The White House is weighing a second round of semiconductor tariffs that would reach laptops, servers and gaming consoles, and may scrap January’s data center exemptions, setting up a collision between reshoring policy and the AI buildout
By the US Trade Desk, Peacock Tariff Consulting | August 31, 2026
WASHINGTON, August 31, 2026. The Trump administration is preparing a second, far broader round of semiconductor tariffs that would extend duties beyond raw chips to the products built around them, including laptops, gaming consoles and the data center servers powering the artificial intelligence boom, according to a POLITICO report published late last week that cited eight people familiar with the deliberations and has dominated technology trade circles through the weekend.
The plan under discussion would mark the most consequential expansion yet of the Section 232 semiconductor program launched in January, and it arrives with a twist that has alarmed the technology industry more than the tariff rates themselves: Commerce Department officials have indicated in private talks that the generous exemptions attached to January’s action, covering data centers, research and development, startups and consumer devices, may not carry over into the new framework.
White House spokesman Kush Desai defended the direction of travel, saying that reshoring chip manufacturing is a top priority for the president. The Commerce Department did not respond to requests for comment on the report. People familiar with the talks cautioned that the framework remains preliminary and fluid, that a phase in period is under discussion, and that the design could change substantially in the coming weeks.
From narrow Phase 1 to economy wide stakes
The starting point is Proclamation 11002, signed January 14, which imposed a 25 percent Section 232 duty on a deliberately narrow set of advanced semiconductors. The accompanying White House fact sheet named Nvidia’s H200 and AMD’s MI325X accelerators, and the document explicitly labeled the action Phase 1, an unambiguous signal that more was coming. The January proclamation carved out imports supporting the US technology supply chain, exempting chips destined for US data centers, R&D, startups, repairs, non data center consumer and industrial applications, and public sector uses.
Those carve outs meant the January tariff, for all its symbolism, touched relatively little commerce. The follow on now being designed would work very differently. By extending duties to finished electronics that contain semiconductors, laptops, servers, consoles and potentially a wider range of consumer hardware, the administration would be taxing product categories in which imports account for the overwhelming share of US consumption, and in which final assembly is concentrated in China, Vietnam, Mexico, Taiwan and Thailand.
At the center of the proposed framework is a mechanism championed by Commerce Secretary Howard Lutnick that would tie tariff relief directly to factory construction in the United States. Under the design described by four of POLITICO’s sources, each company would receive a duty free import allowance, a quota of chips or chip containing products it can bring in without tariff, sized according to how much US manufacturing capacity the company has committed to build. Import more than the allowance, and duties apply to the overage. Officials are also examining tiered tariff rates and import caps that would vary by trading partner and by the scale of each partner’s leading semiconductor manufacturers.
The template is recognizable. Taiwan’s January trade agreement with Washington applies zero tariffs to Taiwanese chips within a quota equal to 2.5 times a company’s current US manufacturing capacity while new plants are under construction, tightening to 1.5 times once those plants are built. The new framework would generalize that architecture across the industry, converting the tariff code into an instrument that meters market access against capital expenditure in the United States.
Industry pushback: the AI buildout as hostage
The technology industry’s objection is not subtle: it argues the administration is taxing the inputs of the very AI race the president says he wants America to win. Data center construction is the largest private capital deployment in the US economy this year, and the servers that fill those buildings are imported, chip dense and precisely the products the new tariff would reach.
Jonathan McHale, digital policy chief at the Computer and Communications Industry Association, whose members include Amazon, Google and Meta, compared the data center buildout to “building the transcontinental railroad” and warned that added cost and unpredictability put that investment at risk, according to the POLITICO report.
Industry representatives have pressed a capacity argument as well. Taiwan produces more than 90 percent of the world’s leading edge chips, and TSMC, despite committing 265 billion dollars to its Arizona site, the largest foreign direct investment in US history, projects only around 30 percent of its most advanced capacity will sit in the United States at full build out. A quota keyed to current domestic capacity, the industry contends, simply cannot cover the volumes hyperscalers are buying during a record AI spending run. One person involved in the talks put the domestic manufacturing build out required to satisfy demand at more than five years, longer than any phase in period the administration has allowed on previous tariff rounds.
Tech lobbyists have met with Lutnick and Bureau of Industry and Security Undersecretary Jeffrey Kessler with growing frequency since the start of summer, according to the report, but three of the eight sources said recent talks have moved against the industry, hence the escalating public alarm. Consumer technology publications amplified the warning through the weekend, with Tom’s Hardware, Shacknews and others cataloguing the potential hit to laptop, console and PC component prices heading into the holiday season.
The consumer price question
For households, the exposure is direct. Unlike the January action, which most consumers never felt, a duty on finished laptops and gaming hardware lands on shelf prices. Retail analysts note that the consumer electronics category has already absorbed multiple rounds of tariff costs this cycle, including the reciprocal tariff regime capped at 15 percent for many partners and elevated China rates, and that margins in PC retail leave little room to absorb another layer. Estimates circulating in the trade press suggest double digit percentage increases on affected categories if duties in the 25 percent range are applied without consumer exemptions.
