The Moore You Know...
We live in a world our ancestors could barely imagine. The Internet, artificial intelligence, self-driving cars, facial recognition, teleconferencing, distance education, cryptocurrency, smart phones, touch screens and virtual digital assistants are but a few of the near-miraculous technologies that have transformed our world in recent years. But as mind-boggling as all of these are one technology has made them all possible: the microprocessor, guided by the self-fulfilling prophecy of Moore’s law. Unfortunately, many people believe this long-lived trend is coming to the end of its run. But is this truly the case or is the pronouncement of its demise once again premature?
It’s hard to believe that Moore’s law has been with us for more than a half century. In 1965, Gordon, then of Fairchild Semiconductor, published an article based on his observations that the number of transistors they could fit on an integrated circuit had doubled each of the past four years. Moore looked at the company’s plans and realized this trend could likely continue for another decade, at which point there would be a then-incredible number of transistors on their chips. This exponential increase in the number of components would eventually result in the 4004, the world’s first central processing unit (CPU). Developed in 1971 by Intel, which Moore and Robert Noyce co-founded three years earlier, that first CPU had a whopping 2,300 transistors! While that was astounding number at the time, the ongoing doubling — or exponential growth — has resulted in chips today that are packed with literally billions of transistors! (Note; While Moore’s law Initially stated the doubling occurred every year, this frequency would adjust between one and a half and two years for several decades.)
The astounding feats of today’s technologies owe their existence to this extremely prolific trend. The resulting massive growth in computer processing power has transformed our society and our world, propelling us from the industrial age into the ever-advancing digital era we know today. In keynotes, lectures and workshops, I often expound on the amazing power of exponential growth and accelerating progress. But while the accelerating change we see in our world has many drivers, including the self-reinforcing positive feedback loops generated by so many emerging technologies interacting with each other, currently the most fundamental driver behind all of it remains Moore’s law.
In recent years, however, predictions of the end of Moore’s law have been growing more frequent. This is nothing new. Pronouncements of the famous law’s demise have been a perennial event for decades now. This has even led some to people to say that the number of predictions proclaiming the end of Moore’s law doubles every two years. By my count, this might be a severe underestimate.
But all exponential progressions in the physical world do eventually reach their limits and Moore’s law will be no exception. For years, this self-fulfilling business trend (which is what it is) has been slowing, advancing more erratically. Fortunately, there are a number of technological breakthroughs on the horizon that are expected to significantly extend Moore’s law, resulting in as much as 50 times greater density on CPUs over the next decade or more. The Fin field effect transistors (FinFETs) that have driven many of the advances of the past decade still have the potential to triple chip density. After that, nanowire transistors (expected around 2025) and later, stacked nanowire transistors will quadruple density still further. At the same time, a number of innovative software advances could result in chips that are a hundred times faster than those we see today.
This doesn’t even factor for a number of other innovations we’re tracking, such as air-channel transistors (ACTs) that could be used in advanced 3-D chip architectures and potentially extend Moore’s law by another two decades. All told, these many developments could take us well into the next era of computing, one that will see our computers emulating and surpassing human level capabilities with increasing frequency.
It’s important to keep in mind is that Moore’s law is essentially a business plan. The capabilities of each new generation of chips make the succeeding generations possible and as long as the economics pencil out, the next iteration of advances will eventually be developed. While it may seem like we could reach an insurmountable wall any day now, in theory there remain many possible ways of continuing for decades to come. As Nobel laureate Richard Feynman stated in his prescient 1959 lecture, “there’s plenty of room at the bottom.”
As our command of the atomic scale advances, it remains to be seen how this will all play out. But for now, we can almost certainly say that the pattern of advances that have so dramatically transformed our world for over half a century are going to continue for the foreseeable future.
