Chips on the Table

This article appears in the February 2025 issue of The American Prospect magazine. Subscribe here.

During the 2020s, semiconductor production returned to the forefront of U.S. industrial-policy objectives, with President Biden’s administration determined to get the country that invented the technology back to manufacturing it. The CHIPS and Science Act was conceived with two main goals: revitalize the domestic semiconductor industry, and bring supply chains back to the U.S. Passed in 2022, the act committed to a one-time payment of $52 billion: $39 billion for “manufacturing incentives” and $13 billion for research and development.

The act has, for the most part, been greeted with enthusiasm, and there is some cause for optimism. U.S. government investment in semiconductor manufacturing in 2024 will surpass total expenditures over the previous 27 years. CHIPS money has been decently distributed beyond the top players like Intel, Samsung, and TSMC. And although semiconductors have long been central to the U.S. military’s vision for dominance, the CHIPS Act is run through a civilian agency—the National Institute of Standards and Technology (NIST) at the Department of Commerce—rather than the Department of Defense.

The CHIPS Act and other Biden initiatives also sought to leverage public investment to make advances in child care, community benefits, profit-sharing, employment for minority populations in economically distressed areas, and the inclusion of labor priorities. For these reasons, its architects have branded the full suite of Bidenomics policies as heralding a restoration of public investment and the end of neoliberalism.

Figures on the Tech Right, who are deeply embedded in Trumpworld, are hoping for escalation with China.

Many Biden administration policies, like its investments in clean energy, had success in increasing homegrown production capacity and creating union jobs. But the CHIPS effort in particular harks back to the industrial policy we’ve seen in the U.S. since World War II, closely tied in to national security. As computer pioneer Grady Booch has frequently stated, all computing is “woven on the loom of sorrow.”

Forged by and for war, computer technologies like semiconductors have never escaped the logic of military Keynesianism, binding the U.S. state and capital more tightly together, producing endless variations of public-private cooperation. And the bipartisan coalition for tech policy interventions like CHIPS always hinges on the imagined capacities of an enemy nation. During the Cold War, it was the USSR; during the 1980s and 1990s, Japan; presently, it is China.

Biden’s anti-neoliberals oversold the progressive nature of the CHIPS Act, whose solutions in practice make little break with the technocratic and anti-democratic methods of neoliberalism. The goals of solving climate change, restoring the middle class, strengthening supply chains, regaining control of leading-edge and legacy chips, and repairing our weakened democracy were not backed up by incentives or enforcement.

Moreover, the act has begun the process of shifting defense money from traditional D.C.-based defense contractors to a newer, venture capital–startup alternative based in the West. As Silicon Valley edges further into military contracting and assumes positions of power in the second Trump administration, the future of CHIPS could wander even further from how it was initially sold.

Making and Remaking a Global Order Against China

In 1984, U.S. companies like Intel, which had long enjoyed a huge market share in semiconductor production, suddenly lost ground to Japanese competitors. Many feared that the U.S. would lose control of semiconductors, the same way they had “lost” the automobile industry to Japan. The Reagan administration responded with political might, even briefly imposing sanctions on Japan for dumping chips on the U.S. market, and demanding that the country purchase U.S.-made chips.

A public-private semiconductor partnership called SEMATECH was founded in the 1980s to collaborate on research and development and revive U.S. manufacturing. But SEMATECH was an old boys club comprising the most historically powerful companies. Because those companies did not want to benefit competitors, SEMATECH refused to test new prototypes. This refusal led directly to the success of the Interuniversity Microelectronics Centre (IMEC) in Belgium, which became the indispensable testing bed for new semiconductor technology.

The United States’ other semiconductor strategy was to inaugurate a new global order built around distributed chip production. As Chris Miller describes in Chip War, the industry boosted firms that filled a vacuum in East Asia after the wars in Vietnam and Korea. Enterprising businessmen, who often got their start in American semiconductor firms in the 1970s and ’80s, brought their expertise home to Korea and Taiwan. The U.S. encouraged these upstarts to undermine Japan.

