In November 2008, I had the chance of interviewing Dr. Saied Tehrani, Everspin's COO. Dr. Saied was instrumental in beginning the MRAM program - back in 1995. Prior to the formation of Everspin Technologies (from Freescale), Dr. Tehrani was director of Analog and Mixed Signal Technologies at Freescale. His R&D team was responsible for the development of power, analog, RF, sensor and MRAM technologies.
Can you explain the decision to spin off Everspin from Freescale?
The decision to form a new company was intended to create a focused entity that can commercialize on MRAM capabilities across the entire suite of new applications. Everspin and its investors believe that an entity focused on MRAM and integrated magnetics will be able to serve customers better and more rapidly expand the MRAM product offering and advance its technology development.
In July 2006, you introduced the first MRAM chip (4-Mbit). This was an amazing milestone for MRAM Tech. What are your own thoughts, 2 years after?
It has been exciting being part of the emergence of a new non-volatile memory technology into the marketplace. We are solving real customer problems by offering cost-effective, high performance and reliable non-volatile memory for broad market applications. In addition, we have seen very strong traction and market adoption of this technology in a very short time.
Can you update us on the current pricing of the MRAM chips?
We have aggressively and continuously reduced the MRAM pricing to increase its adoption in broader applications. We are currently the most cost-effective solution on the market and our pricing is very favorable compared to alternative lower performance and less reliable alternatives like battery -backed SRAM and nvRAM.
When do you plan to launch more MRAM products? Will these have more capacity?
We have just announced our new byte-wide 1Mb and 4Mb MRAM products and introduced our first MRAM in a 48-pin BGA package that reduces board area by three times. These products expand the markets that MRAM serves—specifically in the storage and consumer spaces. Going forward, we have a very aggressive roadmap that includes 16Mb products in 2009.
Can you give us some products that use your chips? We know of the Siemens touch-screen products and the ones in the Japanese satellite...??
We have production programs with industry leaders in the storage, industrial automation, energy management, communications, consumer, gaming, transportation and avionics markets. We are unable to disclose all their names at this time, but we are very happy that we were able to share our success stories with Siemens and Angstrom Aerospace. These are great examples of how MRAM is enabling our customers to provide better and differentiating products in the marketplace.
What are the plans forward for Everspin?
We have an aggressive product roadmap that expands our offering to serve the broader market space. Our recent announcement of the BGA package for the consumer market and byte-wide MRAMs for the storage market is part of our plan to make MRAM available to a broader customer base. We also have a very strong R&D program that allows us to further advance the technology for higher density stand-alone, embedded and magnetic sensors.
Now let's talk a bit about the technology...It seems like MRAM has lost most of the hype lately, with many companies discarding their MRAM research programs. Do you still hope that MRAM or STT-MRAM will be the 'universal memory'?
We have always had a very realistic expectation of what the technology can offer within a given time frame. We are very proud of what we have been able to accomplish with the technology to date. In 2006, we introduced highly reliable, high performance read/write, non-volatile memory in commercial temperature range; and then followed in 2007 with extended (-40 degrees C to +105 degrees C) temperature products. This year we have further expanded the Everspin portfolio with byte-wide products and BGA packages. We believe that we are still at the very early stages of capitalizing on the capability of MRAM technology for stand-alone, as well as embedded markets. MRAM clearly is showing universal characteristics of speed, non-volatility and reliability that is not available in any other memory today. With Everspin’s further advancement of the technology, we believe we can continue to increase the available densities and expand MRAM’s market penetration.
Most companies are working towards some sort of STT-MRAM these days (and not 'plain' MRAM). What do you think? Will you work on that tech as well?
Everspin is one of the pioneers in developing Spin-Torque MRAM from its early demonstration. We have a strong R&D program focused on spin-torque and see some exciting improvements that can be made to MRAM technology by utilizing spin-torque switching. However, the Everspin Toggle Mode MRAM technology was the breakthrough that made our MRAM commercially viable today, and this technology will continue for several generations.
Can you compare your own tech to the new STT-RAM by Crocus, Renesas, Grandis, STT, etc?
There are a number of different approaches that are being considered for future generations of MRAM—all of which are in the early phases of R&D. I would say that what sets Everspin apart is that we have products in the market today, and we clearly understand what it takes to make high quality, reliable, manufacturable MRAM products. We also have our own semiconductor fab that allows us to move the technology rapidly through various phases of development.
I think the major competitor tech for the next-gen memory tech is Phase Change memory. What are your own thoughts on PCM?
There is no PCM product in the market; therefore it is hard to know product reliability and performance. It seems clear from literature that reliability (number of write cycles), as well as write speed, will not reach MRAM performance. However, PCM has the potential to be a contender for future generations of Flash.
Do you think the current world financial situation might affect MRAM development? Will we see less investment in emerging techs? ??
We have excellent financial backing, so I am confident about Everspin’s position.
Where do you see MRAM in about 5 years? Toshiba said in June they expect STT-MRAM to 'take over DRAM' in 2015. I'd like to hear your own projections...??
It is a little early to make that projection, but I do believe that Everspin MRAM will have densities to rival DRAM or NOR Flash in that time frame. Everspin MRAM will establish its own set of unique markets and applications rather than simply replace these other memories. MRAM changes system architectures. It eliminates the requirements for multiple memories. You wouldn’t need things like Flash program memory, EEPROM or battery-backed SRAM if your DRAM was non-volatile. I think MRAM will co-exist with NAND Flash as an external storage device.
Dr. Saied - thank you for this interview. I wish you (and Everspin) good luck.
Ron Mertens, mram-info.com, November 2008.