TSMC has responded to GlobalFoundries accusations of patents infringements. The world’s largest foundry said that it would defend itself in courts and that it considered allegations as baseless. The contract maker of semiconductors said that throughout its history it was granted 37,000 patents and naturally considers itself one of the leaders in the industry.

On Monday GlobalFoundries said that TSMC, a number of its customers, as well as makers of various products infringed 16 of its patents covering various aspects of chip manufacturing. In particular, GlobalFoundries claims that TSMC’s 7 nm, 10 nm, 12 nm, 16 nm, and 28 nm nodes illegally use its intellectual property. Among defendants, the company named Apple, Broadcom, Mediatek, NVIDIA, Qualcomm, Xilinx and many others. GlobalFoundries seeks damages from TSMC and wants courts to ban shipments of products that use infringing semiconductors into the USA and Germany.

GlobalFoundries vs. TSMC et al
Fabless Chip Designers Consumer Product Manufacturers Electronic Component Distributors

Quite naturally, TSMC denies any allegations and claims that it will defend itself in courts. The company stresses that it spends billions of dollars on R&D and has been granted 37,000 patents worldwide. Typically, high-tech companies counter-sue each other in patent infringement cases, so it will not be surprising if TSMC decides to sue GlobalFoundries. In the end, this is what patents are for. Meanwhile, unlike GlobalFoundries, TSMC will unlikely sue fabless designers of semiconductors that use the former’s services to a large degree because the vast majority of chip developers are it slients.

The statement by TSMC reads as follows:

TSMC is in the process of reviewing the complaints filed by GlobalFoundries on August 26, but is confident that GlobalFoundries’ allegations are baseless. As a leading innovator, TSMC invests billions of dollars each year to independently develop its world-class, leading-edge semiconductor manufacturing technologies. As a result, TSMC has established one of the largest semiconductor portfolios with more than 37,000 patents worldwide and a top 10 ranking for US patent grants for 3 consecutive years since 2016. We are disappointed to see a foundry peer resort to meritless lawsuits instead of competing in the marketplace with technology. TSMC is proud of its technology leadership, manufacturing excellence, and unwavering commitment to customers. We will fight vigorously, using any and all options, to protect our proprietary technologies.

GlobalFoundries vs. TSMC et al, GF's Patents in the Cases
Title Patent No. Inventors
Bit Cell With Double Patterned Metal Layer Structures US 8,823,178 Juhan Kim, Mahbub Rashed
Semiconductor device with transistor local interconnects US 8,581,348 Mahbub Rashed, Steven Soss, Jongwook Kye, Irene Y. Lin, James Benjamin Gullette, Chinh Nguyen, Jeff Kim, Marc Tarabbia, Yuansheng Ma, Yunfei Deng, Rod Augur, Seung-Hyun Rhee, Scott Johnson, Subramani KengeriSuresh Venkatesan
Semiconductor device with transistor local interconnects US 9,355,910 Mahbub Rashed, Irene Y. Lin, Steven Soss, Jeff Kim, Chinh Nguyen, Marc Tarabbia, Scott Johnson, Subramani Kengeri, Suresh Venkatesan
Introduction of metal impurity to change workfunction of conductive electrodes US 7,425,497 Michael P. Chudzik, Bruce B. Doris, Supratik Guha, Rajarao Jammy, Vijay Narayanan, Vamsi K. Paruchuri, Yun Y. Wang,Keith Kwong Hon Wong
Semiconductor device having contact layer providing electrical connections US 8,598,633 Marc Tarabbia, James B. Gullette, Mahbub RashedDavid S. Doman, Irene Y. Lin, Ingolf Lorenz, Larry Ho, Chinh Nguyen, Jeff Kim, Jongwook Kye, Yuansheng MaYunfei Deng, Rod Augur, Seung-Hyun Rhee, Jason E. Stephens, Scott Johnson, Subramani Kengeri, Suresh Venkatesan
Method of forming a metal or metal nitride interface layer between silicon nitride and copper US 6,518,167 Lu You, Matthew S. Buynoski, Paul R. Besser, Jeremias D. Romero, Pin-Chin, Connie Wang, Minh Q. Tran
Structures of and methods and tools for forming in-situ metallic/dielectric caps for interconnects US 8,039,966 Chih-Chao Yang, Chao-Kun Hu
Introduction of metal impurity to change workfunction of conductive electrodes US 7,750,418 Michael P. Chudzik, Bruce B. Doris, Supratik Guha, Rajarao Jammy, Vijay Narayanan, Vamsi K. Paruchuri, Yun Y. Wang, Keith Kwong Hon Wong
Methods of forming FinFET devices with a shared gate structure US 8,936,986 Andy C. Wei, Dae Geun Yang
Semiconductor device with stressed fin sections US 8,912,603 Scott Luning, Frank Scott Johnson
Multiple dielectric FinFET structure and method US 7,378,357 William F. Clark, Jr., Edward J. Nowak
Bit cell with double patterned metal layer structures US 9,105,643 Juhan Kim, Mahbub Rashed
Complementary metal oxide semiconductor (CMOS) device having gate structures connected by a metal gate conductor US 9,082,877 Yue Liang, Dureseti Chidambarrao, Brian J. Greene, William K. Henson, Unoh Kwon, Shreesh Narasimha, and Xiaojun Yu
Hybrid contact structure with low aspect ratio contacts in a semiconductor device DE 102011002769 Kai Frohberg, Ralf Richter
Complementary transistors comprising high-k metal gate electrode structures and epitaxially formed semiconductor materials in the drain and source areas DE 102011004320 Gunda Beernink, Markus Lenski
Semiconductor device with transistor local interconnects DE 102012219375 Mahbub Rashed, Irene Y. Lin, Steven Soss, Jeff Kim, Chinh Nguyen, Marc Tarabbia, Scott Johnson, Subramani Kengeri, Suresh Venkatesan

