IBM, Infineon advance magnetic-memory prototype

IBM and German memory maker Infineon unveiled a prototype 16-megabit magnetic-memory chip this week as the struggle to establish new standards of memory rolls on.

The two companies presented the MRAM, or magnetic random access memory, chip at the Very Large Scale Integration Circuits Symposium in Hawaii on Wednesday. Unlike today's mainstream computer memory, MRAM relies on magnetism rather than an electrical charge to store data.

MRAM could significantly advance the state of memory technology, at least according to proponents. Like flash memory, MRAM continues to store data even after its host computer is turned off. It also retrieves data rapidly and can theoretically last forever.

Last June, at the same conference, IBM and Infineon published a paper describing how the companies produced an MRAM chip on the 180-nanometer manufacturing process that held 128 kilobits of data.

At the time, the two companies promised to more fully demonstrate MRAM in early 2004 and predicted that MRAM could go into commercial production by 2005. That deadline, however, has become a tad amorphous. In announcing its prototype, Infineon said MRAM could potentially enter many segments of the memory market "in a few years."

Memory, particularly flash, is expected by many to undergo substantial architectural transformations over the next decade or so because of the technical complexity and cost involved in improving contemporary memory technology a la Moore's Law. Simply put, it's just becoming too expensive and hard to shrink memory chips and add transistors.

MRAM, however, is only one of several contending ideas. Nanochip, a start-up that has received funding from Microsoft, is proposing a flash replacement in which tiny actuators heat microscopic points on a substrate, which can then be read as 1s or 0s. Motorola, meanwhile, says it will incorporate silicon nanocrystals into flash memory, allowing the company to continue to shrink flash chips. ZettaCore, a start-up with several prominent backers, touts designer molecules.

Naturally, each of these companies can explain the virtues of their own approach and problems with the competing ideas. But, as the vague deadline for MRAM's entry into the market indicates, displacing existing technology won't be easy because of the risks and costs involved.

"It is a conservative industry," said Randy Levine, CEO of ZettaCore, "But people are talking about it (the transformation). I don't know a major memory vendor that isn't."