The memristor (short for “memory resistor”) is poised to revolutionize computing. A memristor is a passive two-terminal electrical component that can function as a circuit element relating charge and magnetic flux linkage. When current flows in one direction through the device, the electrical resistance increases. When current flows in the opposite direction, the resistance decreases. When the current is stopped, the device retains the last resistance that it had, and when the flow of charge starts again, the circuit resistance matches what it was when it was last active.
In a nutshell, memristors are resistive devices with an inherent memory. It is this characteristic that makes the memristor an ideal electronic component for neuromorphic computer architectures such as physical neural networks (as opposed to strictly software based neural networks), synaptic devices and AI (Artificial Intelligence) chips.
The memristor has been referred to as the fourth passive electronic component with characteristics of “memristance” that distinguish it from the more familiar resistance, capacitance, and inductance. Professor Leon Chua first conceptualized the memristor in 1971. The first working memristor was not developed, however, until 2008, when HP reported its nano-scale memristor device.
Since 2008, HP and other companies, universities, and government agencies have announced additional memristor developments. Alongside these ongoing memristor refinements is a quiet race to patent the memristor and its many variations and applications.
The first true AI chip will likely include memristive electronic components. Thus, the companies or organizations that succeed in patenting key aspects of memristive devices will be key players in the burgeoning AI industry. A review of recently issued and pending patents confirms that HP is the most prolific memristor patenting entity. For example, a review of the USPTO (United States Patent & Trademark Office) patent database indicates that HP has 103 (as of May 1, 2018) memristor related patents (searching only for issued patents that claim a memristor in some capacity). HP’s 103 patents claim at least some aspect of a memristor.
Interestingly enough, some companies that have made major plays in the AI area lag in memristor related patenting activity. For example, IBM only has 5 memristor-related patents. The same is true for Intel, which also only has 5 memristor patents to date. Samsung has 6 memristor patents, while Google has yet to be issued a memristor patent.
While these numbers only relate to issued U.S. patents, and do not take into account pending applications, it is surprising that such large companies have very little in the way of issued memristor related patents to date. This stands in contrast to smaller companies and organizations (e.g., universities) that have managed to build memristor patent portfolios. For example, Boise State University has been issued 9 memristor patents.
A small company that has managed to develop an impressive AI portfolio and also market and sell memristor chips, is the Santa Fe-based AI startup “Knowm” and its sister IP holding company “KnowmTech, LLC”. To date KnowmTech has developed a patent portfolio of 38 AI related patents that includes 7 memristor-specific issued patents and additional pending memristor related patents.
The race to develop the AI chip involves not only continued research and development in the memristor area, but continued patenting activity. The companies that successfully build patent portfolios covering various memristor aspects will be the companies that eventually dominate the AI market.