Commerce Secretary Howard Lutnick is backing a framework under which companies could receive duty-free access for a specified volume of imported chips, with the allowance tied to commitments to manufacture in the world’s largest economy.

Officials are also considering a phase-in period, while tariff rates, quotas and other details remain unsettled.

One option would impose different rates and quotas on individual countries and provide country-specific guidance for major semiconductor manufacturers.

The framework could still change substantially before any decision is made.

ALSO READ | Big tech’s silicon shift: Designing its own AI chips

The proposal has triggered concern across the American technology circuit, which is already facing tight supplies of advanced semiconductors as companies spend hundreds of billions of dollars building data centres for AI services.

Technology groups are lobbying for exemptions similar to those included in earlier tariff measures, arguing that broader duties could make AI infrastructure pricier and less predictable.

Industry representatives have warned that the timing is particularly difficult. Advanced chip factories cost billions of dollars and can take years to build, leaving American Big Tech dependent on suppliers in Asia while domestic capacity expands.

Taiwan produces more than 90% of the world’s most advanced semiconductors, while Malaysia and South Korea are also important suppliers.

That creates a dilemma for the Trump administration. Tariffs could encourage manufacturers to establish more capacity in America, supporting a strategic objective that has gained importance amid concerns about supply-chain resilience and national security.

But imposing duties before domestic production can meet demand could increase costs for the same technology companies Washington wants to lead the global AI race.

The pressure is particularly acute for data centre operators. The hyperscalers are committing giant sums to new computing campuses and competing for high-end chips, driving the industry into a record investment cycle.

A tariff on semiconductors and equipment containing them could increase the cost of servers and other infrastructure just as companies are trying to expand computing capacity.

The potential duties could also affect US chip designers such as Nvidia and Advanced Micro Devices (AMD), which depend on overseas manufacturing. Apple and other electronics companies could face higher input costs and potentially lose competitiveness against overseas rivals that can purchase the same components without paying US import duties.

ALSO READ | Silicon supremacy: Economic impact of AI chip wars

Technology companies have therefore intensified their lobbying efforts. Industry representatives and lobbyists have met senior Commerce Department officials with increasing frequency, pressing for exemptions for data centres, research and development, start-ups, consumer applications, civil industrial uses and public-sector projects.

Some discussions have reportedly become less favourable to those exemptions, with Commerce officials arguing that tariffs need to be broad enough to persuade chipmakers to increase US production.

The administration’s challenge is to design a system that changes corporate investment decisions without creating an immediate shortage or cost shock.

The proposed approach resembles an agreement reached with Taiwan earlier this year under which Taiwan Semiconductor Manufacturing Co, or TSMC, can import some chips into the US duty-free while committing to expand American production.

TSMC has pledged USD 265 billion for its Arizona operations, described as the largest foreign direct investment in US history.

Even that investment illustrates the scale of the challenge. TSMC expects only about 30% of its most advanced capacity to be located in Arizona once its planned build-out is complete, and much of that capacity remains years away. Industry representatives estimate that a meaningful expansion of US semiconductor manufacturing could take more than five years.

This creates a potential mismatch between tariff policy and available supply. If the volume of chips eligible for duty-free entry is based on companies’ US manufacturing commitments, allowances could fall well short of the quantities required by major technology companies.

The resulting gap could leave businesses paying tariffs on components that are not yet available domestically.

The wider semiconductor supply chain would also experience the impact. Higher import costs could flow through to servers, computers, televisions and other electronics, increasing expenses for manufacturers and potentially consumers.

For companies operating on large technology investment budgets, even relatively small increases in component prices can become significant when multiplied across thousands of servers and large data centre projects.

The administration argues that reshoring is essential for economic and national security. White House spokesperson Kush Desai said semiconductor manufacturing remained a priority and that the administration’s policies had already secured hundreds of billions of dollars in investment in the sector.

Trade officials have also stressed the geopolitical dimension. Heavy US dependence on Taiwanese semiconductor production is viewed as a strategic vulnerability because any conflict involving Taiwan could disrupt supplies of the chips on which the US technology industry relies.

ALSO READ | Wall Street bets USD 500 billion on Nvidia’s AI boom as Big Tech faces debt concerns 

Yet economists and industry groups say tariffs alone cannot create a domestic semiconductor ecosystem.

Building factories requires skilled technicians, reliable electricity and water supplies, qualified suppliers and faster permitting, as well as billions of dollars in capital. Making imported chips more expensive does not immediately provide any of those resources.

The debate therefore goes beyond the proposed tariff rate. It concerns whether Washington can expand domestic semiconductor capacity quickly enough to reduce reliance on Asia without undermining the AI infrastructure boom before alternative supplies are available.

US Trade Representative Jamieson Greer said in May that no new semiconductor tariffs were imminent at that time, although he argued that duties could help protect the sector and encourage production to return to the US. The latest discussions suggest the issue remains active.

For technology companies, the uncertainty itself is becoming a business risk. If tariffs, quotas and exemptions remain unclear while companies commit billions of dollars to data centres, investment decisions become harder to calculate.

For Washington, the challenge is balancing two strategic goals: creating a resilient domestic semiconductor industry and maintaining the supply of chips needed to establish US leadership in artificial intelligence.

The new tariff proposal aims to meet both goals, but the industry says that adding costs before the domestic capacity is ready could make the change harder and slow down the AI growth that the policy is meant to encourage.