As the Japanese threat faded in the late 1990s, Clinton officials inaugurated trade regimes that allowed then-dominant U.S. high-tech firms maximal access to foreign markets. Through luck, savvy, and U.S. hubris, Korean and Taiwanese firms gained on their American competitors. Packaging and assembly moved to nearby locations like Malaysia, Vietnam, China, and Indonesia, where labor was cheap and manufacturing well subsidized. U.S. politicos saw only upside; globalization was a way to spread development, wealth, and stability while ensuring a supply of cheap goods to U.S. consumers.

The CHIPS Act and other semiconductor-related policies thus build on a long-standing feature of U.S. policymaking. They aim not only to reshore leading-edge capacity, but to knit together a new world order through trade, labor, and industrial policies. The older order was aimed at countering Japan; this new vision aims to counter China. The chip industry as a whole is not going to be reshored absent a radical shift. While the U.S. is paying to bring some semiconductor production stateside, the real goal of these policies is familiar: to distribute production among allies to ensure nominal U.S. control, even if it remains abroad. This is sometimes given the anodyne-sounding moniker “friendshoring.”

But it’s not just about investment. The strategy includes extensive export controls on advanced chips and the means to produce them, limiting technology transfer to China, and much more. These measures have not reached the level of belligerence displayed by the Reagan administration, and are less effective than Reagan’s measures, perhaps due to the U.S. being in a less hegemonic position than it was in the 1980s. This ineffectiveness has not deterred their use: U.S. trade representative Katherine Tai claimed that “we really haven’t seen the PRC make any changes to its fundamental systemic structural policies that would make sense for us to provide any relaxation.” Yet there is little cause to believe this might moderate under a Trump presidency, because the effort is bipartisan.

A Defense-Industrial Anti-Neoliberalism?

Policymakers following distorted notions of industrial policy as economic warfare by other means carry echoes of a much older defense-industrial-complex critique of 1980s neoliberalism. Influential ex–Lockheed Martin CEO Norman Augustine blamed the collapse and consolidation of defense firms on 1980s financialization, not 1990s defense cuts. Going back to the early 2000s, Augustine and his allies criticized offshore production and the elevation of China; Augustine likewise was crucial in founding In-Q-Tel, the CIA venture capital firm responsible for Palantir, Anduril, Skydio, and others.

This defense industry critique of neoliberalism attracted influential staffers to Senate Majority Leader Chuck Schumer, a group of State Department figures associated with Biden national security adviser Jake Sullivan, and the new tech defense right. CHIPS itself was molded from two bipartisan bills, the Endless Frontier Act and the U.S. Innovation and Competition Act, with Schumer and several leading Senate Republicans functioning as one of many mouthpieces for the defense-industrial consensus that formed in the late 2000s.

Sullivan allies were crucial in building a broader coalition around the legislation, bringing together climate activists, anti–Wall Street campaigners, organized labor, and the New Right. They have sprinkled money around from philanthropic groups to cement their narrative, and even convinced skeptical left-wing critics like Quinn Slobodian to praise the shift to a more confrontational stance against China.

This faction gained influence in the Democratic Party in the wake of Hillary Clinton’s surprise defeat. They believe in the “China shock” theory of Clinton’s loss: that a disgruntled Rust Belt working class, angry at the Democrats for trade policies that exposed them to foreign competition, rebelled and switched sides. In practice, the theory has merged with an increasing hawkishness toward China, which Sullivan ally Jennifer Harris has said is a threat that “lack[s] historical precedent … [because] the Soviets were never an economic match for the US.” Given the fact that U.S. Cold War mistakes and unnecessary escalations cost millions of lives, this is a weighty claim.

Certainly, removing trade barriers with China led to job loss across many industries in the U.S. The “China shock” narrative, however, offers little in the way of explanation for why leading-edge semiconductor manufacturing went abroad. Nor does it justify new export controls. Labor costs are minimal in leading-edge factories because they are largely automated. Their offshoring, instead, was a product of historical contingencies, the short-term priorities of new MBA-CEOs at Intel and other firms, and U.S. government hubris.

