBM sold its hard disc drive (HDD) business to Hitachi, then in 2005 sold its PC organization to Lenovo.

These moves shocked many in the industry since IBM pioneered development in those areas, launching its first commercial HDDs in 1956 and its first home PCs in 1981. In fact, the company’s designs became so dominant that generations of desktop systems were initially known as “IBM-compatible” PCs. However, a common issue drove the sale of both businesses – component commoditization which made it difficult to differentiate solutions and to produce the margins IBM desired.

Similar issues underscored IBM’s sale of its Intel-based System x (x86) server solutions and organization to Lenovo in 2015. However, they also highlighted the strategic value of these deals to the seller. After divesting core business assets to Hitachi and Lenovo, IBM treated both companies as strategic partners and continued to source HDDs, PCs and x86-based products to deploy for its own needs and use in IBM solutions.

IBM and GlobalFoundries

In other words, at a certain point IBM determined that it was wiser to be the customer of trusted partners than to continue developing and manufacturing certain products. That approach was clearly top of mind in IBM’s 2015 sale of its Microelectronics BU and assets to GlobalFoundries. I won’t go deeply into the technological aspects of the deal here but will focus instead on the strategic value of the sale to both companies, and how that appears to have gone off the rails.

To begin, why did IBM decide to sell its semiconductor and chip fab assets? Mainly because upgrading its aging facilities to support next generation 14nm and 10nm semiconductor processes for the company’s high-performance POWER and IBM Z mainframe systems was cost prohibitive.

In essence, IBM saw two choices: First, engage an established high performance chip maker to supply its needs and then sell or shutter most of its semiconductor assets and fabs (while keeping its chip design and silicon manufacturing R&D teams). Alternatively, engage an ambitious second-tier semiconductor maker and help it step-up to manufacturing high-performance chips, including those used in IBM’s homegrown solutions.

IBM chose the second option and, in GlobalFoundries, found what it believed was an enthusiastic and able strategic partner. While the company had little experience in high performance solution development, its chips commonly used in mobility, automotive, computing, wired connectivity, consumer internet of things (IoT) and industrial applications.

In addition, both companies were serious about the deal. As noted in IBM’s complaint, the proposed agreement was “No mere supply arrangement.” One senior representative of GlobalFoundries told IBM, this was “a marriage, not a date.” That same spirit is apparent in the contract’s generous terms and extended (10+ years) outlook.

As part of the final agreement, IBM divested and GlobalFoundries acquired IBM’s semiconductor business, including engineering teams and other employees, manufacturing facilities and intellectual properties (including over 30,000 patents). IBM also agreed to provide $1.5B in additional cash consideration to help GlobalFoundries transition to high performance chip manufacturing.

In turn, GlobalFoundries promised to use the $1.5B to fulfill its obligations under the agreement, including developing, manufacturing and supplying IBM with high performance semiconductors, specifically, 14nm and 10nm chips. As part of the relationship, GlobalFoundries would also serve as the sole-source supplier of chips for IBM servers for at least ten years.

Overall, the agreement appeared to provide essential value to both companies. IBM acquired an ambitious strategic partner that agreed chips for upcoming generations of IBM servers. GlobalFoundries acquired the tangible and financial assets necessary to step-up up to high performance chip manufacturing, as well as a solidly reliable customer for those new solutions.

When and how things went wrong

So, when and how did things go wrong enough to inspire a multi-billion-dollar legal action? According to IBM, the problems began shortly after the companies closed their deal in July 2015. In September, GlobalFoundries “began to indicate that it did not intend to develop, manufacture or supply” 10nm chips. In October 2015, IBM employees began to hear that GlobalFoundries was planning “to cease its work on 10nm technology and, perhaps, develop a different chip technology at 7nm.”

Not surprisingly, IBM executives requested that GlobalFoundries confirm its intentions and assure IBM that it would meet its contractual obligations. However, in a mid-December meeting between representatives of the companies, GlobalFoundries informed IBM that it had unilaterally decided not to develop, manufacture or supply the 10nm chips.

