Globalfoundries’ Change In Strategy Pays Off

When a company goes through a management or major strategy change, it is often years before the results are apparent and often even longer before everyone notices. Tirias Research recently talked with Gary Patton, the

CTO of Globalfoundries, on the status of the changes at Globalfoundries and a strategic update on the company. Although Globalfoundries is not a public company, Gary was willing to share some financial information. For the first time in the company’s history, Globalfoundries is profitable even without the funds it received from recent fab divestitures to ON Semiconductor and Vanguard International Semiconductor.

When Globalfoundries spun off from AMD in 2009, the company hit the ground running by quickly acquiring Chartered Semiconductor and building a new facility in Malta, New York. With fabs on three continents, Globalfoundries quickly appeared to be building another powerhouse semiconductor foundry offering an alternative and much needed competition to TSMC. However, to keep pace with TSMC still required key intellectual property, packaging technology, design services, and process technology expertise. It acquired all this plus additional fab capacity and valuable talent with the acquisition of most of the remaining IBM semiconductor group in 2015.

Unfortunately, the transition from in-house manufacturing as an integrated device manufacturer (IDM) to a foundry services provider is not an easy one. Intel also struggled in its efforts to offer foundry services. In addition, Globalfoundries faced a changing semiconductor market. With advancements in semiconductor lithography slowing the pace of Moore’s Law and the rapidly increasing costs to manufacture on advanced process nodes, semiconductor companies have turned to leveraging older processes longer and using packaging technology as a new pillar of semiconductor innovation. This left only a handful of computing (processor, FPGA, and GPU) and mobile System-on-Chip (SoC) vendors as potential customers for the bleeding-edge manufacturing capacity. At the same time, Samsung emerged as a foundry powerhouse. As a result, Globalfoundries found itself in a competitive race with both TSMC and Samsung. TSMC is the well-established foundry leader and Samsung is a leading electronics conglomerate, while Globalfoundries was still very much in start-up mode and struggling to make a profit. Something had to give.

With Globalfoundries struggling, especially with the launch of a costly 7nm process, which required the very expensive EUV technology, the new CEO, Thomas Caulfield, made a shift the company’s strategy. The most visible change was to abandon the development of the 7nm process node in favor of licensing from someone like Samsung, a long-time partner of Globalfoundries, at some point in the future. In addition, Globalfoundries divested smaller/less profitable fabs in Singapore and Fishkill New York, the old IBM fab responsible for process development, Power processor manufacturing, and some custom Application-Specific Integrated Circuit (ASIC) manufacturing. What was less visible was the company’s continued focus on specialized process technology like the Silicon Germanium (SiGe) used in analog and RF (especially 5G) and Fully Depleted Silicon-on-Insulator (FDSOI) referred to as FDX by Globalfoundries. The company also continued to focus on packaging technology for 2D and 3D chip architectures. Globalfoundries recently announced a 3D test chip developed with Arm that enables up to 1 million 3D connections per mm² on the company’s 12n FinFET process. With this strategic shift, Globalfoundries has emerged as a foundry that has the ability to assist in the development and manufacture of highly integrated solutions for mobile and IoT, including sensors, connectivity, and power-efficient processing cores.

I have to admit that I was disappointed when I was initially told about the change in Globalfoundries’ 7nm strategy. It looked like the company was set to be a powerhouse. In addition, few companies succeed when they abandon the high-end of the market. We saw AMD retreat to the low end in PCs and servers only to lose market share and teeter on the brink of bankruptcy until the company stepped back and attacked the entire PC and server markers with a competitive new architecture. Likewise, MediaTek has struggled in mobile as it has relinquished the high-end of the smartphone market to Qualcomm and the OEMs with their own mobile SoC design teams, including Apple, Huawei, and Samsung. So, it was natural to doubt the future of Globalfoundries under this new strategy. However, the IoT and other applications outside of mobile and general computing (PCs and servers) have exponentially higher (100x or more) unit volumes with longer product lifecycles. In addition, with companies in every segment moving toward multi-chip mobile or chiplet packaging, there remain opportunities throughout all semiconductor applications. As an example, while AMD is using TSMC to manufacture its processor chiplets and graphics cores, it is using Globalfoundries to manufacture the I/O functions die on its new Ryzen and Epyc processors. The result is the transformation of Globalfoundries into a profitable company with unlimited potential, all in less than a year.

With its current and potential capacity at its current fabs, specialized process technology, and packaging technology, Globalfoundries is larger than all the other technology trailing semiconductor foundries while being well differentiated from its larger bleeding-edge competitors. And for the first time, Globalfoundries has a management team focused on success rather than just competition.

This is one of a series of discussions between Tirias Research and some of the technical leaders from around the high-tech industry. Refer to the Tirias Research link on Forbes for additional articles.