A new Tech Right also claims ownership over the bill. Some conservative groups are much friendlier to industrial policy and state intervention than the paleoconservative right was at the time of Reagan’s intervention into semiconductor production in the 1980s. But the emergence of a Tech Right carries aims that are increasingly anti-democratic and militaristic.

Tech industry oligarchs like Eric Schmidt and Peter Thiel, as well as figures at think tanks like the Manhattan Institute, have a different vision for America’s new industrial policy. Like the Sullivan wing, they want to “end neoliberalism,” but by binding tech companies more closely to the state, eschewing “soft” products like advertising technology in favor of “hard” defense products. Tech companies have more or less always been defense contractors, and certainly function that way today. The heightened focus on semiconductors, advanced manufacturing, and AI functions as an excuse to shift money and power from traditional defense outlets to their own.

Barack Obama, who had very warm relations with tech moguls, brought Schmidt into government advisory and policymaking roles. Government venture capital firms like In-Q-Tel helped steer Silicon Valley funders in the direction of defense by providing long-term lucrative government contracts to their chosen startups. Palantir, Anduril, and other new tech-defense hybrids now see prodigious opportunities for profit in a revived defense focus under Trump. The Financial Times recently reported that the two firms plan to join forces with a dozen competitors to bid on $850 billion in military contracts, to “disrupt” the incumbents. Private equity and venture capital have reasonably concluded government-funded defense firms are a good bet.

Figures on the Tech Right, who are deeply embedded in Trumpworld, are hoping for escalation with China. At a presentation for the now-defunct Silicon Valley Bank by America’s Frontier Fund (AFF), a Thiel and Schmidt venture capital firm, their spokesperson described how they aimed to “profit from several ‘choke points’ in semiconductor and rare earth mineral supply chains.” They also boasted that AFF’s investments would increase “10x overnight, like no question about it” if “the China/Taiwan situation happens” or, more generally, “if there is a kinetic event in the Pacific.”

Schmidt himself has claimed that his underlings helped write significant sections of the CHIPS and Science Act, and his spokesman has also described how government subsidies for priorities like semiconductors are a “fourth pillar” of AFF’s business. These claims should be taken with a serious grain of salt as they are bragging to potential investors, but it’s clear that the tech sector has been prominent in the defense-heavy tilt to semiconductor policy.

The Infrastructure of CHIPS Implementation

A vast array of new industrial-policy subagencies were created to do the practical work of distributing money and crafting regulations and agreements with industry. These include the new CHIPS Program Office at NIST (housed in the Commerce Department), the Undersecretary of Energy for Infrastructure, the Energy Department’s Office of Critical and Emerging Technologies, the Assistant Secretary of Defense for Industrial Base Policy, and the Technology, Innovation and Partnerships (TIP) directorate at the National Science Foundation.

We interviewed several agency participants who were largely optimistic about the CHIPS Act and the “sea change in government philosophy” at work in Biden’s industrial-policy efforts. In our interviews, agency actors were, like Schumer, the Sullivan wing, and the new Tech Right, focused on competition with China.

To the chagrin of some progressives, many of the CHIPS-related offices are, as Dan Kim of the CHIPS Program Office stated, “unique in government … because most people come from the private sector,” specifically Wall Street firms like McKinsey, Goldman Sachs, and Blackstone. Many of these new bureaucrats are motivated in their approach by high-profile failures of the past, where technology was originated but not sustained in the U.S. Likewise, the belief that previous administrations were just “throwing money out” without accountability colors the design of the programs.

CHIPS applications mention community benefits, local involvement in planning, union participation, worker training, environmental stewardship, reinvestment for “excess” profit-sharing, and supply chain resiliency. But the agreements remain noncommittal: They require a plan for profit reinvestment and child care, but not their ultimate provision.

The agencies wanted to create long-term partnerships with carrots and not sticks; to shape but not control U.S. industry. As they see it, these new initiatives are not government capture by industry, but the opposite. But their model builds venture capital further into U.S. industrial policy, offering them even more subsidies and derisking, while asking for little in return.

