Semiconductor Tariffs and the Electronics Supply Chain: What Buyers Should Do

Trade policy has become a component-sourcing variable, which is new. For most of the last thirty years a buyer could treat the geography of the supply chain as settled background. It is now a live input into price, lead time and, in some cases, whether a part can be bought at all.

What follows is the state of play as of June 2026, and — more usefully, because the state of play changes — the sourcing decisions that survive it landing either way.

Where the tariffs stand

On 14 January 2026 the US imposed a 25% Section 232 tariff on certain imported semiconductors and derivative products, effective 15 January under HTS headings 9903.79.01 and .02.

The scope is narrower than the headline suggests. The measure targets high-performance parts meeting defined technical thresholds — tensor processing performance and DRAM bandwidth — which in practice means advanced AI accelerators and their derivative assemblies, not the general run of MCUs, analogue, logic and passives on an ordinary industrial BOM.

The same proclamation directed Commerce and USTR to negotiate with trading partners. A joint report was due on 14 April 2026, and a further Commerce report on the data-centre semiconductor market is due by 1 July 2026. A Phase 2 was signalled: broader tariff coverage at a significant rate, paired with an offset programme for companies investing in US production. As of this writing the Phase 2 determination has not been publicly resolved, and that July report is the next scheduled point at which it could be. A US–Taiwan trade deal focused on reshoring fabrication was announced on 15 January 2026, alongside the tariff.

Two implications for a buyer. First, check your actual exposure before reacting — most industrial and embedded BOMs contain little that falls inside the current scope, and a general response to a narrow measure costs more than the measure. Second, the direction of travel is toward broader coverage, so “currently out of scope” is a statement about today.

Export controls move the market more than duties do

A tariff makes a part more expensive. An export control makes it unavailable to someone — and unavailability, not price, is what reorganises a supply chain.

The same January action added a BIS rule creating case-by-case licence review for high-end accelerators destined for the PRC and Macau. The mechanism to watch is second-order: when a controlled part cannot be sold into a large market, the demand does not disappear. It moves — to the nearest permitted alternative, to domestic substitutes, and to whatever inventory already sits outside the control’s reach. That displaced demand lands on parts nobody expected to be tight, which is why control announcements are frequently followed by shortages in adjacent, uncontrolled families.

This is the pattern worth tracking, and it is the one most sourcing teams miss because they screen their BOM against the controlled list and stop there.

What geopolitical risk actually looks like on a BOM

Not “we buy from Asia.” That is too coarse to act on. The exposures that matter are specific:

• Single-fab parts. A part fabricated in exactly one facility, in one jurisdiction, is exposed to any event affecting that jurisdiction — tariff, control, earthquake, or a change of government. Multiple assembly sites do not help if the wafers come from one place.
• Concentration you did not choose. Several nominally different suppliers frequently share a foundry, a substrate supplier or an assembly subcontractor. The diversification on your approved-vendor list can be notional.
• Parts in end products with export restrictions of their own, where the compliance obligation runs in both directions.
• Long-lived products. A ten-year industrial product will outlive several policy regimes. Its supply plan should assume that rather than the current one.

Decisions that hold up regardless

The honest position is that nobody sourcing components can forecast trade policy, and plans built on a specific prediction fail when the prediction does. What works is reducing sensitivity to the outcome.

Know your fab and assembly geography on the concentrated lines. Not all 400 lines — the 20 that carry the spend and the 20 that are single-sourced. That list is a weekend of work and it is the prerequisite for every other decision here.

Qualify second sources across jurisdictions, not just across suppliers. Two suppliers whose parts come from the same fab are one supplier for this purpose.

Hold more inventory on the specifically exposed lines, not generally. Broad inventory build-up in response to a narrow risk is how working capital gets consumed without reducing exposure.

Time purchases against announced effective dates. Tariff measures come with dates. Parts on the water before an effective date are frequently treated differently from parts that ship after it, and that is a real, legal saving available to anyone paying attention to the Federal Register.

Treat the open market as part of the plan. Inventory already inside your jurisdiction — excess from cancelled builds, distributor stock, surplus held by contract manufacturers — has already cleared customs under the rules that applied when it arrived. During a policy transition that is often the fastest and cheapest supply available, and it is the supply that a plan built only around factory orders never sees.

Verify country of origin properly. Duty exposure follows origin rules, which are not the same as the shipping address. On tariffed lines this is a documentation question with a price attached, and getting it wrong is expensive in both directions.

What not to do

Do not panic-buy across the whole BOM in response to a narrow measure — the carrying cost is certain and the exposure reduction is not. Do not assume a tariff is permanent; several have been moderated by negotiated agreements within months. And do not respond to a price increase on an allocated part by taking the cheapest offer available, because policy disruption and counterfeit supply arrive together, for the same reason: a part that is suddenly hard to get is a part worth faking.

How Makat handles it

We work the supply that already exists rather than the supply that has to be scheduled — independent distributors, excess inventory, and surplus from cancelled builds — which is exactly the inventory that matters most when the factory route is disrupted or repriced. Offers typically come back in 3 to 48 hours, with no MOQ.

Makat buys and resells as principal, so you get one PO and one invoice with us as Vendor of Record and $10M in product liability insurance behind the sale, and parts without full manufacturer traceability are tested to AS6081 and AS6171 by accredited labs before they ship. We are ISO 9001:2015 certified and an ERAI member. Send us your exposed lines.