Executive Summary

Every measure of the Internet shows explosive growth in numbers of users, number of Web sites, bandwidth, and the billions of dollars in e-commerce. However, there is one measure that has been absent from consideration thus far—the quantity of electricity needed to keep the Net hot. Two decades ago, this metric, like all others for the Internet, didn’t matter. The number of computers and microprocessor-based devices on the Internet was counted in the thousands. Now there are at least 100 million.

     Preliminary calculations reveal that the electricity appetite of the equipment on the Internet by itself has grown from essentially nothing ten years ago to 8% of total U.S. electricity consumption today. In all likelihood, the Internet is responsible for one-half to two-thirds of all the growth in U.S. electricity demand in the last decade.

     This analysis finds that for every 2,000 Kbytes of data moving on the Internet, the energy from a pound of coal is needed to create the necessary kilowatt-hours.

When other uses of computers are included (many of which are linked directly and indirectly to the Internet economy), the share of all U.S. electricity consumed by computer-based microprocessors jumps to 13%.

     The turmoil of de-regulation and competition for electric utilities has generated a scramble for structure and revenue in the emerging competitive era for electricity. Underlying much of this activity is the implicit and often explicit assumption that the business of providing electricity is largely saturated, a holdover from the old industrial age and out of place in the new info age.

      Today, utilities and the orbit of regulators, experts and consultants in the electric industry preoccupy them-selves with brand, identity, merger and “stranded cost” issues. Meanwhile, the Internet is building a tsunami of old-fashioned electron demand the likes of which utilities have not seen in half a century. All of this electric growth comes from the avalanche of equipment essential to cre-ate, access and operate the Internet.

     Many are investing on the assumption that use of microprocessors and the Internet in particular has just begun. If so, then the implications for electric demand, reliability and utility architecture portend nothing less than a revolution. Every PC-type of microprocessor is like a light bulb energized by 50 to 100 watts; but unlike lights, many integrated circuits, are on all the time. In addition, on the Internet, demand begets more demand. The microprocessor and the Internet help explain why great strides and billions of dollars invested in traditional electric efficiency have not flattened overall electric load growth. Efficiency gains in lighting, motors and refrigeration—the anchor products of the first electric age—have been more than offset by the electric needs of the products of this next info-electric age.

     Lost in the rhetoric of the power of bits to transform industries is a simple fact: every information technology device has two connections—one to move bits, another for power. Unlike the dominant stand-alone computers of a decade ago, networked computers generate demand for other devices. Every PC on the Internet is connected to a myriad of other electricity-consuming boxes in the network. The worlds of power stations and desktops seem far apart, but are connected by the power cord on the back of every box, and the power of geometric growth. The implications are clear in the arithmetic of the growth of PC use.

     There are already 50 million PCs in households, another 150 million computers in businesses and 36 mil-lion more being sold every year. Not only do the desk-tops and the peripherals need electricity, but so do all the other microprocessor-based boxes in the network that push, amplify, transmit, receive, route and manage the bits. There are millions of these boxes too. Not only is electricity needed to operate these boxes, but they are fabricated by one of the most electric-intensive industries in the country. The $50 billion/year semiconductor industry is now the nation’s largest manufacturing sector, surpassing the auto parts sector in 1995.

     At the worldwide level, this analysis shows that Intel’s vision of one billion PCs on the Internet represents a global kilowatt-hour demand equal to the entire output of the U.S. electric grid. The magnitude of the appetite of the Internet and information age for electricity has powerful implications for those in industry and policy makers. It now seems reasonable to forecast that in the foreseeable future, certainly within two decades, 30 to 50% of the nation’s electric supply will be required to meet the direct and indirect needs of the Internet.

     On top of the sheer need for power, the very nature of the Internet and information age creates an unprecedented demand for reliability. Keeping a gigawatt-based network “up” 24hours per day, 7 days a week sets a new standard for high power reliability. As a consequence, the architecture of the electric supply industry will be forced to adapt to the demands of the Internet. Indeed, reliability will take on an entirely new meaning for electric engineers in the decade to come. Furthermore, issues pertaining to electric price and supply, largely irrelevant two decades ago, now assume a central importance for the companies which comprise the networked part of the economy.

     While environmentalists and utilities have been standing on desks to screw in light bulbs that save 10 watts here and 50 watts there, the owners of the desks have been plugging in PCs and peripherals that gobble 1,000 watts and more—and create an echo on the Internet requiring still more power. The debate over what sources of power we should encourage the market to use, which dominates the electric restructuring debate, will be buried by the market’s info-age driven demand for lots of power, cheap power and increasingly reliable power. Over the next decade, issues like so-called “green” power, will lose urgency in the face of the overwhelming need for “smart” power tailored to meet the Internet economy.