Real-world scenario: A mid-size Taiwanese OSAT — packaging and testing memory chips for US fabless customers — needs another Southeast Asian plant. Two finalists: Vietnam (close to shifting customers) and Malaysia (the long-established OSAT cluster in Penang). The COO says: “We can model tax, but what kills us is missing material suppliers and engineers who know how to run testers.”
He is right. For OSAT — thin margins, living on ramp-up speed + yield — site selection is not a tax problem but an ecosystem problem: within a 200 km radius, how many suppliers of leadframes, substrates, chemicals and specialty gases exist; how many engineers have actually run test handlers; and whether the cluster has enough gravity to pull new suppliers in behind you. This article offers a method for mapping the ecosystem to score both locations, with Vietnam’s legal framework updated to 18 October 2026. The Malaysia section is framework-level; Malaysian local counsel should confirm details.
Why OSAT differs from an ordinary electronics plant
Three traits make OSAT unselectable by “cheap land + cheap labor” logic:
First, materials decide yield. Chip packaging depends on substrates, leadframes, bonding wire, die-attach adhesives and clean nitrogen. These inputs have strict storage and lead-time requirements — importing from afar means large inventories, large inventories mean bloated working capital and moisture/expiry risk. A mature OSAT cluster (like Penang, Malaysia — where major OSATs have operated since the 1970s) has same-day material supply networks in place. That advantage cannot be bought with tax incentives.
Second, people are the scarcest asset. OSAT engineers are not generic electronics engineers — they must run packaging and test equipment, read yield data and troubleshoot processes. Training an OSAT engineer to independent productivity takes 12–18 months. Wherever an OSAT engineer pool already exists, ramp-up shortens by a year.
Third, customers audit the plant before awarding orders. Fabless companies do not hand chips to a new shop on price — they send audit teams to assess the quality system (ISO 9001 minimum, IATF 16949 for automotive), contamination control, security (chips are high-value IP) and lot traceability. A plant inside an already-audited cluster inherits accumulated credibility — a standalone plant builds it from zero.
The method: a five-layer ecosystem map
Instead of comparing two countries abstractly, score each specific locality (e.g. Bac Ninh, Vietnam vs. Penang, Malaysia) across five layers:
Layer 1 — Material suppliers (highest weight). List the 15 most critical inputs of your OSAT process. For each, mark: supplier within 200 km / must import / no source at all. Rule of thumb: if over 60% of inputs already have same-day domestic sources, that location has a clear ramp-up advantage.
Layer 2 — Technical talent. Count working OSAT/test engineers in the region; relevant training schools; hiring competition (a mature cluster means more people but also aggressive poaching and higher wages).
Layer 3 — Technical infrastructure. Power (OSAT needs absolute stability for cleanrooms — one dip can scrap a running lot), ultrapure water (UPW), specialty gases, chemical wastewater treatment. Check the IZ’s actual spare capacity, not the brochure figure.
Layer 4 — Customers and partners. Where your current fabless customers already are; other OSATs in the region (both hiring competitors and a “certificate” for the region); semiconductor-specialised logistics (ESD-safe packing, temperature control).
Layer 5 — Legal framework and incentives. This layer comes last, not first — but it still must be computed correctly.
Vietnam legal framework for an OSAT plant
Investment: IRC under Investment Law 143/2025/QH15 (effective 1 March 2026) and Decree 96/2026/ND-CP; enterprise registration under Decree 168/2025/ND-CP. Semiconductor projects fall in the high-tech category — check the specific incentive lists for the applicable CIT exemption/reduction level, while noting the global minimum tax: Resolution 107/2023/QH15 (from fiscal year 2024) and Decree 236/2025/ND-CP (effective 15 October 2025) may collect the top-up to 15% for in-scope groups — headline incentives must be re-modelled into real tax.
Tax: standard 20% under CIT Law 67/2025/QH15 (1 October 2025). An OSAT-specific note: tolling models where the customer supplies the wafers (customer-owned material) carry different tax and customs treatment from buy-sell models — structure the contract and customs declarations correctly from day one under Customs Law 54/2014/QH13.
Land: sub-lease inside an IZ under Land Law 31/2024/QH15 (1 August 2024) via Decree 35/2022/ND-CP. OSAT plants need large cleanroom footprints — negotiate an expansion option in the land lease up front; it costs far less than leasing more later.
Digital industry: Digital Technology Industry Law 71/2025/QH15 (effective 1 January 2026) — semiconductor/OSAT projects should review eligibility under this law alongside general investment incentives.
Labor: work permits for foreign specialists (process engineers, yield experts) under Decree 219/2025/ND-CP — OSAT specialists typically meet the expert criteria; prepare early because these specialists “switch on” the whole plant.
Reading the map: where each side wins
Apply the five layers to the two finalists (framework-level — locality specifics need field surveys):
- Malaysia (Penang): wins Layers 1 and 2 — a 50-year OSAT cluster with dense material supply chains and a deep engineer pool; wins Layer 4 — most major fabless firms already have a footprint, easing customer audits. Loses partly on cost (Layer 5): engineer wages and operating costs run higher.
- Vietnam (Bac Ninh–Bac Giang–Hai Phong): wins Layer 5 (cost, incentives) and growth momentum — a fast-expanding electronics cluster where big names already create gravity pulling suppliers in; still catching up on Layers 1–2: the deep semiconductor material chain and the OSAT engineer pool remain thinner than Penang’s, though the gap narrows yearly.
The decision therefore hinges on your time strategy: need ramp-up within 12 months with stable yield from day one → Malaysia; accept an 18–24-month ramp-up in exchange for lower long-run cost and a first-mover position in a rising cluster → Vietnam. There is no right answer for every company — only the right answer for your risk appetite and timeline.
OSAT companies evaluating sites in Vietnam and needing a review of the investment–tax–customs–labor legal framework for a semiconductor project can speak with our team.
Frequently asked questions
1. What tax/customs points matter in a tolling model (customer-supplied wafers)?
Customer ownership of inputs changes the transaction’s nature versus buy-sell — affecting customs declaration, VAT treatment and revenue recognition. Structure the tolling contract and declarations correctly from the first lot; fixing it after running is very expensive.
2. Do Vietnam’s semiconductor tax incentives get “eaten” by QDMTT?
It depends on whether the group is in global-minimum-tax scope. If in scope of Resolution 107/2023/QH15, the gap to 15% is collected as a top-up under Decree 236/2025/ND-CP. Incentives still hold value above 15% and in non-tax forms (land, infrastructure) — but model the real tax.
3. How hard is hiring OSAT engineers in Vietnam?
The pool of engineers with real OSAT experience is thinner than Malaysia’s. The playbook of earlier entrants: bring foreign specialists to “switch on” the plant (work permits under Decree 219/2025) while signing training partnerships with local technical schools from year one — do not wait until the plant is finished to hire.
4. What do fabless customers audit before awarding chips?
The quality management system (ISO 9001 minimum), cleanroom contamination control, physical and information security (chips are high-value IP), lot traceability, and the business continuity plan. Prepare six months ahead — fail an audit once and it takes six more months to get re-audited.
5. How important is proximity to customers for OSAT?
Very. The engineering loop between the OSAT shop and the customer’s design team — fast or slow — decides time to market. Geography plus time zones plus language barriers compound into a large hidden cost.
6. Long land lease now, or short lease then expand?
For OSAT, negotiate an adjacent-land expansion option in the very first lease. Cleanroom expansion needs contiguity — leasing elsewhere later destroys operating efficiency. An option clause today costs far less than relocating or retrofitting after three years.
This article has been quality-reviewed by the FLAW AI Legal Council