The gaming segment illustrates the stakes. Consoles are assembled almost entirely in Asia, and the segment’s three major platform holders have already shifted some production among Asian countries in response to earlier tariff rounds. A chip content based duty would follow the semiconductors wherever assembly moves, which is exactly the design intent, and exactly why the industry sees no easy mitigation short of US assembly that does not currently exist.
Data center operators face a subtler but larger bill. Hyperscalers are spending at an annualized rate in the hundreds of billions of dollars on AI infrastructure. Even a modest effective duty on imported servers compounds across that volume, and the threatened removal of the January data center exemption converts what has been a protected category into perhaps the single largest pool of dutiable value in the program. Analysts warn the cost would either slow the buildout, get passed through to cloud and AI pricing, or, in the administration’s preferred outcome, get avoided entirely through accelerated US chipmaking and server assembly commitments.
A pattern of escalation across the chip file
The tariff expansion is one thread in an increasingly dense weave of semiconductor trade actions. The administration’s January proclamation directed Commerce to report to the president by July 1 on the market for data center semiconductors, a review that put the January exemptions formally in play. An April 14 report from the US Trade Representative and Commerce covered tariff negotiations with Taiwan, South Korea and Japan, the allied producers whose quota treatment the new framework would systematize.
On a separate track, the government concluded a Section 301 investigation into China’s drive for dominance in foundational, older technology chips, finding Beijing’s conduct unreasonable and actionable. The administration announced it would impose tariffs under that authority but delayed the action, an approach designed to preserve leverage while truce negotiations with Beijing continue. Washington and Beijing are separately maneuvering ahead of a September 24 leaders summit, with a 7.5 percent overcapacity tariff on Chinese goods reported to be in preparation, and the bilateral truce set to expire November 10.
The Section 232 semiconductor program itself, meanwhile, interacts with the reciprocal tariff architecture in ways that matter for importers: several Section 232 sectoral tariffs, including those on semiconductors, lumber and pharmaceuticals, are capped for many trading partners at the flat 15 percent reciprocal rate negotiated in framework agreements, according to analyses by PwC and other advisers. Which cap applies to which product from which origin has become one of the most technical, and most consequential, classification questions in US trade practice.
Carrots, sticks and the post CHIPS Act playbook
The quota framework represents a philosophical break with the previous decade of US semiconductor policy. The CHIPS and Science Act of 2022 approached reshoring with subsidies, appropriating 52 billion dollars in grants and tax credits to attract fabs. The current administration has soured on that model, renegotiating some awards and arguing that subsidies socialize the cost of reshoring while tariffs privatize it, putting the burden on companies and foreign producers rather than taxpayers. Lutnick has made the argument explicitly in public appearances this year: access to the American market is the leverage, and the tariff schedule is the pricing of that access.
The results of that leverage are visible in the deal ledger. Beyond TSMC’s 265 billion dollar Arizona commitment, Samsung has expanded its Taylor, Texas project, Micron has accelerated New York and Idaho memory fabs, and GlobalFoundries, Texas Instruments and SK Hynix have all announced enlarged US footprints since the tariff program began taking shape. Administration officials cite that pipeline as vindication. Industry economists counter that most of the announced capacity is years from wafers, that advanced packaging, substrates and assembly remain overwhelmingly Asian, and that a tariff which bites in 2026 cannot be paid with a fab that opens in 2029.
The partner by partner tiering under discussion would formalize a hierarchy that already exists in practice. Taiwan negotiated its quota agreement in January. South Korea and Japan, whose framework agreements cap many reciprocal duties at 15 percent, are pressing for equivalent chip treatment, and the April USTR and Commerce report on allied negotiations laid groundwork for exactly that. The European Union, a smaller chip exporter to the US but home to critical lithography supplier ASML, has signaled it expects its trade framework to shield it from any Phase 2 rates. Where partners land in that hierarchy will shape investment flows well beyond semiconductors, since server and device assembly tends to follow the tariff optimized path.
Echoes of List 4, with higher stakes
Veterans of the 2018 to 2019 trade war hear a familiar rhythm. The original Section 301 China actions began with capital goods and intermediate inputs, and only in the List 4 round did Washington threaten laptops, phones and consoles, a step it ultimately softened precisely because of the consumer price optics. The current deliberation is List 4’s logic returning under a national security banner, but with three differences that raise the stakes. The tariffs would apply globally rather than to China alone, closing the Vietnam and Mexico assembly detours that defused the last round. The product scope centers on the AI infrastructure the administration itself has designated a national priority. And the legal vehicle, Section 232, gives the president wider discretion and courts have so far granted it more deference than emergency statutes received.
