Atomic resolution storage will probably rise to the fore in spacecraft before it takes off in PCs, but that doesn't make its potential any less meaningful to the computer market. The technology is derived from atomic probe microscopy, and its mechanics sound appropriately space-age. A storage medium moves along under tiny probe tips that emit lasers no wider than one atom. To read already-recorded patches of medium, the probe tips produce a weaker, even more sensitive laser. At present, the chief limitations to this technology are ones of precision: not only do probe tips need to be super-refined; the medium itself must be made of the right stuff. Current theories see the medium comprised of two physical states; and probe tips would record data by changing the state of individual bits. No less challenging is the technology's need for strong packaging. A vacuum would probably be required to hold in all those flying electrons. Add to that the marketing challenge that such technology would present (think unhappy magnetic-disk producers), and you've got a product that probably won't be available anytime during the next decade.

5.IBM's Millipede

IBM is currently working on another alternative to the hard-disk drive. The "Millipede," as it calls this new device, can store up to 500 gigabits per square inch (as compared to the 35 gigs a magnetic-disk inch can presently store). Still in its testing phase, the technology calls for over 1,000 styluses producing tiny indentations in a flat plastic surface. The stylus' tip is no more than 20 nanometers in radius. Short, sharp waves of electricity run through it, heating it to 400? Celcius, causing it to melt the plastic and create a small divot. In order to read those divots, the styluses run along the plastic surface and when they encounter an indentation, they lose heat and the resulting change in their electrical resistance can be mapped. The Millipede's chief weakness is durability: the plastic surface is susceptible to the sort of wear and tear that could seriously corrupt delicate digital information. IBM plans to investigate the possibilities of this technology over the next two years or so, concentrating on its viability in watches, cell phones and digital cameras -- products in which size does matter.

Most approaches to computerized memory use the paradigm of a needle that writes data onto or reads data off a secondary surface. By contrast, multi-layer technologies look at increasing memory by introducing multiple layers of that secondary surface into a single device. Layers are coated with fluorescent material to prevent interference between lasers. These coatings change the lasers' wavelengths as they pass through layers, thus preventing cross-talk. Fluorescent coating currently enables devices to hold up to a hundred layers of readable material. This translates to memory of hundreds of gigabits per single disk. Plus, parallel reading of different layers means faster access times -- something that is not frequently addressed in conjunction with increased memory capacity. These days research tends to favour either memory space or access speed -- so much so that the industry seems to be producing two separate types of devices: ones that are big, and ones that are fast, without attempting to reconcile them. While computer capacities increase by 130% annually, speeds creep along with an increase of only 40% per year. Accelerating access speeds is a tricky business, though -- whether you're talking about single-layer devices or multi-layer ones. It'll be a few years yet before computers are remembering more, and a few years after that before they're remembering more, faster.