The focus of R&D efforts is shifting at some companies, in some cases as part of a strategic plan and in others as a result of the economic downturn. At Cisco Systems, the emphasis in the last year has been on enhancing existing networking systems rather than developing entirely new ones and on developing new add-on products. That's because local area network switches, routers, and other networking platforms have a life of five to seven years, and fewer customers are buying new systems. What's more, Cisco has found that its customers get more productivity out of their systems in years four through six than in years one through three, giving them less of an incentive to buy new ones every few years.

With that in mind, a lot of Cisco's R&D dollars go into the development of new security technology, such as intrusion and virus detection and support for converged voice, data, and video networks. "There's no discussion anymore about whether IP voice, data, and video will come together. The answer is yes," Cisco president and CEO John Chambers said last summer (for more on voice and data convergence, see our story, "Vox Popular").

EMC has been shifting its R&D focus from hardware to developing new software, such as its AutoIS for managing arrays of storage devices. Software is an increasingly important component of EMC's storage systems; standalone software such as AutoIS accounted for 22% of the company's sales in 2001, up from 12% in 1999 and 16% in 2000. Lewis says 75% of EMC's R&D budget is devoted to software development.

R&D spending averages 5% to 10% of sales for computer hardware companies and 10% to 20% for software companies. The latter is comparable to the 10% to 20% of sales that pharmaceutical companies spend to develop new drugs.

The lion's share of Oracle's R&D resources are devoted to enhancing existing software, CFO Henley says, and that's been the case for several years. Oracle occasionally develops new products, such as its recently introduced collaboration software suite, or new modules for its 11i set of applications. But Henley says "sustaining engineering" -- developing new releases of established products -- accounts for most of the vendor's R&D expenditures.

Much of Microsoft's research and development aims to deliver software that integrates PCs and mobile computers more easily into everyday life. Microsoft's ClearType technology for easy-on-the-eyes on-screen reading shows up in the Tablet PC version of Windows, due Nov. 7, and in the next version of Microsoft's Outlook E-mail program, due next year. Tablet PCs will include the ability to search and edit handwritten notes. And Gates predicts people one day will point to desktop LCD screens instead of using a mouse. A top R&D initiative, Always Works, seeks to establish better feedback between Microsoft and its users. And a future version of Windows, code-named Longhorn, will include a file system that recognizes metadata about documents so users can dynamically organize their files by content, without memorizing folder names.

Sun's R&D to-do list includes creating tools to improve the administration and use of the IT infrastructure and building portal, Web server, E-mail, and calendar applications into its Solaris operating system. The company continues to develop new Java and XML technologies and is increasing its R&D investments in storage and storage-management technology, Tolliver says. Still, Sun has yet to make good on McNealy's vow to boost development spending: R&D expenditures totaled $437 million, or 16% of sales, in the first quarter ended Sept. 30, compared with $473 million, or 17% of sales, one year earlier.

While most companies don't distinguish between research and development expenses, but instead lump them together on their income statements as "research and development," most of that money goes to product-development work. But a few, notably IBM, Lucent Technologies' Bell Laboratories, and Microsoft, still conduct basic research performed by scientists to discover new technologies or processes. Such work generally isn't on a fixed schedule, Harvard's Chesbrough says. Development, in contrast, is led by engineers and is focused on turning ideas and technology into money-making products. And it's usually on a tight timetable.