27-10-2021 | By Robin Mitchell
Recently, Micron announced that it will be investing up to $150 billion over the next decade to expand its fabrication capabilities. What developments in technology have placed unprecedented demand, what will Micron plan to do, and could such expansion give rise to ASIC and other custom semiconductors?
The semiconductor drama
When it comes to drama and media attention, most would immediately think of celebrities and the unusual relationship between various artists. However, the semiconductor industry has become increasingly involved in the world of drama with the shortages in semiconductors, the challenges faced by many industries due to this shortage, and the rising tensions between Taiwan and China.
While the COVID-19 pandemic was the spark that ignited the semiconductor shortages, it certainly was not the cause. Sure, COVID-19 halted large portions of the industry that saw semiconductor manufacturers stop producing key components, but what COVID-19 really did was demonstrate how sensitive the semiconductor industry was to change. Furthermore, it also showed how closely intertwined semiconductors are to everyday life and how semiconductor manufacturers do not have clear communication channels with major businesses that use their products.
These problems compounded when it was revealed that most businesses have now moved to a just-in-time production model. This model is perfectly fine when everything works, but any hiccup in the production line ripples throughout the entire supply chain and can have major knock-on effects.
Micron to invest $150 billion in semiconductor production
Recently, Micron announced that they will be investing over $150 billion over the next decade to step up the production of advanced semiconductors. This investment will be the largest ever made by Micron, and there is talk of Micron approaching various governments for grants to achieve this goal. The need for grants and subsidiaries comes as a result of the expensive nature of producing semiconductors in countries such as the US and UK, which can see a price increase of up to 45% (a result of tax and regulation).
While the current semiconductor shortages drive this expansion, another is the rapid introduction of large-scale computing and AI. Such technologies are highly dependent on processors and memory, and Micron’s new plants will be focused on producing such parts. More specifically, Micron is the world’s third-largest DRAM producer and is also responsible for producing NAND flash memory. However, the $150 billion announcement does not detail exactly how the investment will be split between these various technologies.
Could large-scale semiconductor production encourage custom silicon?
The semiconductor shortage has demonstrated how the entire world is dependent on a handful of nations for advanced technologies. Governments are now considering moves to try and encourage locally produced semiconductor devices with national security in mind.
However, it should be understood that while the most advanced semiconductor companies continue to shrink devices into single nanometer sizes, older technologies (such as 20nm) are still perfectly valid and usable in modern technology. If the semiconductor industry follows the same trends found in almost every other industry, the equipment and costs associated with older semiconductor technologies should start to fall in price.
What would cost billions 10 years ago may now only cost a few hundred million. Combined with the fact that more semiconductor companies are becoming fabless, you have the perfect conditions for smaller-scale semiconductor foundries to start operation.
Large foundries such as those owned by TSMC and Intel are generally interested in large-scale production to reduce the cost of parts, but smaller foundries using pre-established technologies can focus more on smaller production scales and even individual devices. While custom silicon chips already exist (usually used by Universities and research labs), the introduction of more small-scale foundries could encourage engineers towards entirely custom silicon.
Of course, it is also possible that FPGAs and CPLDs could become more commonplace in modern electronics as they enable hardware to be programmable, reduce the overall number of components needed, and allow for upgrades over time. Overall, it is hard to say what the future holds for the semiconductor industry, but personally, I hope that custom silicon becomes an economical option for engineers.