The Price of the Power Curve

Component complexity has roughly doubled every year...this rate can continue

for at least 10 years. By 1975, the number of components per integrated circuit

can be 65,000. I believe that such a large circuit can be built on a single

wafer.

Gordon E Moore published this in 1965 in Electronics. The magazine is dead,

but the prophecy lives on as Moore's Law. It was less of a law than a rough

prediction that chip density would double each year, later updated (in 1975) to

"every two years".

Forty years down, this rough guide still maps the trend curve in computing.

Some say it's a self-fulfilling prophecy, a benchmark to which Intel, which

Moore co-founded three years later, has been living up to, and thus a target for

the industry to match.

Such doubling of 'power' is seen in one other area: memory (and storage).

My first computer had 16 MB memory. My camera today takes a 4 GB CF2 card.

You've heard the analogy: if cars had done that, we'd have had a $10 Merc

going a thousand miles on a gallon. But hold on. I still don't have a $10

computer. (Or do I, embedded into my car, camera, MP3 player?) The PC is still

expensive, even though each year it packs more power into a lower price.

I asked Gordon Moore (in an April conference call): could this amazing

progress have been better applied to drive down cost? Could we have had a $100

laptop if the specs had stabilized-perhaps with less 'power', but a much

wider penetration?

Possibly, Moore said, but the industry needs the money to keep investing in

newer tech. If it had been selling the same number of laptops at $100, that

wouldn't allow progress as we know it today ... and the economics could

collapse.

Even so: we know that tech needs to be affordable (and compelling) to impact

the earth's billions. The mobile phone showed us that. A cheap computer has

the potential. A $100 laptop can cause more than a tenfold increase in usage-if

it doesn't mean a big sacrifice in specs. (Growth doubled when laptop prices

dropped in India in 2004, but people still don't buy the Rs 35k laptop as much

as the 70k laptop.)

Moore's Law's author expects the law to survive a decade or two more

before it hits the barriers of physics, or of cost. That means a continued

doubling of power...

But developing economies are where the growth is. And they need cheaper tech.

And a more 'cost-driven' edition of the law, such as: "The median

end-user cost of tech will halve every four (or five) years". We're not

very far away, but I believe such a guiding principal can drive the next big

explosion of tech usage-across the planet.

Prasanto K Roy