The End of Moore’s Law? Don’t Bet on it.

Digital MindIn Moore’s Law and The Future of Health Care, I offer a vision of healthcare based on exponential advancements in tech innovation as described by Gordon Moore. Moore is an Intel cofounder and is credited with observing that computer circuits have shrunk in size and doubling in compute capacity every two years. Moore’s Law is what drives down costs & size, but that logarithmic trend is not easy to grasp. So let’s look at two analogies explaining a 60,000 improvement in cost and 90,000 improvement in speed since Intel started tracking computer chips in the 1960’s.

COST — If the price of cars and gas improved exponentially at the same rate as computer chips, we’d be able to buy a new car for about 8-cents today and would only spend 2-cents per year on gas. At that rate, cars would be disposable, and we might just buy a new one for each trip, as a fashion accessory matched to our outfit.

SPEED — If the speed of air travel advanced at the same exponential rate as computing, today we’d be able to fly from the U.S. to Japan in less than a second, but the plane would be just over 1-tenth of an inch long.

Industry analysts keep predicting the end of Moore’s Law, arguing for many reasons that computer chips can only get so small or so cheap, and today I responded to another article about The End of Moore’s Law. Here’s my response, which shows optimism from my 30 years at IBM (I retired in 1999) and my interest in technology as a futurist.

The End of Moore’s Law? Don’t Bet on it. 

Moore’s Law defined transistor size, but the more important trends go beyond that. Transistors are just one part of processor chips, and processors are just one part of servers, so I’m not worried about physical limitations of semiconductor manufacturing that might limit transistor size. Other computing architectures will surely emerge, including organic, optical, parallel, and quantum computing, moving us beyond information processing and toward shared human knowledge and digital consciousness.

Consider software evolution as it transitioned from machine language to assembler language, 3rd generation languages (FORTRAN, COBOL, PL/1), application generators, and the open-source movement with online sharing of ideas and code. Each new tool enabled the creation of better tools so we could create ever-faster and more powerful systems more quickly.

Next may be artificial intelligence and machine learning. Just as the brain evolves with ever-more synapse connections between neurons, computing may evolve exponentially with more connections between each processor, clusters of processors, and computing systems. And each processor will eventually be the size of cells, implanted and connected to neurons, providing an efficient brain-computer interface and enabling a brain-brain interface that supports the idea of shared consciousness. Futurist Ray Kurzweil describes a vision where tiny embedded processors may each have the processing power of the human brain, and then the human race.

Now consider that there are 100 billion neurons that they might individually be connected to. No, I don’t think the computing evolution will slow. It will accelerate exponentially as it has with Moore’s Law and probably faster.

How Small can a Transistor be?

Already it’s the size of an atom, at least in research labs. This infographic (and linked article) suggest that Moore’s Law has a limit, but while transistor size certainly may, I don’t think that will limit computational ability because, as I said above, new architectures will evolve.

How small can a transistor be? If we Abandon Transistors?

The following video from SciShow explores some of the new materials and architectures that I mentioned while arguing against the end of Moore’s Law. It’s fast-paced and entertaining.

 Related Articles