Environment: Toxic Chemicals and Carbon Emissions

Public discussions around CHIPS have almost entirely ignored environmental problems, which have been at the center of labor and community organizing around electronics since the 1980s, both in the U.S. and abroad. Samsung in Korea and Vietnam has faced considerable criticism and activism around the hazards it knowingly exposed workers to, the most important of these being toxic chemicals like PFAS used in the production of semiconductors and the energy that powers it. Many workers have either contracted cancer or incurred serious reproductive harms. In Santa Clara County, California, alone, there are 23 Superfund sites from the historic production of semiconductor chips. The state has passed special toxic gas ordinances as a result.

Until recently, the CHIPS Office viewed environmental regulations in terms of potential delays to implementation, and has assembled a team to avoid “roadblocks” that would slow down potential projects. Last October, Congress passed the Building Chips in America Act, which adds exemptions from federal environmental review for every project receiving CHIPS money.

CHIPS agencies have indicated they are committed to making semiconductor fabrication plants (known as fabs) more efficient in energy and water use, making the projects cheaper. Likewise, application guidelines express a preference for projects that are net-zero; applicants must “submit a climate and environmental responsibility plan.”

Rather than focusing on high-quality jobs, agencies in charge of CHIPS have advanced industry priorities like workforce development.

But despite claims and pledges of net-zero energy use, this does not seem to be the case for several planned fabs. New York, for example, encourages Micron to source clean energy, but local grid operators are concerned about where this will come from. Commitments to renewable energy frequently rely on unbundled renewable-energy credits that don’t put any new renewable energy on the grid. Micron doesn’t plan to invest in new clean-energy sources and will only use renewable-energy credits. This is true for Intel in Ohio as well. While grids can handle load growth, in many cases this will increase demand for dirty energy. In Arizona, for example, local utilities are using new demand to justify increases in fracked gas and to slow the retirement of old coal plants.

In addition, environmental protection goes beyond climate issues. Industry wants environmental exemptions on even regulated chemicals, though many chemicals they use are presently unregulated and untested. Local agreements tend to lack transparency about chemical use, making the understanding of the full impact to human health unclear.

More recently, the issue of PFAS and related toxics has garnered more attention, including from several Democratic senators, prompting meetings among CHIPS agencies. The new prototyping and research entity, Natcast, recently released a program called PFAS Reduction and Innovation in Semiconductor Manufacturing (PRISM). Notably, PRISM is not explicitly about removing PFAS from semiconductor manufacturing entirely. Alongside this effort is a longer-term set of funding opportunities for PFAS alternatives, which may not survive the Trump administration.

Developing cleaner chips is a long-held goal of the environmental-labor coalition. However, unions and the Sierra Club worry that even if PFAS and related chemicals are phased out, they will be replaced by other chemicals like PFBS, which still has some level of toxicity but will be allowed for use without public testing and regulation.

Labor: Worker Training vs. Union Involvement and Representation

When the CHIPS Act was announced in 2022, the Biden administration committed to place labor at the center of its new industrial policy. The CHIPS Office funding documents ask that companies follow the “good jobs” principles outlined by the Departments of Labor and Commerce. To what extent did the administration deliver on its promises?

The unionized building trades, especially the pipefitters unions, have won contracts to build and maintain new electronics plants. But only one company has committed to labor neutrality (an agreement not to fight unionization) for other long-term workers: Akash Systems, which only received $18.2 million of the $39 billion in manufacturing incentives for its West Oakland facility. Although Micron announced in 2023 that it would talk labor peace with the Communications Workers of America (CWA), a year later the union said the discussions were “disappointing.”

Rather than focusing on high-quality jobs, the agencies in charge of CHIPS have advanced industry priorities like workforce development and training. For example, design is one of the few areas of the semiconductor industry where the United States still dominates. The average worker in this field is in their mid- to late fifties. Over the past five years, Apple, which has been very concerned about this predicament, has developed and funded its own curriculum and training program. Intel has taken similar steps to address workforce problems.

