Where? Likely in 10-to-50-kVA layouts for the typical system user, and where it makes sense in terms of single point of service, cost, performance, reliability and reduced complexity. One need only look at the present generating capabilities of the United States and the power draw of the Internet to understand how integrated UPSes may yet become.

Aside from the Y2K challenge, the supply and power-transmission network will be adequate for the next three to five years, as indicated by the North American Electric Reliability Council and its many affiliates. That means few if any new power-plant buildings of major proportions and a more or less fixed power budget, leaving the UPS as the major plug in the dike to ensure "24-by-seven-by-365" service.

And a bigger "go-to" resource at that (with UPSes already at several hundred kVA), in large part thanks to Internet demands. Throw in the increased emphasis on power management across the board-which includes greater dependence on Web linking to networks to monitor and control the activity of distributed networks- along with the increased complexity of powering those networks, and the trend toward more individually compact building blocks appears inevitable.

When? As regional power-capacity margins erode below 10 percent for starters, this scenario is seen as a factor along about 2003. That means tighter integration as we surge into the new millennium. The technical and business aspects of "packaging" a deal with UPSes and switch-mode power supply systems will undoubtedly be a bear, and we're not certain how meaningful collaborations and architectures will take shape.

But they probably will. With power in the driver's seat today, that old jazz about sourcing switch-mode power supplies for OEM equipment initially and adding UPSes as an afterthought seems likely to change.