US CHIPS Act is only a small step toward a very big problem

The passage of the CHIPS and Science Act of 2022 represents a major U.S. government intervention in industrial development and economic policy — particularly the semiconductor sector. It promises to inject $52.7 billion over the next five years to stimulate a return to national manufacturing, which has been gradually lost over the past half century. While strong in chip design, the U.S. manufactures only about 12% of the world's processors, with East Asia now responsible for three-quarters of worldwide production. In addition to attempting to strengthen manufacturing, the CHIPS Act is intended to help fix supply chain weaknesses that have plagued chip-dependent industries — particularly automotive, but also computers, smartphones, and defense — and to bolster national security.

 In some respects, the passing of the bill is already showing some results, with commitments from various companies to construct or expand a half dozen or more chip fabrication plants in the U.S. These projects probably wouldn't have gone ahead without the stimulus, or the investments would have been made abroad, actively encouraged by foreign governments. There are also plenty of challenges that the CHIPS Act can't directly address — a complex brew of economic factors, logistical bottlenecks and disruptions, trade wars, shifting geopolitics and (not least) technical barriers — that could mitigate any positive impacts attributable to the Act.

The semiconductor sector has become unbalanced. Wide-ranging industries, especially those that rely on just-in-time production strategies, have been severely impacted by supply chain issues. There is surely justification for wanting to strengthen U.S. chip manufacturing and packaging. The problem is the complexity of doing so. Few believe the federal government has enough working knowledge of this industry to effectively pick winners and losers in the semiconductor supply chain and ensure the investments are made in the most critical areas. (Remember solar energy company Solyndra?)

 Protectionist conditions attached to the subsidies might force the chip industry to restrict overseas partnerships and sales. If U.S. chip companies can no longer sell to China, they miss out on the largest market for their products. (Of course, such restrictions may not be enforced by a globally connected U.S. Congress). And given the massive rates of spending required to build and maintain chip plants, $52.7 billion is nowhere near enough to make a meaningful difference in the long term.

Supply chain, chip design presents more challenges

Perhaps the two most worrying assumptions behind the CHIPS Act are that a return to local manufacturing will address the current shortage of silicon — a factor that has surely helped drive its passage through Congress after a year's delay — and that manufacturing is somehow more important than design. The latter is a refutation of the long-established and highly successful fabless semiconductor model, through which chip designers outsource the manufacturing process to highly specialized and highly efficient chip fabrication plants, most of them in Taiwan.

 A return to local chip manufacturing won't fix the chip shortage because supply chains are now far too complex and international in nature. Vendor concentration will remain a huge issue in the foreseeable future. The largest chip fabrication plants are mainly in Taiwan, which has unique geopolitical concerns. While that's a central justification for building out capacity elsewhere, the chasm in levels of investment remains huge. TSMC on its own is investing $44 billion this year on new manufacturing facilities in Taiwan. The proposed TSMC Arizona factory will likely represent about 1% of its global capacity. Even then, once chips are made in the U.S., they will have to be shipped back to Asia for testing and packaging, and then to China for assembly into smartphones and PCs.

This leads to the second concern: Is manufacturing really more important than design? Some of the most successful chip companies in the business — Advanced Micro Devices Inc., Apple Inc., Arm Holdings, Broadcom Inc., Marvell Technology Inc., Nvidia Corp. and Qualcomm, not to mention cloud companies with their own cloud-native chips and all the AI chip startups that have emerged over the past decade — have deliberately outsourced all manufacturing, not wishing (or not able) to invest massive amounts of capital in manufacturing themselves. Aside from Qualcomm, which has aligned itself with GlobalFoundries for a portion of its future output, these companies have shown little enthusiasm for the CHIPS Act. Some have been lobbying for equivalent subsidies and tax breaks for chip designers.

CHIPS Act will offer uneven benefits

It is hard to argue that the CHIPS Act won't be beneficial to the U.S. semiconductor industry, creating many more jobs and spurring additional investments in the sector that would otherwise not have happened, or would have happened offshore. However, the advantage will be spread unevenly, not helping some of the most innovative companies in the market, while shoring up some of those that have fallen behind. TSMC and Samsung, already dominant chip manufacturers worldwide, will win subsidies to expand onto U.S. soil, but their primary activities and skillsets will remain elsewhere. Chip testing, assembly and packaging are unlikely to return to the U.S. in any form that will make them economically viable for volume business.

 Meanwhile, the cyclical nature of the chip industry continues independently of the new funding. We now appear to be entering a downturn in demand, as the buying frenzy of consumer technology during the lockdown ebbs. In a year or two, we are likely to see a glut of capacity, at least temporarily, which will test the resolve of those companies that have committed to expensive buildouts.

 Over the long term, worries about continuity of supply will likely continue as geopolitical tensions intensify. These tensions could fuel further national security concerns, stall international scientific collaboration, and potentially affect the establishment and development of international standards, which can easily become politicized. Then there is climate change, stretching fragile supply chains to their limit and requiring a radical rethink of chip manufacturing processes. For example, can the state of Texas supply enough water to serve Samsung's 11 new plants?

It is good that the U.S. government has woken up to the fact that the semiconductor industry is vital to its future and must be supported, but has it fully grasped the scale of the problems? Likely not. Is it willing to invest much more extensively in the future to achieve at least some of its goals? Perhaps, but only if there is also a clear advantage to be gained.