What the End of Moore's Law Means
Nature recently published an article on the forthcoming end of Moore's Law. In case you don't keep up with tech buzz, this refers to Intel founder Gordon Moore's empirical observation that the number of transistors in a microchip doubled every 24 months, running from roughly 1965 to the present. To the layman, this meant that computers roughly doubled in power every two years and became 1000 times more powerful every two decades.
Moore's Law is more than just geek arcana. It's a philosophical statement about the progress of technology. We live in times with such rapid growth in scientific knowledge, manufacturing capability, and global economic power that we have come to expect technology to improve so fast we can't even keep up. We're used to our computers and phones and entertainment systems and internet resources upgrading at a blinding tempo. So much so that we take it for granted.
Well, things are about to change.
Many experts and writers have forecast that computer power will grow exponentially forever, eventually bringing about the replacement of human intelligence by computational power. Instead we may soon be reminded that computers, too, are natural systems, subject to natural laws. Processes based on harnessing the resources of nature don't grow exponentially forever; they plateau. Mushrooming growth gives way to a more measured progress.
We witness this exact trend in many areas of human technology.
The first rocket carrying a human being to orbit launched in 1961. By 1969 a man was on the moon. Within a year of its launch in 1982, the space shuttle program lofted 10 missions into space. NASA planned to put men on Mars by 1983, or 1987, or possibly 1999. Colonization would follow. Technology, politics and economics came together to slow things down. Now we pay Russia to put people in orbit.
Orville Wright's first flight (1903) traveled 120 feet and failed to escape the chasing Wilbur. By 1912, planes hit 100 mph. Lindbergh's 1927 flight covered 3600 miles. In 1947 we broke the sound barrier; by 1953 we doubled it. The 1960s ushered in the mighty SR-71 Blackbird, capable of cruising at 90,000 feet at velocities topping 2200 mph, and the Boeing 707, carrying nearly 200 passengers across a continent at 500 knots. But then the 60-year revolution in speed, range, and size tailed off.
Today's planes are better, but not radically more capable. A modern passenger jet looks just like a 707. It's no faster either, but it's lighter, more powerful, and can carry passengers further in greater luxury. No production plane has ever bested the SR-71's speed and altitude capabilities; warplanes now advance more subtly in other ways.
Other inventions that make our modern lives so easy have followed a similar path. Refrigerators don't get any colder than they used to. But they are more efficient, a little bigger, and they fire ice cubes and spray water at you on demand. Cars are not capable of massively greater speed or range than they were decades ago, but they are moderately more efficient, more comfortable, more luxurious, and more safe. Those are just the practical improvements. Cars are also more powerful, more sporty, more blingy and more electronically sophisticated.
Computers will likely face something similar: a period of measured growth and increasing refinement in their finer aspects.
Having sprinted nearly to the limits of speed and power within current technology, devices may now consolidate their gains by growing in sophistication instead of sheer power. A smartphone won't be so much faster as it will more carefully pack more features in, with better integration, while using less battery power. Personal computers have already begun to do this: they are less about doing one task with greater speed than they are about doing more tasks at once, running longer on a battery charge, and doing more things in a more integrated fashion.
This isn't entirely bad news. A gentler pace of change may give us time to breathe and become more used to what we have. Perhaps we won't have to throw out our 8K TVs that replaced our 4K TVs that replaced our HD TVs that replaced our tube sets quite so soon.
The technical reasons to expect this slowdown are clear: silicon is nearing its limits both scientific and economic, and there is no heir.
The heart of a modern chip is made of crystalline silicon, carved into nanometer-sized transistors and topped with copper connections and wiring. This technology has been so fruitful that the investments in scientific research, specialized equipment, specialized techniques, specialized chemicals, and most importantly dollars in the trillions have advanced it vastly beyond any competitor. In fact, most competing schemes have never passed beyond the dreams of primitive research prototypes.
If a successor to silicon is coming to re-fuel Moore's Law, it's nowhere to be seen today. And given the decades of monumentally successful work behind silicon, it's probably not going to pop up overnight.