The supposed Holy Grail of computing memory, MRAM is a "gee whiz" technology that could combine the speed of SRAM (Static Random Access Memory), the density of DRAM (Dynamic Random Access Memory), and the stability of flash memory. MRAM could allow your computer to turn on as fast as your TV and improve battery life for cell phones and PDAs.
Large tech outfits such as Motorola
As in the legend, though, only the pure of heart get to see the Grail. MRAM could eventually become a major semiconductor business, but NVE will be lucky to see a nano-nickel from it. The real work on MRAM had been conducted independently by technology giants such as IBM
The highlight of what NVE calls its "watershed" MRAM patents is something known as a "one-transistor-per-bit read addressing scheme." This invention relates to the electrical circuitry supporting a memory cell, but has nothing specific to do with MRAM. Further, the claim is based on so much prior art that it's meaningless and unenforceable. The DRAM in your PC has relied on this approach for decades. Motorola did license technology from NVE in 1995, but this was before NVE even had these patents. Motorola pays NVE no fees today and has no reason to pay NVE royalties in the future. Meanwhile, Cypress Semi appears no closer than it has been to producing a commercially viable MRAM-based product. No other company with a major MRAM development program has licensed NVE's patents.
Although NVE has product sales today, none pertain to MRAM. The company generates $12 million a year in revenue, mostly from "work-for-hire" contract R&D. Its scant product revenues come from selling sensors and couplers. These products employ giant magnetoresistor (GMR) devices, a form of magnetic memory. However, you cannot make commercially viable MRAM using GMR. Instead, you have to use magnetic tunnel junctions (MTJ), an entirely different memory cell technology.
Motorola walks away
Consider this: When Motorola signed the license with NVE in 1995, Herb Goronkin, director of Motorola's Physical Research Laboratories, joined NVE's board. Goronkin oversaw Motorola's MRAM development program for the next eight years until his retirement in June 2003. Curiously, when Goronkin resigned from NVE's board in July 2002, Motorola opted not to install another representative. Between June 2003 and January 2004, Motorola sold its entire 8.4% stake in NVE. If NVE's technology had value to Motorola, surely it would maintain its board position and hold on to its shares.
Motorola apparently ditched NVE. In June 2003, Goronkin told the Forbes/Wolfe Nanotech Weekly that NVE's technology process was transferred to his lab in 1995 to "kick-start" Motorola's MRAM efforts. "We saved a year in learning the technology," he said. But Motorola "had a number of problems and had to make numerous discoveries along the way." Luckily, Goronkin explained, "someone at MIT and Japan discovered magnetic tunneling junctions and I went over to Japan and we decided we are going to switch horses and get on magnetic tunnel junctions." On its own, Motorola discovered the toggle (US patent 6,545,906), which solved the "soft error" problem caused by the fact that magnets flip each other randomly when crammed in close proximity.
NVE had long insisted it would be due royalties if Motorola commercialized the MRAM device described in its technical papers, but NVE recently backed away from this claim. Last summer, Motorola spun out part of its Freescale Semiconductor
What about Cypress?
Next we have Cypress Semi, which signed a royalty-free license with NVE in April 2002 in conjunction with a $6.2 million purchase of stock and warrants amounting to 24% of NVE. Cypress CEO T. J. Rodgers had said in May 2002 that Cypress would be sampling its MRAM chips by August, with chips in production by year-end. That didn't happen. In September 2003, Cypress announced its MRAM product would be delayed again. The chipmaker also disclosed it had sold all of its 686,849 shares of NVE, retaining only its stock purchase warrants.
This past August, Rodgers predicted that Cypress would sample its 256Kbit chip by year-end and generate revenues by the first quarter of 2005. But the specs show Cypress still doesn't have a commercially viable product. Contrary to the norms of semiconductor development, Cypress' MRAM memory cell has now doubled in size to a non-nano 24 microns. This chip is many times larger and more expensive to make than the higher capacity Motorola/Freescale 1.55-micron chip or 1.42-micron chip under development by Infineon and IBM. Cypress continues to experience "soft error" issues that cause data to be lost or corrupted, forcing it to add massive amounts of redundant circuitry. But even that hasn't solved the problem. Manufacturing yields also remain terrible. Even worse, the Cypress chip uses 5 volts, whereas most current memory chips and competitors' MRAM chips use 3.6 volts or less. Cypress also employs a 3-transitor-per-bit design that doesn't even appear to use the alleged invention in NVE's "watershed" patents. Cypress looks poised to produce a chip so big, expensive, and electronically unsuitable, relative to current standards, that its product will have an extremely limited commercial market, if any.
So NVE's key MRAM patents seem irrelevant, its potential royalty-paying licensee appears not to be using NVE's intellectual property, and its other licensee can't produce a commercially viable chip. NVE's MRAM efforts are worthless today, and they seem to have no demonstrable future value.
What NVE does have is a CEO with an alarming history. Prior to joining NVE in January 2001, Dan Baker had spent more than 10 years as CEO of Printware Inc. (Pink Sheets: PRTWZ.PK), a company that missed earnings targets the second quarter after its July 1996 IPO and spent the next five years failing to match Baker's optimistic forecasts. Baker's reign at Printware ended with angry shareholders voting him out. Baker has expressed his own confidence in NVE's future by selling nearly all of his then current shares last January, pocketing $4 million. He even exercised options that didn't expire until 2011. NVE's management now owns just 8% of the stock.
NVE has also enjoyed the sponsorship of a number of stock promoters, including most prominently the Forbes/Josh Wolfe nanotech newsletter, which first mentioned the stock in June 2003. Wolfe subsequently anointed it a "Nanosphere" play and the stock took off. Wolfe has shown panache as a self-promoter. Before he had made a single nanotech investment, he modestly heralded himself as one of the "true business visionaries of the nanotechnology sector."
Blind them with science
This is the model of how a former penny stock like NVE becomes a darling of speculators. Take allegedly important patents and big name licensees who may never use the technology or produce a marketable product. Add a promotional CEO and mix in a self-styled "visionary" with a wide readership of speculators.
But dubious penny stocks have a habit of returning from whence they came. The truth is that everyone best positioned to understand NVE's technology and future business prospects -- Motorola or Cypress or even Baker -- has cashed out already. Investors who get caught in this trap have simply ignored the evidence before them.