In an interview, Dan Kim of the CHIPS Office said that creating a broader semiconductor ecosystem in the U.S. will encourage workers to circulate among companies, develop best practices, and improve techniques. In theory, these exchanges can create the engineering expertise that, according to many, produced TSMC’s dominance.

The Chips Communities United (CCU) coalition is the most active left pressure group, comprising various unions, environmental groups, and other independent activists. CCU aims to unite its coalition to steer industrial policy around chips toward a Green New Deal–style framework, where workers and communities get a fair shake. This kind of push would make these investments more palatable to communities and solve some of the labor shortages often cited by companies, by creating more attractive, better-paying jobs that don’t expose workers to toxic chemicals.

But the coalition formed only recently, was not active in drafting the legislation, and doesn’t yet wield the kind of influence that more established left groups used to shape Inflation Reduction Act allocations in clean energy. Legislative staffers involved in CHIPS whom we interviewed note that they have felt little left pressure and think it’s a mistake that more powerful and established left organizations have not mobilized around CHIPS money.

According to Judith Barish of CCU, the coalition has been “trying to get the Department of Labor at the table,” as well as building local coalitions to get signed community benefits agreements (CBAs). But no preliminary CHIPS agreements have a minimum living wage, or detailed minimum standards like community benefits, worker voice in production, worker-led elected health and safety committees, whistleblower protections, or “green and clean” R&D. Several companies that have received preliminary agreements have either union-busted or resisted worker demands. In a recent interview, workers at Analog Devices, Inc., in Oregon, said it is a “real struggle to survive” on their salaries.

In addition, right-wing (and some centrist) attacks on the child care provisions in CHIPS have been effective, despite not being credible (labor costs are minuscule compared to other capital requirements). Labor organizers we interviewed were upset that Democrats did not fight back.

Some state grants, like Ohio’s agreement with Intel, require a minimum average pay floor and a certain number of jobs created. But companies exploit averages to hide that the bulk of jobs pay poorly. As the Dayton Daily News reported, 70 percent of the 3,000 new employees at an Ohio Intel plant are slated to be technicians, who can make as little as $50,000 a year, which is well below the average wage.

According to labor sources, the largest share of workers on the production side are operators and assemblers, who are paid a median wage of $18.13 an hour. Most of the production jobs pay around $40,000 a year. Many of the most advanced plants are automated, meaning there are fewer production workers and more engineers, who make around $70,000, not the six figures the Biden administration intimated. Fabs often run 24/7, and engineers are frequently called in to work weekends and nights. But they don’t make overtime, despite working these extreme hours. This creates lots of turnover.

For its part, industry has complained that it can’t find enough domestic workers, in order to justify more H-1B visas, which can trap people in bad jobs at risk of deportation. Unions claim there are plenty of workers, including many in the marginalized communities that the Biden administration wanted to target. But it’s hard to attract workers for poorly paid jobs with terrible hours and potential exposure to carcinogens.

It is unclear if Biden administration claims about how many jobs CHIPS will create will pan out. It is possible there will be even fewer jobs created than the (meager) 125,000 cited, unless industry is held accountable with strong contract metrics. Meanwhile, the more labor-intensive pieces of semiconductor production, like packaging and assembly, are not coming to the U.S. without more political pressure or subsidies; attempts to reshore these tasks have been much less significant than the attempts to regain advanced chip manufacturing. East Asian governments that house most high-end fabs provide prodigious subsidies, tax breaks, free land, and cheap labor; it will be difficult to dislodge all facets of production.

The longevity and political benefits of this kind of industrial policy rely on whether the jobs are good union jobs, whether workers can have a say in production to make sure companies are not just looting government subsidies, whether profits get reinvested in production, and whether communities get benefits. All of these factors are apparent in the vision of CHIPS, but were not prioritized in the rollout.

