Sunday, March 25, 2012

Crossbar, Inc. - a memresistor startup

Crossbar, Inc. is the latest start-up company focusing on resistive memory (ReRAM) as a replacement for Flash, DRAM, and SRAM. The company was formed based on research conducted at the University of Michigan led by Dr. Wei Lu (link) and has begun filing patent applications over the past few years for new forms of ReRAM based on silicon materials. Some pros and cons of this company are as follows:


1) Several major industry players (HP, Hynix, Samsung, Sharp, Sandisk 3D, Toshiba, Panasonic, Rambus) appear to be getting behind ReRAM and building an infrastructure and support system for ReRAM development as a replacement for Flash, DRAM, and SRAM. If successful this will certainly benefit start-ups in the ReRAM space by generating market acceptance.

2) Silicon is the most common semiconductor material used in industry. Gaining a strong patent position for silicon-based ReRAM may give Crossbar, Inc. an advantage with semiconductor memory manufacturing companies who want to stick with silicon instead of switching to metal oxide or chalcogenide materials used in other forms of ReRAM.

3) Dr. Wei Lu has some connection with researchers at HRL Labs who have been funded by DARPA in the SyNAPSE project to develop neuromorphic circuitry based on memresistor electronics (link). This relationship could lead to spin off applications outside of the semiconductor memory market in areas where there is more potential for exponential growth.


1)  There is no evidence yet of a strong management team for Crossbar Inc. A lack of experience and relationships within the semiconductor memory field could make deal-making difficult especially  with much larger companies.

2) The value of the patent portfolio for Crossbar, Inc. is unclear. Most of their patent applications have not yet been examined and some earlier examples of silicon-based ReRAM exist such as discussed in US 7345295 issued to Infineon Technologies. This may severely limit the scope of the claims for Crossbar, Inc.'s patents.

3) Dr. Wei Lu has foolishly adopted the "memristor" idea of Leon Chua in describing his technology. While this may seem beneficial in the short term to attract attention the actual scientific merit of Chua's memristor is highly questionable and is based more on the reputations of Chua and the researchers at HP than on legitimate reproducible evidence (link). In the long term companies such as HP and Crossbar that improperly rely on Chua's theory for attention may weaken the legitimacy of their own technology by association.