Bottom line first: PCB sourcing is shifting from "lowest price at a single point" to "switchable across regions". Orders have not disappeared — they have been redistributed, with part of the volume moving to Southeast Asian assembly and mid-to-low layer capacity, while high-end, high-layer boards that need tight registration are concentrating in the few factories with a complete process chain. The real action for a buyer is not waiting for prices to move, but building a qualified supplier pool of two regions and two process grades, and moving the engineering review forward into the RFQ.
Structural change: the orders moved, they did not vanish
Three changes stand out over the past two years. First, the US market has added tariffs on PCB from China, and the reciprocal tariff disruption since 2025 has changed the cost structure of exporting finished equipment and sub-assemblies quite visibly. For an intermediate product like PCB, however, most of the duty is borne by the importer and assessed on origin and degree of processing, so the landed cost of the same board can vary far more between customs routes than between factory quotes.
Second, PCB and assembly capacity in Southeast Asia — Vietnam, Thailand, Malaysia — keeps expanding, but the expansion is concentrated in 2~8 layer mid-to-low-layer boards and in assembly. That capacity is price-sensitive and cheap to switch. Conversely, capacity for high-layer boards (20 layers and above), HDI, thick copper and metal-base boards stays highly concentrated, because those boards depend on lamination cycles, registration accuracy, via filling and impedance consistency built up as a whole process chain, which cannot be copied quickly.
Third, AI compute demand is pulling high-end capacity further away. In public industry discussion, AI server carrier boards and switch backplanes have moved into the 24~32 layer range, and such orders consume the tightest part of the capacity base — high-layer lamination lines, and lines that can hold registration and back-drilling at high layer counts.
Prices step up by tier, lead times stretch at the tail
These two are often discussed together, but the mechanisms differ completely. Price behaves as a tiered step-up: low-layer boards on standard materials still face fierce competition because capacity is ample, while high-layer, low-loss, thick-copper and metal-base boards have much less price elasticity, since capacity is constrained and the materials themselves — low Dk/Df hydrocarbon or PTFE systems, high-Tg FR-4 — are in tight supply.
Lead time behaves as a stretched tail: the fastest channel for standard boards is still fast, but complex boards take considerably longer than standard parts and are affected by both scheduling and material arrival. The caveat is that lead-time claims must be stated honestly — as fast as 24-48 hours, actual timing depending on product complexity. Treating "fastest" as "default" is the single most common way a procurement plan loses accuracy.
Four executable actions for purchasing and engineering
- Split the quote into three parts: processing price, the tax difference created by origin and customs route, and logistics plus capital tie-up. An FOB quote alone distorts the comparison. Asking suppliers for both an origin declaration and the available customs routes is far more effective than arguing about duty later.
- Allocate by process grade, not by spend: low-layer boards and assembly suit multi-region price comparison and fast switching; high-layer, HDI, thick copper and metal-base boards should lock stable capacity first, with price subordinate to consistency. Build a tier table keyed on layer count, HDI or not, copper weight and metal base or not, and name a primary and a backup supplier for each tier.
- Move the engineering review into the RFQ: supply stackup requirements, impedance targets, copper weight and hole structure with the enquiry so the factory can give process feasibility feedback while quoting. A common practice is to require a one-page DFM summary with the quote, settling "price" and "can it be built" in one exchange.
- Hold safety stock and a qualified alternative for long-lead materials: low-loss and special laminates typically arrive more slowly than standard FR-4. Assess 4~8 weeks of safety stock for critical part numbers, prepare an electrically acceptable alternative grade, and complete impedance and reliability validation in advance instead of changing material only when supply breaks.
Where this goes
Over the next 12 months the core tension will not disappear; it will move from price negotiation to capacity availability and switchability. Change the supplier strategy from single-point optimum to a matrix of two regions (domestic plus Southeast Asia) and two process grades (standard plus high-end): keep standard tiers competitive on price, and put consistency first in high-end tiers while retaining backup capacity. Whoever brings engineering review forward and standardises part numbers and stackups sooner will take the smaller hit in the next round of price and lead-time volatility.
To make that operable, the RFQ package itself has to carry the parameters that decide the outcome: layer count and stackup, whether HDI is required, copper weight, metal base or not, impedance tolerance, the qualification standard and the expected annual volume per part number, alongside origin declaration and customs route. Writing those into the quotation request and the acceptance criteria is what turns a regional strategy from a slide into a purchasing rule. Suppliers that can quote process feasibility, DFM notes and a realistic lead-time window against that same package are the ones worth keeping in both tiers.
Send your Gerber files and stackup requirements; a 24-hour expedite channel is available.