CHIPS at the State Level: Ohio as a Case Study

Ohio government, like many state governments, is notoriously corrupt. Yet the CHIPS legislation relies heavily on the state and local level for implementation. Land deals around CHIPS allocations in Ohio offer a window into a political economy where oversight is scant and crony capitalism and corruption are often ignored.

In his 2022 State of the Union address, President Biden highlighted “a thousand empty acres of land” east of Columbus, Ohio, as “the ground on which America’s future will be built. That’s where Intel, the American company that helped build Silicon Valley, is going to build a $20 billion semiconductor ‘mega site.’”

These “thousand empty acres of land” were not originally empty. Much of it was leased out for farming. But for over a year prior to Intel’s announcement of an advanced chip manufacturing plant in western Licking County, this emptiness was quietly created through pressuring local residents to sell. With increases in land prices stemming from speculation around new semiconductor plants and data centers, landowners who inherited this farmland have been selling rather than leasing land to working farmers. Many felt coerced.

In the summer of 2021, several farmers and homeowners in unincorporated areas in western Licking County, such as Jersey and Monroe Townships, received unexpected letters and knocks on their doors from representatives of the New Albany Company, a private real estate business owned by billionaire Les Wexner. These representatives made offers to option their properties at possibly 50 percent above market rate, with payment up front on the condition that owners sign a nondisclosure agreement. MCVGCM Holdings and MBJ Holdings—two business entities of the New Albany Company—would then buy the properties.

Within six months, these entities acquired over 3,190 acres of farmland in Jersey Township and eventually moved to annex the land into the City of New Albany, a city that Wexner built in the ’80s. The Jersey Township trustees approved the annexation, albeit reluctantly. Trustees claim they were kept in the dark about the purpose of these changes.

The electronics industry as a whole remains skeptical that this industrial-policy push will be sustained enough to do much.

In May 2022, the New Albany City Council voted for emergency legislation to annex and rezone 1,689 acres as a “technological manufacturing district.”

The whole process was shrouded in secrecy; residents were unhappy with these arrangements but felt helpless to stop them. “You either work with them or they run over you,” said Jeff Fry, a trustee in Jersey Township. Many people privy to the plans had signed nondisclosure agreements.

Wexner and his company benefited from the plan to bundle together thousands of acres for Intel and other companies. JobsOhio, a private entity, supplied the funds for Intel’s land. Responding to Intel’s new facilities, Amazon, Microsoft, Meta, and Google all pledged investments to expand data centers; for this purpose, Amazon and Microsoft bought 1,700 acres of land from Wexner’s company for $400 million. These land transactions yielded massive profits for Les Wexner’s company; he still owns over 1,000 acres in the area.

Intel’s $28 billion fab is now scheduled to be operational by 2030, five years after the original estimate. Its announcement came with the promise of 7,000 construction jobs and 3,000 manufacturing jobs, and has found enthusiastic responses among politicians across party lines, state agencies, unions, community colleges, universities, and local media. It will benefit from $1.5 billion in federal grants, including nearly $2.5 billion at the state and local level (see chart).

Former Ohio Sen. Sherrod Brown saw Intel’s factory as an opportunity to “bury” the “Rust Belt” tag. “We know how to speed up our supply chains, lower prices, and better compete with China: make more things in America,” Brown wrote in a statement in 2022, “and there’s no better place to do it than Ohio.” A Brookings Institution report from 2023 promoted central Ohio as a place “to facilitate large-scale economic inclusion by ensuring sizable numbers of workers can access training pathways toward semiconductor manufacturing occupations.”

At the forefront of this effort are community and technical colleges in Ohio. They see their task as to “fulfill the expectations of the semiconductor industry,” in the words of Richard Woodfield, chief academic officer for the state’s community colleges. Woodfield wants as many of his member institutions to participate with a common semiconductor curriculum, but he claims that hostility to higher education in the state makes it hard to produce enough trained workers.

In 2022, Intel announced a $100 million investment, part of which established the National Semiconductor Education and Research Program. Half of this was promised to Ohio higher education. Intel’s new investments, administrators hope, will provide the state with equitable prosperity. The state is expediting the rollout of this new educational infrastructure at a historic pace, Woodfield said, because they believe “if we don’t produce the workforce … then all those additional fabs … won’t be built.” He added that less rigorous programs would facilitate the inclusion of women and minorities: “The longer the curriculum, the more likely that they won’t finish … life gets in the way.”

