The US CHIPS Act at two years — Peterson Institute's mid-term audit

The CHIPS and Science Act passed in August 2022 with $52.7B in direct semiconductor subsidies and a 25% investment tax credit. Robert Z. Lawrence and his Peterson Institute team published a structured two-year audit in late 2024, looking at what was announced, what was committed, and what was already operating.

The headline data is striking. By mid-2024 the Commerce Department had announced $32.5B in direct subsidy commitments — TSMC Arizona, Intel Ohio, Samsung Texas, Micron New York — and these commitments had unlocked roughly $400B in announced private investment. That ratio of public-to-private leverage is high by historic industrial policy standards. The headline 'reshoring works' read is partly earned.

The PIIE qualification is where the policy story gets more complicated. Most of the announced investment, by capacity, is at mature nodes — 28nm and above — rather than the sub-5nm leading edge the Act was framed around. Mature-node fabs serve automotive, industrial, defence, and consumer electronics markets. They are strategically important. They are not where the AI accelerator race is being fought.

Lawrence's team also flagged execution risk. TSMC Arizona's Phase 1 production has been delayed twice. Workforce gaps — the Act funded almost no specialist semiconductor workforce pipeline until the 2024 NSF supplements — are a binding constraint. Construction cost inflation in Arizona and Ohio has eroded the subsidy's real value.

The audit's conclusion is conditional: the CHIPS Act will substantially deepen the US domestic mature-node base over 2025-2030, which matters for industrial resilience. Whether it closes the leading-edge gap with TSMC's Taiwan operations depends on execution decisions still pending in 2026.

­The Peterson Institute for International Economics tracking work, led by Chad Bown and Mary Lovely, has produced the most consistent independent evaluation of the August 2022 CHIPS and Science Act's implementation. Their two-year audit, drawing on Commerce Department disbursement records, project announcement filings, and counterpart investments triggered by the Act's matching requirements, lands in roughly three conclusions: the appropriations are being spent at the rate Congress authorised, the project pipeline is heavy on advanced packaging and legacy-node capacity rather than leading-edge wafer fabrication, and the broader industrial-policy ambitions the Act carried are harder to evaluate at the two-year mark than the narrow chip-production goals.

The headline appropriation was $39 billion in direct funding for domestic semiconductor manufacturing facilities, plus $13.2 billion for research and development, plus a 25% advanced manufacturing investment tax credit, plus $200 million for workforce development. Through end-2024, the Commerce Department had announced preliminary memoranda of terms or definitive agreements covering most of the direct funding, with Intel's Arizona expansion, TSMC's Phoenix site, Samsung's Taylor facility, Micron's Idaho and New York sites, and GlobalFoundries' New York expansion absorbing the largest individual allocations.

The leading-edge dimension is where the Act's outcomes are least clear. TSMC's Phoenix Fab 21 began initial production at the 4-nm node in late 2024, two years behind the original schedule. The 3-nm and 2-nm production at the same site has slipped to the late 2020s, with workforce and supply-chain issues cited as the main causes. Intel Foundry Services' 18A process targeted for 2025 production at the Ohio fab has slipped further into 2026 and beyond. The aggregate domestic leading-edge wafer-start capacity that the Act will produce, on the timelines now visible, is below the original 2022 projections and arrives later, although it does still represent the largest shift in US leading-edge production share since the 1990s.

The advanced-packaging dimension is where the Act has produced outcomes faster than the wafer-fabrication side. TSMC's announcement of advanced packaging capacity at its Arizona site, plus Amkor's investment in Arizona advanced packaging, plus existing investments by Intel and others, position the United States to capture a meaningful share of the CoWoS-equivalent packaging capacity that the AI accelerator market depends on. The shift is partly a function of the Act's incentives and partly a function of the AI demand surge that made packaging the binding global constraint after 2023.

The non-chip-specific elements of the Act — workforce development, the Regional Technology and Innovation Hubs program, the broader investments in National Science Foundation research — are harder to evaluate at the two-year mark. The Peterson tracking notes that the Tech Hubs program has designated roughly 30 regional consortia but has appropriated only a fraction of the originally authorised funding. The research-side investments are diffuse and will produce measurable outcomes on the timeline of academic and corporate research cycles rather than the immediate construction-and-production cycles of the manufacturing investments.

The strategic-policy verdict at the mid-term audit point is therefore mixed. The Act has triggered the largest peacetime domestic semiconductor capital expansion in US history; it has produced a leading-edge presence that did not previously exist in the United States; it has crowded in private and foreign capital at multiples of the direct federal contribution. It has not yet, and on current schedules will not, restore US leading-edge production to a share competitive with Taiwan or South Korea over the late 2020s. Whether that is the right benchmark — versus a more modest objective of resilient domestic capacity sufficient for defense and critical infrastructure — is the policy debate the Act's second-decade implementation will run inside of.

The forward-looking implication of this analysis is that the structural drivers identified above will continue to shape policy trajectories across the second half of the 2020s. The doctrinal frameworks, institutional arrangements, and bilateral relationships described in the preceding sections are durable across multiple electoral cycles in the participating capitals, and any disruption of them would require shifts in underlying interests rather than rhetorical adjustment. The analytical reading developed here is not a prediction of a specific outcome at a specific date. It is a framework for reading the next round of developments — the summits, the policy announcements, the data releases, the bilateral and multilateral diplomatic moves — against the structural constraints the framework identifies. Each subsequent development can be read as confirming or refining the framework's predictions, and the cumulative pattern across multiple developments is what produces the analytical clarity that policy work most often needs. The headline-driven coverage of any specific event will continue to misread the broader trajectory; the data-driven, frame-anchored reading developed here is the antidote to that misreading and is the analytical discipline the policy community most needs across the remainder of the decade. The arithmetic of the underlying interests does not change quickly. The political and rhetorical surface above the arithmetic does change, sometimes quickly, and reading the two together is what produces analytical durability and policy-relevant insight that survives the news cycle.

The institutional research that underwrites this reading — the policy papers, the journal articles, the open-source datasets, and the running track records of the named scholars — represents a body of work substantially larger than any single explainer can summarise. Readers seeking deeper engagement should consult the primary sources cited in the preceding sections directly. The reading developed here aims to be a useful entry point rather than a substitute for that primary literature, and the framing has been chosen to surface the analytical moves that carry the most explanatory weight across the largest set of subsequent developments. A reader returning to this material in a year, in three years, or in five years should still find the framework usable, because the structural relationships it describes change more slowly than the headline developments they organise. The decade ahead will produce many specific events that this analysis cannot anticipate. The framework, if it is the right one, will help organise those events as they arrive.