Though IBM refused to relieve GlobalFoundries of its obligations to deliver those chips, it supported the company’s 7nm development plan. During the following two years, IBM continued to provide GlobalFoundries the assets, support and cash payments contained in their agreement. Unfortunately, the company encountered considerable challenges in delivering volumes of workable 14nm chips for IBM’s new Power Systems and IBM Z servers, thus delaying the introduction of those solutions.

In December 2017, IBM delivered the final $250M payment of $1.5B in additional cash consideration. But early in 2018, IBM began hearing that another change in strategic direction was being planned by GlobalFoundries’ newly appointed CEO, Thomas Caulfield (who was VP of IBM’s 300mm Semiconductor Manufacturing group from 1989-2005). In July, IBM representatives requested a face-to-face meeting with Caulfield to discuss rumors surrounding the 7nm chips.

In its complaint, IBM notes that during that encounter, Caulfield suggested that GlobalFoundries “might go forward with the development of a 7nm high performance chip if IBM paid … another $1.5B.” IBM refused and shortly thereafter, “GlobalFoundries’ leadership effectively ceased communicating with IBM and refused to discuss its development plans or the future of the parties’ alliance.” In mid-August, GlobalFoundries announced that it would not proceed with development of its 7nm process technology.

Final analysis

Not surprisingly, GlobalFoundries has a significantly different view of the events and actions detailed in IBM’s complaint. In a countersuit against IBM, the company stated that “Development of the 7nm technology was far more challenging and expensive than had been anticipated, which caused delays in reaching the project’s targeted milestones.”

The company failed in attempts to find funding via bank loans or government grants and determined that spending its own assets would have resulted in significant losses and risked the company’s future. “With this backdrop, (GlobalFoundries) made the difficult, but prudent, decision to cut its losses.”

So where does that leave things? A few months after GlobalFoundries’ decision to exit 7nm process development, IBM announced that it had engaged Samsung to manufacture its upcoming 7nm-based POWER10 processors. If that partnership succeeds, it is likely the companies will continue working together on future IBM Z and POWER solutions. The company also continues to develop leading-edge, high-performance semiconductor technologies, including a recently announced 2nm technology that IBM says, “Can fit 50B transistors onto a chip the size of a fingernail.”

For its part, GlobalFoundries continues to make the mass-produced chips that it does best. Plus, it has sold off some of the assets it acquired from IBM, including the East Fishkill fab (to ON Semiconductor for $430M) and the ASIC business (to Marvell Semiconductor for $650M+). Finally, since the global semiconductor shortage has created a field day for silicon manufacturers of nearly every sort, the company is reportedly working with Morgan Stanley on an IPO that some believe will value GlobalFoundries at up to $30B.

If one were looking for lessons learned, the deal illustrates how even the most willing arrangements, generous terms and idealistic partnerships can fail. In some cases, that happens due to unforeseen events or circumstances, including unanticipated changes in course or strategy. In others, it occurs because one or more of the partners involved fails to keep promises or meet agreed to obligations.

IBM’s complaint succinctly sums-up its allegations: IBM depended on GlobalFoundries after investing heavily in a long-term mutual relationship. GlobalFoundries responded by taking IBM’s money, and benefitting from IBM’s knowledge, skill and assets. Though GlobalFoundries repeatedly assured IBM that it would meet its commitments, GlobalFoundries instead abruptly and without any justification walked away from IBM while IBM was reliant on GlobalFoundries. GlobalFoundries has demonstrably failed to act as a reliable partner and supplier.

Absent a negotiated settlement, the two companies’ claims, and counterclaims will need to be settled in court. But looking back at IBM’s past divestitures and the overall success of those deals both for the company and the strategic partners who received IBM’s assets and businesses, it is difficult to see GlobalFoundries’ acquisition of IBM’s Microelectronics business and semiconductor assets as anything besides an aberration or outlier.

In the end, a judge and jury will likely determine a just conclusion to the two companies’ disagreements. History, however, appears to be on IBM’s side.

© 2021 Pund-IT®. All rights reserved.