Wall Street has begun pricing the possibilities. Server original design manufacturers and PC brands with concentrated Asian assembly saw estimates trimmed after the report circulated, while shares of companies with announced US assembly operations outperformed. Analysts at several banks noted that hyperscaler capital expenditure guidance for 2027, already the market’s most scrutinized number, now carries a tariff asterisk: a duty on imported servers without a data center exemption would either compress cloud margins or lift AI compute prices at the exact moment investors are questioning the buildout’s return profile.
What it means for importers and the supply chain
For importers of electronics, the immediate task is scenario planning rather than reaction, since no proclamation has issued. Trade advisers are recommending several concrete steps.
First, map chip content exposure across product lines now. A duty structured around semiconductor content will require documentation that most electronics importers have never assembled: which chips, from which fabs, in which finished goods. Companies that build that data set early will clear customs faster and argue exemptions more credibly than those that wait for the Federal Register.
Second, model the loss of the January exemptions. Data center operators, R&D importers and startups that structured procurement around exempt status should price the scenario in which that status disappears, and consider accelerating deliveries of long lead time hardware into the window before any new proclamation takes effect. The January action moved from signature to effect in one day, a precedent that argues against assuming a comfortable runway.
Third, watch the quota design. If duty free allowances are pegged to committed US manufacturing capacity, the commercial advantage will flow to companies buying from suppliers with large announced US projects, TSMC, Samsung, Micron, Texas Instruments and GlobalFoundries among them, and away from supply chains anchored in facilities with no US footprint. Sourcing decisions made this fall may determine tariff exposure for years.
Fourth, remember the phase in is a negotiation. The administration has repeatedly used comment windows and effective date gaps to extract investment announcements. Companies with credible US expansion plans have found that announcing them buys tariff relief; the new framework would formalize that exchange rate.
The exposure is not confined to giants. System integrators, value added resellers and the thousands of small businesses that build custom PCs, industrial controllers and point of sale hardware import chip dense components in volumes far too small to negotiate quota allowances or fund US fabs. Trade groups representing smaller electronics firms have asked Commerce for a de minimis style small importer accommodation in any Phase 2 design; none of the reporting so far suggests one is under consideration. For that tier of the market, the practical options reduce to paying the duty, passing it through, or consolidating purchases through larger distributors with quota access, a dynamic that would further concentrate the electronics supply chain.
The calendar to watch is reasonably clear even if the decision is not. Commerce’s data center semiconductor report, directed by the January proclamation with a July 1 due date, is complete and sitting with the White House, according to people familiar with the process. A proclamation could follow the president’s review at any time, though officials describing a phase in suggest the administration wants the framework announced before the September 24 China summit and effective in stages thereafter. Importers should also watch for a Federal Register notice establishing the quota mechanics, which would function as the program’s operating manual and its first real test of administrability.
The politics of Phase 2
Politically, the expansion tests whether the administration’s reshoring coalition holds when tariffs reach consumer prices. The January action was cost free for most voters; a laptop and console tariff in an election cycle is not. Administration officials argue the quota design answers that concern by making duties avoidable: companies that invest do not pay. Critics respond that the investment cannot physically arrive before the duties do, and that the gap between those two timelines is measured in consumer price increases and delayed AI infrastructure.
There is also a legal backdrop. With the Supreme Court having invalidated the administration’s IEEPA based global tariffs earlier this year and forced roughly 81 to 100 billion dollars in refunds, according to figures reported by Investing.com and court filings in the ongoing refund litigation, Section 232 has become the load bearing statute of US tariff policy. That raises the stakes of every expansion: a Phase 2 that sweeps in finished consumer electronics will invite fresh challenges over whether national security can plausibly stretch from a fighter jet’s chips to a teenager’s gaming console. The Court of International Trade has so far read presidential authority broadly in related cases, including an August decision upholding the elimination of the de minimis exemption, but the litigation environment remains the least predictable variable in the program.
Allied governments are watching with their own anxieties. Seoul and Tokyo negotiated framework agreements this cycle partly to buy predictability, and a Phase 2 that reopens semiconductor treatment would test how much predictability those frameworks actually purchased. Taipei faces the sharpest dilemma: its quota deal rewards US investment by its champions, but a Washington policy that succeeds in pulling leading edge production across the Pacific erodes the silicon shield logic that has anchored Taiwanese security thinking for a generation. Taiwanese officials have said publicly that the most advanced nodes will remain at home; the quota mathematics are designed to pressure exactly that commitment.
The bottom line
No decision is final, and the framework could soften as lobbying continues. But the direction is unmistakable: the administration regards January’s narrow accelerator tariff as a proof of concept, not an endpoint, and it is prepared to put the price of laptops, servers and consoles on the table to force semiconductor manufacturing onshore. For the technology industry, the coming weeks are a race to shape quota mechanics before they harden into a proclamation. For importers, they are a window, possibly a brief one, to build the chip content data, inventory positions and sourcing options that will separate the prepared from the surprised when Phase 2 lands.
As one industry participant told POLITICO, the domestic build out this policy demands will take more than five years. The tariffs, if January is any guide, could take effect overnight.