Pages

11 November 2022

Moore’s Law and Its Practical Implications

Gregory Arcuri

Q1: What is Moore’s Law?

A1: While popularly referred to as a “law,” Moore’s Law is better understood as an empirical observation regarding advancements in computing. In a 1965 Electronics Magazine article, the cofounder of Fairchild Semiconductor International, Inc. and Intel, Gordon Moore, projected that the ideal number of transistors per square inch on a microchip would double each year while the manufacturing cost per component would halve. Ten years later, Moore revised his original projection and said chip density would, instead, double every two years for at least the next decade.

More transistors and components, in layman’s terms, means more computing power, higher efficiency, and more complex functions. A corollary of Moore’s Law is that the cost of computing has fallen dramatically, enabling adoption of semiconductors across a wide span of technologies. Today, semiconductors are the technology platform underpinning how the world works, communicates, and consumes.

Q2: Is Moore’s Law still viable?

A2: Moore’s Law has largely held true into the twenty-first century, though it has begun to slow down as engineers reach the limits of shrinking circuits within the laws of physics. Even so, the computing power of a single integrated circuit today is roughly 2 billion times what it was in 1960.

As the exponential increase in the density of transistors per square inch on a chip decelerates, some observers have proclaimed Moore’s Law demise. However, advances in chip packaging and design may allow a form of Moore’s Law to “survive” into the 2020s. In 1995, Moore himself admitted that “the definition of ‘Moore’s Law’ has broadened to refer to anything related to the semiconductor industry that, when plotted on semi-log paper, approximates a straight line.” A reconceptualization of Moore’s Law—sometimes dubbed “more than Moore” or MtM—prioritizes system complexity over chip density as a more accurate path for progress in computing technology and has extended the continued viability of Moore’s Law.

As of 2022, advancements along the lines of MtM, including the advent of the so-called three-dimensional integrated circuit (3DIC), heterogenous integration, and “chip stacking,” as well as the potential for quantum-enabled semiconductors, may hold the key to the persistence or even acceleration of Moore’s Law—albeit in different form—well into the twenty-first century.

Q3: What are the practical implications of Moore’s Law for the U.S. economy and for U.S. policymakers?

A3: Moore’s Law drives innovation-based competition within the semiconductor industry. With Moore’s Law setting the pace, semiconductor firms understand that they have a predictable timetable—roughly two years—for when they must conceive the next generation in functionality before their competitors almost inevitably will; they must either keep up or lose out. Annual roadmaps— set by the industry itself—enable firms to meet this pace of innovation.

This pace of innovation in the semiconductor industry also drives innovation in industries powered by semiconductor chips. This includes the consumer electronics sector that produces the latest smartphones and laptops, and emerging technologies such as artificial intelligence (AI) and machine learning, cloud computing, internet of things (IoT) infrastructure, and next-generation telecommunications. Advances in chips also facilitate breakthroughs in emerging and precommercial industries like quantum information science and quantum computing—which will, in turn, enable more advanced chips.

However, like all commercial and scientific endeavors, adherence to Moore’s Law in the semiconductor industry relies on significant investment. In fact, each doubling of chip computing power requires a parallel doubling in capital investment. While annual increases in research and development (R&D) expenditure within the semiconductor industry since 2000 have been relatively modest, the struggle to maintain production in an increasing number of legacy or trailing-edge nodes (older chips) while continuing to compete and manufacture at the leading-edge at-scale carries significant costs. In 2021, the semiconductor industry collectively spent $146 billion in capital expenditure, 60 percent of which came from the world’s three biggest chipmakers: Intel, Samsung, and TSMC.

While the astonishing growth of the semiconductor market—especially since 2020—has recovered these expenses, each semiconductor firm must participate in, and benefit from that growth for the investments to remain economically viable. According to the Advanced Electronics Practice at McKinsey & Co., the semiconductor market follows a winner-takes-all model: “If a company’s product or service is even slightly better than a competitor’s, it typically captures an outsize portion—or even the vast majority—of industry revenue.”

This means that a country whose semiconductor industry falls behind in both the investment and technological leadership needed to keep up with Moore’s Law can cede the related downstream economic benefits and the national security advantages that come from that industrial leadership. “Catching up” after falling behind, therefore, becomes an impossible proposition, as the required iterative investment must be made after forfeiting the previous generation’s profits to a competitor. With each missed opportunity, the price tag for regaining leadership grows exponentially higher.

Q4: Where does the United States stand and what is the path forward?

A4: The United States has historically been an active participant in Moore’s Law advancements. Silicon Valley was the birthplace of the first transistor; many of the semiconductor industry’s pioneering firms such as Intel, Texas Instruments, IBM, and others were U.S.-based and, even until 1990, the United States alone was home to 40 percent of the world’s semiconductor manufacturing capacity.

Today, the United States is still a major player in the semiconductor industry, with 7 of the top 10 chip design firms by annual revenue headquartered in the United States. This leadership is a critical national asset, with chip design constituting the highest return-on-investment niche within the industry.

However, as in other industry sectors, U.S. leadership in semiconductor manufacturing has atrophied over the past three decades. Global market trends began to favor specialization and the outsourcing of certain processes overseas, particularly chip fabrication. East Asian firms in Taiwan, South Korea, and China were the major beneficiaries, with that region now accounting for nearly 80 percent of global chip fabrication capacity. According to industry veteran Richard Elkus, Jr., by relinquishing leadership in chip manufacturing, the United States has ceded revenue critical to financing continued investment in Moore’s Law advancements.

This realization comes amid growing concerns about the stagnation of the U.S. innovation system. Research conducted by experts at Duke University found that total factor productivity growth in the United States—a metric driven largely by innovation—has decelerated over the years despite steady annual increases in basic science R&D spending and the number of science and engineering PhD graduates. While the factors behind this trend are manifold, its consequences point to a vulnerability: if the U.S. innovation economy is not as dynamic or competitive as it once was, its ability to translate increased investment into technological progress and—by extension—its ability to secure U.S. participation in Moore’s Law is attenuated.

Q5: What are the consequences of the CHIPS and Science Act?

A5: With the passing of the CHIPS and Science Act (also known as CHIPS+), policymakers have taken the first critical step in addressing the nation’s ability to maintain Moore’s Law leadership. Incentives for constructing semiconductor fabs on U.S. soil are already enticing major chip companies to make substantial investments in facilities across the United States, including Intel in Ohio, TSMC in Arizona, and Samsung in Texas.

These investments, in addition to creating thousands of high-skilled, well-paying jobs, build local manufacturing capacity and renew the nation’s semiconductor innovation ecosystem. Likewise, the significant R&D initiatives authorized in Division B of CHIPS+–which would add up to around $200 billion if fully funded—constitute a major federal commitment to revitalizing U.S. science and innovation infrastructure. Even so, other nations are forging ahead with their investments. TSMC alone plans to spend $100 billion over the next three years to expand its chip fabrication capacity in Taiwan.

If implemented effectively, the various CHIPS+ programs will ensure that the United States stays ahead of the Moore’s Law curve—along with the related downstream economic and strategic benefits. However, such an outcome is still contingent upon sustained federal support for U.S. technological leadership. While the passage CHIPS+ was a significant start, it represents only the beginning of what should be a long-term U.S. strategy for facilitating a more competitive innovation system and a vibrant domestic semiconductor industry.

No comments:

Post a Comment