We have been keeping an eye on developments in non-volatile memory for a while, and this week a startup called Crossbar came to our attention, coming out of stealth mode to unveil its resistive RAM technology, having passed the theoretical stage with an example already in silicon.
RRAM is a new type of non-volatile memory which has the potential to completely disrupt the current status quo, and seems at the moment to be the most likely successor to NAND. The current NAND market, worth $20 billion, could become obsolete if the technology is brought to market, if it is all it claims to be.
The Crossbar technology is based on a back-end-of-line stackable crossbar that
can be put on top of standard CMOS. The stacking means a lot of memory can fit on a tiny amount of silicon, or as Crossbar says ‘up to 1TB of storage on a single chip’. It says that the technology will deliver 20 times faster write performance, 20 times lower power consumption, and 10 times the endurance at half the die size compared to NAND, which is proving difficult to scale below 20 nanometer geometry, with a predicted absolute maximum limit of 7 nanometers.
Crossbar, whose backers include Kleiner Perkins Caufield & Byers, claims to be the exclusive holder of RRAM patents, from the University of Michigan, and has filed for 100 patents, with 30 already issued. It is not, however, first to the party in the RRAM market. Rambus ‘acqui-hired’ a company called Unity Semiconductor in February 2012, to ‘continue developing innovations and solutions for next-generation non-volatile memory,’ and Panasonic announced only last week that it has started mass production of microcomputers with mounted RRAM, to be used in portable healthcare, security, and sensor products (at a rate of 1 million a month). Along with these, Elpida, IMEC, Micron, Samsung, Sharp, Sandisk, SK Hynix, and others, have all been working on their own iterations of RRAM.
But it is the CMOS compatibility which is the strong suit of Crossbar, as it allows easy System-on-Chip integration. Crossbar highlights possible uses in consumer electronics, enterprise storage, solid state drives and cloud computing, and low power requirements make it suitable for the Internet of Things, wearable computing, not to mention secure payments because Crossbar reckons that it is easier to embed secrets into the chip, than into Flash. But at this stage, despite having demonstrated the product in a commercial fab, there are many more hurdles, not to mention competitors, which Crossbar will have to overcome. We expect to see the company going to the market for more funding in the not too distant future, and perhaps being acquired by a larger player that can foot the bill of commercialization.