Information Technology Will Change Everything

Are you ready for a world in which your key asset is the information stored in your people's heads and in their pocket computers? OVERVIEW: Information technology will change everyone and everything in society: how we work, live, play, and communicate with others. The next 25 years will see dramatic cost/performance improvements in technology, and these improvements will lead to a very different world. It will be a world in which computation will be both pervasive and invisible, leading to an environment where everyone and everything will be connected to the Internet. Vast amounts of data will be stored digitally online, creating huge opportunities for those companies that can find valuable information in a sea of apparently random data. It will also be a world in which the difference between the "real" and the "virtual" will become blurred, allowing for new experiences for recreation and new ways to conduct business. Although any kind of estimate of where the world will be tomorrow, let alone 25 years from tomorrow, is necessarily speculative, I believe we can say one thing with confidence: Information technology will change everything in the world in which we live. There will be no institution, no person, and no government that will be unaffected. To understand why I believe this, let's look back and try to do some reasonable extrapolations about where we are going. I start with the storage of information. If you go back to the early 1950s, before there was a magnetic disk-drive, you find that it took a hunk of material almost a millimeter on a side-about a billion billion atoms-to store a single bit of information. Extrapolate from that past through today and out to 2025, and you get down to 1,000 atoms per bit of information. Is this really going to happen? Can we continue to push this technology down to 1,000 atoms per bit of information? First, observe that there are five different technologies in Figure 1. As one slows down, another takes over. We go from ferrite cores to magnetic bubbles to disk drives. The technology has evolved from the inductive head, to the thin-film head, to the magnetoresistive head, to something else. Soon there will be the giant magnetoresistive head, and pretty soon the supercolossal magnetoresistive head! Technology finds a way to keep on grinding. If you extrapolate this technology trend, you go from a cost of $10,000 to store a megabyte of information in 1956, when we had the first disk drive, down to $0.1 to store a megabyte of information today. And if you were sufficiently adventurous and extrapolated this all the way out to 2025, for $1.00-the loose change in your pocket-you could store a terabyte (1,000 gigabytes) of information. Will this really happen? I believe it absolutely will happen! We can, in the laboratory today, store one bit of information in an individual atom. And that's not even the limit. There is no Heisenberg uncertainty principle that says you must stop at a single atom. You can put multiple bits of information in the electronic structure of the atom. We don't quite know how to do that yet, but the technology of using individual atoms to store information is not science fiction. It exists today in the laboratory. We are a little slow in getting the information in and out of atoms, but that will come too. Let's look at another trend. Figure 2 shows the energy that it takes to do one piece of computation. It dropped by some 12 orders of magnitude from 1940 to 1990. If you extrapolate this out to 2025, the energy it takes to do one piece of computation goes down to kT, the energy in the atoms themselves. Actually, that's not a limit either. We can show theoretically that it really doesn't take any energy to do computation, so long as you don't throw away excess information. The concept is called adiabatic computing. We can actually build circuits today that use adiabatic computation tricks. The Promise of Moore's Law In the semiconductor arena, there is Moore's Law, which says we are going to keep cramming more and more information, higher and higher density circuits, into and onto silicon chips. …