Just as at the national level, in Ohio critical questions of job quality, union rights, and workplace safety and health are missing from discussions of “workforce development.” Melissa Cropper, president of the Ohio Federation of Teachers, lamented that administrators are “not interested in labor rights or health issues or quality jobs; [they want to] prepare the workforce but not quality jobs.”

Intel sought and received a “trade secret” exemption from Ohio’s Environmental Protection Agency, allowing them to avoid disclosing information contained in their air pollution control permit application. Communities residing around the proposed plant will not receive adequate information about pollution, nor will they be briefed on what to do in an emergency if there is an accidental emission or discharge. Communities and advocacy groups have not received responses to their inquiries and concerns.

Targeting marginalized communities for haphazard worker training for jobs that might not be particularly good or safe may prove a flawed workforce strategy. Intel has only offered vague information about who is getting these jobs, despite grant stipulations about inclusion.

Anti–Wall Street Priorities

As Vivek Chibber argues, successful industrial policy requires the state to be able to discipline its capitalist class, preventing them from simply looting subsidies. Likewise, for industrial policy to succeed, constituencies beyond industry have to benefit. But none of the anti–Wall Street priorities have manifested in the CHIPS rollout, which is not completely surprising considering the predominance of Wall Street financiers at the core CHIPS agencies.

Many of the fights about allocating CHIPS money have been about whether it should go to U.S. or foreign companies. Both options have major drawbacks, which could be tempered by nationalizing more crucial pieces of the infrastructure. Existing players like Intel and Micron have less-than-inspiring recent records. As Adam Tooze noted, Intel has already had to indulge in creative financial engineering to supplement government subsidies. Meanwhile, Samsung has been having serious internal issues, and both TSMC and Samsung are in talks to build fabs in places like the UAE, which may undermine the anti-China export controls the Sullivan wing has been so intent on rolling out.

Progressives wanted a national infrastructure not controlled by current chip monopolists, hoping that this would encourage the emergence of more functional alternatives to Intel, which has staggered over the past couple of years. A nationalized infrastructure could also prioritize Green New Deal–style values: not just making chips smaller, faster, and cheaper, but also making them and their production more environmentally sustainable, in terms of both toxic by-products and energy usage. Likewise, while industry tends to prioritize incremental improvements over genuine novel research (it’s less risky), more government oversight over national infrastructure could help course-correct.

But the fight over whether the new testing and prototyping entity, the National Semiconductor Technology Center, will be open and democratic has been lost. Industry has succeeded in instituting a pay-to-play model for incumbents that has failed in the past, for example with SEMATECH.

Meanwhile, stock buybacks have historically hampered investment in the industry. The American Economic Liberties Project estimates that through stock buybacks, growing executive pay, and other similar issues, “$698 billion” has been squandered—“an amount 14 times the total CHIPS Act funding, or enough to build 70 leading-edge fabs.” The Institute for Policy Studies and Americans for Financial Reform Education Fund express similar concerns; they recently reported that the first 11 CHIPS Act awardees have “spent more than $41 billion combined on stock buybacks.” Intel spent $30.2 billion on stock buybacks from 2019 to 2023. Texas Instruments stockholders recently balked at the idea it would invest in production over stock buybacks.

Unfortunately, much like the care economy investments, the CHIPS Office guidance merely suggests it prefers that awardees refrain from buybacks.

Conclusion: Is There a Path Forward?

After the election loss, the main priority for the Biden administration was getting all $52 billion in CHIPS money out the door. But that level of cash is not proportional to the capital-intensive nature of the industry (a single new fab costs between $4 billion and $20 billion). AI companies are now seeking CHIPS funding, which would further reduce available funds.