Related Reading:

Source: TSMC

Comments Locked


View All Comments

  • Dr. Swag - Tuesday, August 27, 2019 - link

    New nodes allow for higher power efficiency and higher transistor density. There is definitely still demand for new nodes. Otherwise why would AMD move to 7nm if it didn't help them make better CPUs? Why are they also moving GPUs to 7nm? Why are qualcomm and apple moving to 7nm for their smartphone SoCs?
  • FreckledTrout - Tuesday, August 27, 2019 - link

    That is completely false. Smaller nodes do provide large benefits. Sure at some point this may stop but man you are way off base. When we get to true 5nm using a gate all around approach we will have another huge transistor density jump and you think the engineers will look around and say I have no idea what to do with all of this?

    You will see more chip integration like higher end APU's negating many from buying dedicated GPU's. Things like system memory brought on chip. You will see Apple an other make true SoC's with no need for any other chips.
  • FunBunny2 - Wednesday, August 28, 2019 - link

    "I have no idea what to do with all of this?"

    fact: they haven't made anything like the tsunami of demand from 1-2-3 which kicked off the PC generation. and without that demand, ever more transistors becomes a niche business. just read the recent reports on AT about various productions being slowed or halted. don't you think there's a reason for that? much, if not most, of the use of the transistor deluge in the last 15 years has been to bring off-chip function on chip (CPU). that 'innovation' can go only so far. the multi-processor/thread revolution has faded away, modulo a niche or two, just because there aren't a lot of embarrassingly parallel problems in (consumer) user space. what new function(s) do you expect to see in mid-level SoC implementations as a result of yet more transistor availability? what about for retail computers? if the foundries can get more CPU from a wafer with smaller nodes, what does that do to the tool makers? well, it decimates their business model of course. how many 450mm wafer lines are out there, decades after they were supposed to rule?

    and so on. John Barth wrote a great book (well, lots of them), "The End of the Road". again, without a clear and demonstrable use for smaller nodes not doable at current nodes...
  • name99 - Wednesday, August 28, 2019 - link

    You seem to believe that the PC industry (throttled by the dead hand of Intel and Microsoft) is the same thing as the semiconductor industry...

    Apple has engaged in PLENTY of innovation over the past few years. There are a variety of interesting ways to use transistors on an A12 SoC or in an Apple Watch, or even in Airpods.
    As for "tsunami of demand", every year about 7 times as many cell phones are sold as PCs.
    Phones, Wearables, Hearables, IoT, computational photography, smart cars, any of this sound familiar?

    The problem is not that " 'we' don't know what to do with all of this "; the problem is that Intel doesn't know WTF it is doing.
    'You think that's too harsh? Where are the NPUs on Intel SoCs? When will they get there? 2025?
    Where has Intel been in providing their portables with sensors? They want you to buy 2-in-1's that can split into a tablet, but an iPad has a range of sensors (location and orientation, fancy ISP for camera, cell modem, ...) whereas Intel has, oh yeah, gutted their modem division, doesn't offer GPS on their x86 SoCs, and no-one's ever heard of their ISP.

    Intel gave up on innovation about ten years ago. Everything since then has been same old same old. Even the areas where they want to claim they are innovating, everything they're doing comes from a company they purchased (Nervana, Movidius, Altera) because they didn't have the vision internally to understand that transistors could do more than just x86.
  • FunBunny2 - Thursday, August 29, 2019 - link

    "Apple has engaged in PLENTY of innovation over the past few years. "

    as Steve said with the iPhone debut (paraphrasing), it's just 3 existing functions stuffed into one chassis. that's innovation? moving off-chip function on-chip is innovation? VR has gone nowhere. AI is going nowhere. what, exactly, is in need of smaller nodes, and is something other than a toy or mere consolidation of existing function on-chip? I dare you.
  • Vitor - Wednesday, August 28, 2019 - link

    "Semiconductor device with transistor local interconnects" sounds generic and vague AF.

    "Semiconductor device having contact layer providing electrical connections" this one is so absurd that it becomes funny.

    "Introduction of metal impurity to change workfunction of conductive electrodes" so they want monpoly on combining different materials with different electrical properties.
  • Oxford Guy - Wednesday, August 28, 2019 - link

    I hope the comments system here is able to handle your analysis of TSMC's 37,000 patents.
  • Arnulf - Wednesday, August 28, 2019 - link

    What do TSMC's patents have to do with ridiculously absurd patents listed above?

    You're the only one jacking off to the 37000 figure so why don't you present your analysis to us (along with explanation why this matters in the first place, of course)?
  • Nebuzaradan - Wednesday, August 28, 2019 - link

    Why don't you read the entire patent rather than trying to paste in a single line and acting like you understand technology? Also you can argue that the patent may or may not have been "original enough" to be granted by the USPTO, but it's a fact that this is a patent held by GF. If so then the questions the courts will answer is 1. Was this patent infringed upon and 2. What are the financial damages by said infringement.

    Also all the folks in here yelling about how TSMC has 7nm and GF doesn't so it doesn't apply. Please keep in mind FINFET designs aren't size specific. If you make a device and shrink it, it's just a smaller version of the same device. It's still fair game.
  • errorr - Wednesday, August 28, 2019 - link

    Read the patents they are much more specific. For example the 3rd one you mentioned requires Halfnium and that the the process of choosing p or n happen in the same step as a conductive electrode instead of separate steps.

Log in

Don't have an account? Sign up now