by Scott Hamilton
Senior Expert in Emerging Technology
I have predicted since 2001 that a well-known law in the integrated circuit business was coming to an end very soon. I have argued several times that it, in fact, ended in 2001. This well-known law is called Moore’s law, and it had two parts in its predictions. The first part was the rate at which the number of transistors in a single chip would double and the second was related to the cost of manufacturing the integrated circuits.
Moore’s law predicted that the number of transistors in a single chip would double every two years, and the cost would drop in half in an equal time frame. My argument has been that in relation to microprocessors, the advantages of Moore’s law ended in 2001 with the invention of the first multi-core processors. Prior to 2001 the increase in the number of transistors and equal reduction in the transistor size equated directly to an increase in microprocessor performance; this held true for over 40 years. In 2001 the smaller transistor sizes started to negatively impact microprocessor performance, meaning the advantage came to an end. Intel extended the usefulness of Moore’s law by bending the rules. They started putting more than one microprocessor on a chip. Even though the processor speed remained the same, they claimed a two fold increase in performance. It was several years before operating systems and software took advantage of multi-core technologies, so there was a slow spell in computers for a few years.
In the late 2010s another problem hit Moore’s law, the single layer transistors were now too small to shrink any further. They had reach nearly the size of three silicon atoms wide, effectively meaning that Moore’s law could no longer be in effect; the transistor could get no smaller. Once again technology at Intel changed the rules and they began manufacturing multi-layer transistors, or 3D transistors. This allowed them to cram more transistors in the same space, keeping the first part of Moore’s Law alive. In 2020 or early 2021, another limit was reached and Intel once again modified the transistor, allowing for an even smaller size by stacking transistors rather than layering a singe transistor; they were making multiple layers of transistors. This is where the second part of Moore’s Law began to break down. In 2018 the cost of manufacturing transistors on chips was at a historic low of $1.49 per 100,000; as the newer technology made the manufacturing process more complex this price began to climb. Today it is at nearly $5 per 100,000, which is similar to the price in 2001.
There is little room to argue that Moore’s Law is no longer in effect, and many are predicting that it is the reason for the current chip shortage impacting the auto industry. In the past, retired microprocessor plants were used to manufacture the older technology in the chips for the automotive industry, making them nearly free for the manufacturer to produce. Since new technology development has slowed there are fewer retired plants, causing a rise in price due to limited supply of older chips. Chip manufacturers now have to invest in production infrastructure for both the leading edge chips and the older technology, which has never happened in the history of the industry. This leads me to believe that the decades long boom of the integrated circuit industry is at an end. The Boom has went Bust and we will all be paying for it as the expense of technology increases across the spectrum.
Until next week, stay safe and learn something new.
Scott Hamilton is a Senior Expert in Emerging Technologies at ATOS and can be reached with questions and comments via email to firstname.lastname@example.org or through his website at https://www.techshepherd.org. You can also follow his channel on rumble at https://rumble.com/c/c-1141721.