SAN FRANCISCO—Taiwanese chip foundry United Microelectronics Corp. (UMC) has licensed technology from IBM Corp. to expedite the development of its 20-nm CMOS process, including FinFET 3-D transistors, the company said Friday (June 29).
Under the terms of the agreement, IBM will license its 20-nm process design kit and FinFET technology to UMC so the foundry can accelerate the availability of these processes for UMC customers, UMC (Hsinchu, Taiwan) said.
UMC is not joining the Common Process Platform—the IBM-led process technology alliance that includes Globalfoundries Inc. and Samsung Electronics Co. Ltd.—as the result of the license deal, a UMC spokesman said. The agreement covers only 20-nm and FinFET technologies, the spokesman said.
Though he declined to disclose the specific cost of the license deal, the spokesman described the terms as "reasonable and insignificant to our financial statements after expense amortization." The deal will enable UMC to save the considerable costs that would have been involved in developing 20-nm and FinFETs on its own, the spokesman.
The 20-nm and FinFET technologies will be offered as UMC processes, not as IBM processes, the spokesman said. The IBM license will enable UMC to expedite its own internally developed technology, he said.
UMC's internally developed 20-nm planar process will be aligned to IBM's design rules and process/device targets, while UMC's FinFET will be offered as a low-power technology enhancement option for mobile computing and communication products, UMC said. Implementation will take place at UMC's R&D site in the southern Taiwan city of Tainan, UMC said.
FinFETs are three-dimensional, fin-based multi-gate transistors. Intel Corp. is currently in production of 22-nm devices featuring FinFETs—known by Intel as tri-gate transistors. Taiwan Semiconductor Manufacturing Co. Ltd., UMC's biggest rival in the foundry business, is not planning to implement FinFETs until the 14-nm node. Globalfoundries and Samsung are also planning to implement FinFETs at the 14-nm node.
"UMC’s position as a world-leading foundry involves timely introduction of leading-edge processes to enable next generation customer chip designs," said IC Chen, vice president of Advanced Technology Development at UMC, in a statement. "Leveraging IBM’s technology expertise to shorten our 20-nm and FinFET R&D cycle will create a win-win situation for UMC and our customers."
Do not underestimate IBM. IBM still does a great deal of fundamental research. No other semiconductor company has Capabilities as IBM research does. IBM Still generates more patents than any other Semiconductor company. And Intel or TSMC are never going to give away their core asset Process Tech, coz that is all they have, IBM makes lot of money from lot of different busineses
Dylan, any reason why TSMC, Globalfoundries and Samsung are postponing the implementation of FinFETs at the 14-nm node. ? Is it because they dont haven't developed the technology to produce the FinFET's at higher-nodes ?
This is just the way of telling everyone that UMC is not absent in the fancy leading edge technology development. Same as they joined the SEMATECH before (now is out).
Does this imply they are serious in developing into a full scale, manufacturable platform? I bet not.
I posed your question to UMC directly. Here is what UMC's spokesman said:
On not joining the Common Process Platform:
-As customer needs evolve, we wanted the flexibility to have all options open for future opportunities (including Common Process Platform, but others as well). This was a UMC decision.
On resources and support:
-Besides the design rules mentioned in the press, IBM will provide technical consultation, in person or as needed through teleconference. IBM will not send a dedicated engineering team to Taiwan.
-UMC does not publicy disclose technology roadmap information; however, we are on track to accommodate first wave customers’ 20nm product rollout schedule.
So let me see if I got this right - IBM's demonstrated ability to produce FinFETs at ANY node? Zero. Their income from said ability (or lack there of)? Greater than zero. Sweet deal if you can get it. I am still amazed at IBM's ability to derive license revenue on process technology at least a decade after they stopped being relevant to process technology.