Power Report

polluted by an are furnace over a hundred miles away. With only a couple of prototypical exceptions, all switching on the grid is still electromechanical. A spring helps speed-up the switching, which minimizes - but does not eliminate - arcing, and a concomitant burst of electrical noise. Such Rube Goldberg devices fire up and down the length of the grid, hundreds of times a day.

For all that, our trillion-dollar electric network is a rernarkdble achievement. Some 15,000 central power plants deliver over 3 trillion kWh per year, with (roughly) 99.9% reliability. That's "Three Nines" - or about 8 hours of outage a year for a typical consumer. Before SmartChip, that was good enough. The electricity system was built in response to three main waves of demand, each a function of a new invention: the light bulb, the electric motor, and the air conditioner. For each of these technologies 99.9% reliability was acceptable-considering the high cost of further improvement. It gets very expensive, very fast, to boost reliability much above that.

Hospitals, airports, and military bases have demanded better than Three Nines for years, and have deployed their stand-by generating systems accordingly. Phone companies deploy huge battery banks and gas-fired stand-by generators alongside their large central offices, to keep phone lines up even when the lights are out. But SrnartChip has changed the power world in two ways. As microprocessors and Web links penetrate deep into the economy, in to the facilities of even "ordinary" manufacturing and commerce, power quality and reliability become as important to "old economy" companies as they are to Bell Atlantic and Amazon. Residences are next, as SmartChips come to permeate the home. The sheer magnitude of devices and users that depend on SmartChips is unprecedented, and accelerating.

SmartChips, the devices that use them, and the networks that connect them, create a new standard far beyond Three Nines. Reliability demands start at Six Nines for the telecom and dot-com world - 99.9999% up time. At the fifth Nine we're into minutes of down time a year, tolerable for a homeowner, unacceptable for any serious dot-com company. At Six Nines, we're talking 30 seconds outage a year.

But Six Nines is still orders of magnitude away from the reliability required to power SrnartChip industries and companies. At the seventh nine, several seconds year, we're protecting against a spectrum of minor distortions on power lines; common occurrences that last less than a minute, but are still seen by SmartChips as network crashing events. At the eighth nine - 99.999999% - interruptions are measured in hundreds of milliseconds - a trivial flicker for a light bulb, but still enough to crash a SmartChip. It's only around the ninth and tenth Nine that the power can really he labeled clean enough for SmartChip purposes.

Here's the rub. Practically speaking the traditional power grid will never he able to provide much better than Three, perhaps Four Nines of quality. The grid, distributing power over vast distances, is necessarily too exposed to catastrophic events to ever satisfy SmartChip's real needs. Rebuilding it to do so would make it too expensive to support the economy power needed by the doniinant 'dumb' appliances on the grid. To add Nines, we must turn to an array of systems, ranging from capacitors and inductors mounted on a motherboard or in a UPS (uninterruptible power supply) to fix problems of milliseconds duration; to batteries, flywheels, and super-conducting coils to compensate for outages of seconds to minutes; to diesel generators and turbines to supply back-up for hours or weeks. Every step requires a switch that can operate fast and cleanly enough to make the switching process invisible to the SmartChips. That's PowerChip's job.

Building in reliability is expensive and it gets more expensive for every additional Nine. But SmartChip companies have no choice. Clean power, information- quality power, is becoming a sine qua non of the information economy and thus one of the greatest business opportunities of our time.

Information-quality power is one of the greatest business opportunities of our time.

How much will people pay for the next Six Nines? Well, what does it cost a company like Schwab to go off-line for an hour? Or a day? What does it cost the cell phone company that loses a cell site in mid-town Manhattan? What does it cost the harried home-office worker, on a deadline to complete a report? That's how much the extra Nines are worth. Which is why information-quality power is already being sold at rates and in volumes that make clear this market will yield profits far surpassing anything dreamt of in the electric industry for decades.

power1.gif (37968 bytes)

True, the volume remains in the Three Nines market, the market for regular old grid power now increasingly traded coast to coast as a true commodity. But as we add Nines the cost, and the value, of the electrons rises rapidly. A single central power plant up 99% of the time delivers Two-Nines electricity wholesale into the grid at 2c/kWh. A number of such plants woven into the grid allows utilities to deliver 99.9% reliability at about 10c/kWh retail at your plug.

For every Nine after that, the costs soar. Three extra Nines can bring a 200 to 1000 fold price premium. But that's a premium that chip-centered businesses which will soon mean all viable businesses will readily pay. Information-based businesses have been doing so for some time. You have almost certainly done so yourself, though you probably didn't think of it that way. When you buy an American Power Conversion (APCC) UPS to keep your desktop PC isolated from ubiquitous line voltage sags, what you're really doing is paying $20/kWh; roughlv 200 times retail. True, you buy only one kilowatt-hour per year at that price - spread out over 20 five second events each using 0.05 kWh. But with inilons of customers APC is a $1 billion a year power company. The challenge and the costs are modest at the desktop though. Costs and value rise exponentially with power.

Companies at the core of the SrnartChip economy require very large quantities of High Nines power. At the semiconductor manufacturing plant level, or the major dot-com level, the per kWh cost of more Nines hits the stratosphere.

If a voltage sag becomes a sustained outage, you kick into an on-site diesel generator, in effect a Five Nines device. The imputed electric cost can he as high as $5/kWh for a few hours of annual use.

But to make a seamless transition from the grid power to the diesel generator, a transition measured in milliseconds, requires batteries (sometimes flywheels) and PowerChips. The resulting Six Nines reliability comes at a cost of perhaps $ 1,000/kWh.

At the high end, factory-power levels, PowerChips combine with superconducting storage devices to supply the capability to pump megawatts into the power curve in fractions of a 60-eycle wave. The implicit cost of power delivered at these Nine and Ten Nines levels ranges beyond $ 1 00,000/kWh.

Clearly, there will alway's be a lot more Three Nines kWh sold than those at Six Nines and up. But with these price spreads, you dorn't have to sell a lot of High Nines kWhs. Caterpillar (CAT) in selling diesel generators is really selling Five Nines. APC is selling Six Nines. American Superconductor (AMSC) sells Seven Nines. Silicon Power sells Nine Nines. At each successive tier of Nines stands another multi-biflion-dofiar market.

In aggregate, the total market for kilowatt-hours above Three Nines is already 20% of the Three Nines market and growing at double-digit rates.

As SmartCIiip continues to reorder the economy, the aggregate profits in premium power and PowerChip companies will soon exceed those in the Three Nines market. This transition will turn the electric industry upside down.

One finds High Nine companies falling into three categories: Clean Power Systems; Ride-Through Systems; and Stand-Alone Local Generators.

Clean Power Systems

PowerChips clean up power by switching it fast, rnediating between electrons caning off a primary power source and electrons temporarily stored in good-sized capacitors and inductors alongside the PowerChip itself. It's impossible to store much total energy on a circuit hoard, but you don't need a lot to clean up blips and dips. What you do need is very fast, accurate, and intelligent electronics to mediate between the dirty source and the on-prernises reserves. PowerChips do the job.