INFER announced a strategic focus on microelectronics in April 2022 and launched the Microelectronics Technology topic featuring 13 questions. We also hosted our first Fireside Chat with a leading researcher and professor in the field, who discussed how the U.S. could reassert leadership in this space. This “topic brief” aims to give everyone (especially non-experts) a starting point into forecasting on questions pertaining to microelectronics. We will first discuss how microelectronics fits into our broader focus of assessing future global competitiveness in AI. Then, we will break down key terms and resources to help you understand how microelectronics are manufactured, trends, and major players. Read through each section in order or skip around:
- What is microelectronics?
- Why does microelectronics technology matter for the development of AI?
- What are some potential implications of advanced microelectronic technology?
- Key terms and resources: microchips, lithography machines, and semiconductors
— — —
1. What is microelectronics?
Microelectronics is a subfield of electronics focused on the study, design, and use of tiny complex machines or “microchips” (defined further down in this post) that allow most devices to function. Microelectronics are used in many personal electronic devices, such as smartphones, computers, and gaming consoles, and they have permeated many critical areas in medical, aerospace, and military applications.
2. Why does microelectronics technology matter for the development of AI?
Microelectronics serve a critical role for innovation in AI. The most obvious connection is that AI requires hardware, such as microchips, to run. What can be done in the future with AI depends on how microchips evolve to support it. Understanding the future capabilities of microchips allows us to better forecast what the “art of the possible” will be.
Modern and more advanced AI techniques require smaller and faster microchips referred to as “AI chips,” as explained by Georgetown University’s Center for Security and Emerging Technology in
AI Chips: What They Are and Why They Matter:
“The success of modern AI techniques relies on computation on a scale unimaginable even a few years ago. Training a leading AI algorithm can require a month of computing time and cost $100 million. This enormous computational power is delivered by computer chips that not only pack the maximum number of transistors—basic computational devices that can be switched between on (1) and off (0) states—but also are tailor-made to efficiently perform specific calculations required by AI systems.”
As the computational power of microchips increases, this will continue to drive AI progress. According to an
Open AI analysis, since 2012, the amount of computational power used in the largest AI training has increased exponentially, doubling every 3.4 months. With even more advanced microchips, the capabilities of machine learning and AI will be far outside of what can be done today. For example, whereas ten years ago we were just beginning to create natural language processing tools, we now have AIs that are
better readers than some students at Stanford! Current AIs are so strong, that
they’re writing their own novels that could rival top literature within the next decade.
3. What are some potential implications of advanced microelectronic technology?
Besides the topic’s obvious connection to AI, another reason that INFER is focusing on microelectronics is due to some of the policy and security implications that are tangentially related to the advancement of AI. For example, the demand for trusted microelectronics (those that can be verified to not have nefarious exploitations by competitors) will grow as public and private groups continue to incorporate AI into mission-critical systems. Will we be able to have trusted microelectronics at scale? What may have to be done to achieve this? Currently, there is only one company,
ASML, that the world relies on for the production of “lithography machines” – the devices used to create microchips. Understanding more about the future of trusted sources will be critical in helping us understand if our AI applications can be powered in a secure way.
Another example, all this computing produces an enormous amount of data. It’s estimated that by 2025, the world will produce
roughly 163 trillion gigabytes of information – where will we store that data? Some microchips are designed to hold large amounts of data, but can manufacturing keep up? Perhaps most importantly, can we do it in a way that is accessible to those needing to use that information?
4. Key terms and resources: microchips, lithography machines, and semiconductors
This section goes into relevant terms and resources to help you understand how microelectronics are manufactured, trends in the field, and major players. After going through each section, we encourage you to try forecasting on relevant questions within INFER’s Microelectronics topic.
Microchips
You may already know that microchips power computers and the AI algorithms that run on those computers. But what exactly are they – and what do they do? Microchips are physical devices you can hold in the palm of your hand (not to be confused with microelectronics, which refers to the field or industry that studies them). Microchips (or “chips”) are units of packaged computer circuitry (also called integrated circuits) usually made from silicon. They come in the form of logic or microprocessing chips that process information to complete a task, or memory or RAM chips that store data.
How are microchips evolving? As demand for increased chip performance and efficiency continues to rise, manufacturers have turned their attention to making smaller chips. With smaller chips, you can fit more of them in a limited space, thus dramatically increasing performance. The microchips currently being produced range in size from 5 nanometers to 10 nanometers. IBM
recently announced the creation of a 2-nanometer chip with smaller, more advanced processors. Although not yet in volume production, 2-nanometer chips
could quadruple cell phone battery life, speed up a laptop drastically, or even help decrease reaction time in autonomous vehicles.
Additional resources on microchips:
Lithography machines
What are lithography machines and what is their importance? Lithography (or photolithography) machines are one of the core pieces of technology in chip manufacturing. They etch the integrated circuits found within the chips. Mass production of extreme ultraviolet lithography machines – the most efficient kind of machine –
has only come about in the past five years, even though lithography technology
has been in development for roughly 40 years.
Who are the key players in developing these machines? ASML, a company in the Netherlands, is the largest maker in the world of lithography machines, providing them to the most advanced chip makers. ASML developed
extreme ultraviolet (EUV) lithography machines that can achieve unprecedented precision. Due to its price tag of $150 million per machine,
only a few of the top chipmakers can afford them: Intel (U.S.), Samsung (South Korea), and the Taiwan Semiconductor Manufacturing Company Ltd. (Taiwan).
Additional resources on lithography machines:
Semiconductors
What is the relationship between microchips and semiconductors? Semiconductors are mostly silicon materials that conduct electricity in some cases but not in others, a property ideal for controlling electrical circuits and why they are used to create a majority of microchips. When we talk about AI or any advanced technology, we are solely focused on semiconductor-based microchips (also called “semiconductor chips”). Semiconductors are both thin and lightweight (good for use in smaller devices) while also not compromising processing power. The name for Silicon Valley actually comes from the fact that silicon is the most popular and prolific commercial semiconductor used in modern microprocessing chips.
Who are the major players in the semiconductor industry? The semiconductor industry is the aggregate of the companies engaged in the design and fabrication of semiconductors and semiconductor devices, such as integrated circuits and chips. The industry is dominated by companies in the U.S. and China, and trade disputes between the two giants have repercussions for both international and domestic industries.
Other countries are leaders in the manufacturing of chips. That sector is led by companies from the U.S. and Southeast Asia, the largest of which are Intel (U.S.), Samsung (South Korea), and the Taiwan Semiconductor Manufacturing Company Ltd. (Taiwan). Amid the semiconductor chip shortage in 2021, spurred by the global COVID-19 pandemic, governments around the world realized Taiwan’s outsized role in chip manufacturing: More than 90% of the world’s most advanced chips are made in Taiwan. Now other countries, including the U.S., are addressing those vulnerabilities and increasing investments in semiconductors to ramp up manufacturing capability.
Additional resources on the semiconductor industry:
Thank you for reading, and happy forecasting!