Project Description

What’s current in semiconductors—
can patent law keep pace?

Moore’s Law has long been the golden rule of growth in computing. Transistor density in integrated circuits has doubled roughly every two years for decades—and computing power has grown exponentially as a result. Today’s smart phone users are holding more than a billion transistors in their hands when they snap a picture, schedule a rideshare, or Google a restaurant recommendation.
To find out more about the future of semiconductors and computing, we asked Dr. Stanley Shanfield to provide some insight into increased computing power and the challenges this continual innovation presents to intellectual property law. For his work on semiconductor research, Stanley has been awarded the University of California Regents Award for Research, and he has participated as an expert witness in multiple intellectual property patent disputes.

What does Moore’s law have in store for the future?

Clearly, the number of transistors cannot continue to double every two years, but innovation will continue to expand—even amaze. “The drivers of innovation have always been speed, size, and cost, and that hasn’t changed,” Stanley said. Chipmakers continually find more inventive ways to squeeze additional computational power into smaller spaces. The finFET (a fin-shaped field-effect transistor) is an example of that innovation. FinFETs have been around commercially for about 10 years. In simple terms, it is a field effect transistor that resembles a tiny mountain range, where you can build transistors up and down, not just side-to-side, allowing for Moore’s Law-level increases to continue—FinFET devices have faster switching times and lower leakage current than older planar chip technology.
Increasingly, the leading edge of chip technology is being aimed at building more efficiency into parallel processing and artificial intelligence (AI). An example is cell phone GPS receiver chips, which determine location through hundreds and thousands of parallel, low-level calculations. Mass parallelization of chip-level computation also allows for increased AI opportunities. (To get answers fast and to learn fast, you need a lot of chips operating in parallel.) Essentially, some chip designs of the future will try to match what “neural networks” accomplish from a software perspective. The end results: “You’re going to see some very smart equipment in the next few years that will surprise everyone,” Stanley said.

How will intellectual property keep pace?

In the face of continuous innovation, intellectual property law faces some challenges. Patents, for example, are often issued based on design or on how something is made, such as a single transistor or the arrangement of sub-systems in a chip.
However, the highly integral functions that go into the design and manufacture of transistors or chips complicate intellectual property legal matters. For example, in chipmaking there are certain widely understood standards in the chip fabrication sequence or in circuit design that everyone used at the time a patent was issued – sometimes those standards end up incorporated in claim language. References that highlight such a feature might be hard to find because the standard was so well understood as a requirement, so a contextual understanding of the prior art at the time the patent was issued is needed.
To add another significant complication, there are thousands of steps involved in building chips, so a particular aspect, such as an insulating layer, that was part of the process may be difficult to detect, even with the best instrumentation. “Understanding the context of a patent requires a lot of knowledge about how these things are usually made. Additionally, the combination of claim language and the context of the specification is meant to be understood by a specialist and can be unfamiliar to even technical experts in closely related fields.” Stanley explained.
An issued patent might be proven invalid based on obviousness (prior art), but it may have been prior art that wasn’t apparent to the person getting the patent or to the reviewer at the time of application. In the last 10 years, the shift to inter partes reviews (IPR) helps alleviate some of the complications by providing an efficient alternative to traditional patent litigation and because the review board judges are much better equipped to understand the highly technical details on which many of these patents are based.
As new technologies such as chip level AI advance, such patent disputes can only become more complex, requiring a strong background in the entire manufacturing process as well as a solid understanding of past technologies, in order to provide an effective analysis in any individual case.
Dr. Stanley Shanfield is frequently called as an expert witness in intellectual property cases across the globe involving computer hardware, semiconductors, and electronics. He is a named inventor on numerous patents, the author of dozens of published articles, and the leader of scores of scientists in bringing groundbreaking new technologies from invention through manufacturing. He also has specialty training in US patent law.
About Rubin Anders
For almost 20 years, Rubin Anders has delivered scientific experts to support high-stakes litigation. Our research team is comprised of scientists and attorneys uniquely positioned to understand our clients’ needs, and we leverage our global network of more than 20,000 scientists to reach the expert best positioned to support each client’s success.