OPINION:
President Biden’s industrial policies to revitalize domestic manufacturing and meet Chinese competition in Artificial Intelligence recently took a gut punch.
Cutting-edge chips are essential for training and applying AI models like ChatGPT. Beijing has prioritized catching and surpassing the United States, Japan, Korea, the Netherlands and other Western nations that host globally integrated supply chains for the design, tooling, fabrication and packaging of advanced semiconductors.
Intel, the largest recipient of Chips and Science Act grants to rejuvenate the fabrication of leading-edge chips in America, recently reported a $1.6 billion second-quarter loss because its foundry arm is bleeding cash. Now, it’s considering spinning off that manufacturing activity and, like rivals Nvidia, AMD and others, focusing only on chip design and software.
That casts serious doubt on the U.S. goal of achieving 20% of global leading-edge chip production by 2030 and threatens to make the American industry dependent on China for advanced chips one day.
Intel, TSCM and Samsung have been awarded $8.5, 6.6 and $6.4 in grants from the $39 billion authorized by the Chips Act to rejuvenate cutting-edge facilities in the United States.
However, as current U.S. efforts to limit Chinese access to Japanese and Dutch machine tools essential to manufacturing chips illustrate, boosting capacity even among those owned by allies firms creates delicate to manage security vulnerabilities.
Hence, the Department of Defense correctly seeks to focus critical technology purchases on domestically owned suppliers. Intel is the sole recipient of a $3.5 billion Chips Act program to provide advanced semiconductors for military and intelligence applications.
Considering that most significant other high-end foundries are owned by the Taiwan Semiconductor Manufacturing Corporation and Samsung, finding a domestic suitor for Intel’s fabs would be tough.
Competition in semiconductors is central to America’s complex rivalry with China—whose heft is central to the mischief-making of the broader Axis of Autocrats, which includes North Korea, Russia and Iran—because high-end chips, such as those designed by global industry leaders like Nvidia, Qualcomm and Broadcom, are critical to every security and industrial activity where AI is deployed.
American policy falls short in four ways.
First, funding is inadequate. China is offering $144 billion in subsidies, and Beijing can direct bank loans to the industry as it has for other priority activities.
The U.S. advantage is primarily in semiconductor design and R&D, but the Chips Act allocates only $11 billion to this.
Second, the Chips Act fails to address the fundamental problem in the U.S. foundry business—it costs 44% more to manufacture chips in the United States than in Taiwan owing to a myriad of environmental assessments, permitting procedures, community engagement requirements and mandates to employ union labor.
The Biden administration exacerbates these by requiring firms receiving Chips Act and Inflation Reduction Act assistance to fund childcare, negotiate project labor agreements, invest in K-12 and higher education, set aside employment and training opportunities for underrepresented groups and serve other social justice initiatives.
Third, Intel’s losses partly stem from the cyclical nature of semiconductor demand. As the weaker of the major fabricators, its vulnerability is exposed when the U.S. and Chinese economies slow, especially with the EU economy caught in a funk.
Intel would benefit from industrial alliances that recognize its potential role in an overall AI supply chain that included, in addition to Nvidia’s high-end processors, a broader range of semiconductors, energy infrastructure, data centers, software, and the like that could carry it through periods of globally slack demand.
Sam Altman is seeking support from governments and sovereign wealth funds in North America, Japan and the Middle East and major businesses like Nvidia, Microsoft and Apple to create an integrated supply matrix that will effectively compete with China.
Fourth, that initiative raises antitrust concerns, but administration competition policymakers are often motivated by political objectives misaligned with more salient American industrial policy goals.
Nippon Steel’s proposed acquisition of U.S. Steel would inject billions in new capital and advanced technology to preserve manufacturing jobs. However, to appease organized labor, the Biden administration opposed the deal in favor of a tie-up with Cleveland-Cliffs, even though it would control 100% of U.S. capacity to make steel for electric vehicle motors.
Nvidia is reaching the limits of its market share in high-end AI processors and is seeking to expand into related software, for example, by acquiring Run: AI.
This is consistent with the Altman vision, yet the Justice Department is in the early stages of investigating the tie-up.
An effective semiconductor policy should give broad deference to integrating initiatives like Mr. Altman’s and refrain from the political impulse to divvy up the spoils for political constituencies to ensure a prominent place for U.S. chip manufacturing in global markets and stay ahead of China.
• Peter Morici is an economist and emeritus business professor at the University of Maryland, and a national columnist.
Please read our comment policy before commenting.