It’s no news flash to say that computers are going to keep getting smaller, as they have for the past 50 years. But even as they vanish from sight, computers will, in an important sense, grow much larger.

That’s because the time is coming when computing devices connected in a wireless web will permeate our entire physical environment, toiling behind the scenes to monitor and manage our houses, factories, roads, vehicles-even our bodies. But this lofty vision will be realized only through a series of small improvements in computing’s nuts and bolts. Some researchers, for example, are developing ways to bring new capabilities to the existing Internet, such as powerful network-based services that can link a company’s inventory systems with its accounting and customer databases. Others are studying technologies to broaden the Web’s physical reach-among them more power-efficient microchips and high-quality broadband wireless systems.

In the world created by these converging trends, networked computing devices will surround us-but we will no longer think of them as “computers.” They’ll simply be part of the furniture. We’re already well down that road. “Your car has dozens of processors that adjust all kinds of things, yet you just think of them as the heating system and the air bags and the brakes,” says Richard Burton, who manages distributed-systems research at the Palo Alto Research Center in the heart of Silicon Valley. “You’re not aware of all the computation there.”

This movement-toward what’s variously known as “ubiquitous,” “pervasive” or “embedded” computing-is hardly new. But it is gaining momentum. Thanks to recent advances in underlying technologies such as semiconductor manufacturing and networking software, proponents have moved beyond the stage of spinning gauzy theories and started tackling the technical problems. “Ubiquitous computing will be the dominant paradigm in information technology,” predicts TR100 judge Juzar Motiwalla, a partner at Green Dot Capital in Singapore.

At first blush, it might seem that computing is already ubiquitous. After all, the World Wide Web transformed the Internet from the province of academic scientists into history’s biggest town library, village marketplace and sidewalk soapbox. Now, though, software designers, including several members of this year’s TR100, are turning the Internet and the Web into the media we’ll use to stay connected, share our favorite content, tap into distant computing resources and run our businesses-and do it all faster.

Justin Frankel of AOL Time Warner, for instance, is the originator of Gnutella, an ingenious program that lets PC users link directly to each other’s hard drives through the Internet. The result is a Napster-style file-sharing free-for-all, without a central database or server that ticked-off copyright owners can shut down. But music sharing is only the beginning of what these “peer-to-peer” programs could do. A computer employing such software uses the Internet to locate a handful of other machines running the same program; these machines are connected to even more machines, and so on, eventually forming vast webs that can propagate search requests and files. Gnutella’s power to easily copy and move documents around the network could make it easier to store information wherever disk space is available, for example, as well as to keep one step ahead of potential censors.

At the other end of the computing-power scale from Gnutella, researchers like Steve Tuecke of the Argonne National Laboratory in Illinois are writing software that unifies supercomputers around the world into a single “grid.” Tuecke was the lead software architect for Globus, open-source “middleware” that provides a common language for accessing distant supercomputers, data-gathering instruments and scientific databases. Globus includes tools for automatically locating the hardware and software scientists need, authenticating legitimate grid users and parceling out parts of a computational task to whatever facilities have spare processing cycles. While Globus is now used mainly by research scientists, IBM, Microsoft and other companies have adopted it as a step toward new and potentially lucrative network-based services.

Such services use newly standardized Web protocols to give users access to e-business software running on any kind of computer on the Internet, taking over data-intensive tasks like inventory management, scheduling and accounting. In addition to the big firms already exploring this area, “A whole host of new companies will come along to provide Web services,” predicts TR100 judge Anthony Sun, a general partner at Venrock Associates, a Menlo Park, CA, venture capital firm.

Case in point: Bang Networks, a San Francisco startup founded by Tim Tuttle. Recognizing that the performance of Internet-distributed software might suffer due to network bottlenecks and lost data packets, Bang developed “intelligent routing” that maintains secure communications. “In dollar terms, these business-to-business and business-to-consumer services are going to remain the dominant aspects of ubiquitous computing for the near future,” says TR100 judge Philippe Janson, who works in IBM’s Zurich Research Laboratory on the kind of computer networking hardware that forms the hidden “back end” enabling such services.

Making the computer networks we have faster and smarter makes good economic sense. But technologies like peer-to-peer file sharing, grid computing and Web services may only reach their full potential when we no longer have to stay riveted to our desktop PCs to use them. “Until anybody can have access to broadband content anytime, anywhere, we are not done with the infrastructure,” says Sun.

This challenge hasn’t escaped the attention of infotech researchers. MIT’s Vahid Tarokh, for example, has invented a way to keep wireless signals strong long after they’ve left a transmitter by broadcasting the same signal from multiple antennas. Such technology, combined with emerging standards for packing more data into radio transmissions, could extend bandwidth-hogging Web services to cell phones and handheld computers.

Chip makers are betting that such technologies will unlock the Internet in a way that businesses and consumers can’t resist. This spring, Intel announced plans to build radio transceivers into all of its silicon chips by 2010. This development could reduce the number of components in-and hence the cost of-mobile, connected devices. And Sunnyvale, CA-based National Semiconductor has created an entire division dedicated to building energy-efficient chips for devices like lightweight, tablet-sized Web terminals. The company’s latest Geode chips, which feature a control processor that puts components to sleep between bursts of activity, use about one-tenth the power of the microprocessors inside today’s PCs. Such chips should fuel the development of portable information appliances-as well as the networked sensors and controllers that will extend our awareness into our surroundings.

These devices promise to help with the chore of running the technological infrastructure and to bring us varieties of information never before available-for example, real-time data on the structural integrity of bridges or buildings during earthquakes or terrorist attacks. But to be practical, such highly distributed systems will need the ability to diagnose and fix their own bugs and to reroute messages around lost nodes. The software to accomplish this remains very much on the drawing board. “We have a lot of work to do on the plumbing,” says Gaetano Borriello, head of an Intel-sponsored ubiquitous-computing lab in Seattle.

Which is another way of saying that the TR100 and their information technology peers will have to keep innovating-finding new ways to furnish the future with intelligent machines that draw their power from their very invisibility.