There have been successes. The TSMC plant in Arizona has produced yields on par with, or even better than, those in Taiwan. The U.S. government has successfully used political pressure and has leveraged Apple’s monopolistic position, including a promise to buy TSMC’s Arizona-produced chips, to bring production of some cutting-edge chips to the U.S.—at least temporarily.

But without continued subsidies and political pressure, production may shift back, as happened with Intel’s Ireland facilities. Donald Trump has criticized funding for TSMC’s U.S. plants. And TSMC’s recent failure to keep its chips from Chinese users demonstrates that anti-China crusaders may not get what they want from CHIPS.

Reagan’s pro–big business industrial policy in semiconductors and other high-tech goods was unable to sustain itself past the late 1990s, and led to much of the consolidation and dysfunction that motivated the current government interventions. Neglecting progressive priorities aimed at increasing the longevity of industrial programs could produce the same failures.

Why do policymakers seeking to revitalize U.S. tech-industrial policy keep trying the same methods long after they’ve failed? Perhaps it is related to the dulling effects of access and overclassification. In his book The Theatre of Operations, Joseph Masco recounts what Daniel Ellsberg told Henry Kissinger upon Kissinger’s entrance into the classified world of government, that “it will become hard for you to learn from anybody who doesn’t have these [security] clearances … The danger is, you’ll become something like a moron. You’ll become incapable of learning from most people in the world, no matter how much experience they may have in their particular areas that may be much greater than yours.” In short, policymakers have become increasingly unable to take in new information, ideas, and expertise.

A renewed willingness to do industrial policy and spend fiscally is encouraging and, contra Jonathan Chait, not a short-term fad. But in itself, industrial policy is not anti-neoliberal; Reagan’s industrial policy for semiconductors was in many ways the model for CHIPS. Many of the climate initiatives in the Inflation Reduction Act more successfully included labor, maybe because these fields are more suited to the theories and practices of the new industrial policy and garner more attention and lobbying from the left. In CHIPS, China-hawkishness among policymakers has overcome any serious efforts at moving on from the failed industry-centered industrial policy of the past.

The electronics industry as a whole remains skeptical that this industrial-policy push will be sustained enough to do much, and will merely continue a cycle of sporadic government funding. Industry has quietly claimed since its inception that CHIPS lacked enough money. A second CHIPS bill has quietly been proposed, but the Republican takeover of government and a focus on cutting spending makes that prospect remote. The semiconductor business is cyclical and often volatile; scale and the cost of capital can determine success. It is not an industry that can survive without government coordination, funding, and trade and foreign policy assistance.

Tax breaks may be the future of CHIPS funding at the federal, regional, and state levels. The CHIPS Act included a 25 percent tax break on manufacturing equipment for semiconductor production. As The Wall Street Journal reported, former Intel chief executive Pat Gelsinger has cited the tax incentives as “the most important mechanism to keep the momentum of the program going in the long run.” Official and unofficial state and regional subsidies could also help sustain these investments.

The biggest problem for the industry in the short term, which may spell doom for the entire CHIPS policy venture, is what will soak up all the chips they are producing in these new plants at home and abroad. The question of what to do with the excess capacity is vitally important. Otherwise, semiconductor prices will collapse, and the U.S. will be in the uncomfortable position of having excess chips to dump. The U.S. government could contemplate something along the lines of a strategic chip reserve for use in grants, government uses, and even foreign aid to make sure prices don’t melt down once capacity goes up.

Newt Gingrich’s dream that high tech might obviate the need for the state is dead; the opposite is true. Semiconductors are one of the fundamental motors of tech, built now into every facet of society. Their production, it is now understood, requires a robust state infrastructure and massive state investment and planning. The fact that the genuine technological marvel that crucially undergirded claims about the efficacy of the post-Reagan political-economic model—the Moore’s Law–compliant chip—explicitly violates the tenets of the ideology it is enlisted to defend is a wondrous contradiction. Nonetheless, due to a fundamental lack of imagination and unwillingness to re-examine escalating anti-China sentiment pervasive in both parties, old models prevail in semiconductor policy, over attempts to escape the neoliberal.