Semiconductors: Let's talk

Editorials featured in the Forum section are solely the opinions of their individual authors.

If we take a step back and analyze the pandemic-accelerated supply chain shocks, we can see that they were inevitable. In the specific context of semiconductors (integrated circuits/transistors/processing units), the disruptions to the fabrication process have demonstrated the need to explicitly define the intersection of innovation and interests. The discussion needs to shift from “semiconductors are vital to everyday life” toward a more grounded realism. Every step of the fabrication process, from the design to diffusion, presents unique challenges that require more nuanced discourse in the public policy sphere.

Investment in more advanced domestic semiconductor fabrication should be mainstream, particularly when discussing strategic competition with countries like the People’s Republic of China (PRC). Trade wars and military buildup do not exist in a vacuum of instrumental modality. Semiconductors, both their capabilities and security, are vital components of communications networks, imagery and reconnaissance satellites, threat detection networks, naval ships, and of course, smart refrigerators. For this reason, the Department of Defense awarded Intel a contract to develop, in conjunction with the Defense Advanced Research Projects Agency (DARPA), new structured application-specific integrated circuits for more efficient and secure military-use microelectronics.

However, that partnership represents only a fraction of the total amount of secure microelectronics and networks that the military, and other agencies, utilize. Taiwan Semiconductor Manufacturing Company (TSMC) produces the most advanced processors that guide Lockheed Martin’s F-35 Lighting II and supplies other manufacturers such as Xilinx (freshly acquired by AMD). For several CPU architecture generations now, Intel has been falling behind in node technology compared to TSMC, specifically for desktop and server-grade processors. As modern capabilities require more computational power, strategic competition with the PRC continues to unfold in the military technology space.

It is critical that we continue investment in more advanced semiconductor technology to stay ahead of hardware vulnerabilities that stagnant development presents. In 2018, the public became aware of security vulnerabilities Meltdown and Spectre, raising discussions about the penetration process and payload type. While the impact of these vulnerabilities is undetermined, it is well-known that malicious actors can exploit a processor’s internal buffers to extract encrypted information and keys. Aging military technology and the vast quantities of equipment that can store encrypted information mean that potential targets and exploits are likely more numerous than mainstream discourse suggests.

Within this context, I argue that current partnership programs and Congressional funding for domestic semiconductor fabrication are woefully inadequate in addressing the impact of sourcing the majority of electronic and circuit components from Taiwan, which has been subject to constant encroachment by mainland China. The Creating Helpful Incentives to Produce Semiconductors (CHIPS) for America Act fails to address the timeframe in which its lofty ambitions fall subject. Although passed by Congress last January, the act has yet to authorize any funding for the programs laid out in its mandates. As I mentioned previously, Intel is years behind the bleeding-edge node technology that TSMC has made readily available at the consumer level. For example, Intel has finally transitioned from nearly a decade of 14 nanometer FinFET technology to 10 nanometer (which is conveniently labeled Intel 7) whilst TSMC enters the final phases of its risk production of three nanometer node technology. Intel designs and manufactures its own semiconductors, whilst TSMC supplies the physical die to fabless companies like Advanced Micro Devices (AMD). All these discussions continue as Congress has yet to agree on how to actually allocate funds to the programs designed to alleviate the issues I highlight.

The one element that permits this lengthy and ongoing discussion is the fact that China has yet to successfully produce highly-advanced semiconductor technology domestically. It has largely been forced to import the more advanced processors from the US and has significantly invested into the design, testing, and diffusion of semiconductors that are beyond the low-level logic chips that it currently produces in bulk. And yet, that certainly is not to say that the current state of domestic semiconductor production in the U.S. can continue indefinitely. It is time for Washington to seriously consider the future of its capabilities through the lithography of advanced computing.