TSMC and Amkor Sign 10-Year Packaging Agreement; ASE Accelerates Expansion to Counter Competition
Miles Bennett
TSMC and Amkor locked in a 10-year advanced packaging partnership in Arizona, directly challenging OSAT leader ASE; ASE is countering with ~15 new facilities and a record capex above $8.5 billion, signaling a full-scale battle over who can mass-produce packaging at global scale.
Why is TSMC pulling Amkor into an alliance?
TSMC and Amkor's deal is anchored in Arizona, spanning 10 years, focused on advanced packaging.
This means → TSMC is pairing its U.S. wafer fabs with co-located packaging — the entire advanced manufacturing chain is migrating onto American soil.
This reflects a shift: customers now treat supply-chain diversification not as a preference but as a prerequisite for placing orders.
Why does ASE call itself "optimistic"?
ASE COO Tien Wu framed the contest in one line: "The real competition is not about the first chip or the tenth — it's about the hundred-millionth and the billionth."
In plain terms = anyone can produce samples; the moat is stable packaging at the billion-unit scale.
ASE's playbook is "Taiwan first, replicate when mature" — automate and integrate packaging-and-testing in Taiwan, then copy the model to Malaysia, the Philippines, South Korea, Japan, and the U.S.
In the U.S., ASE already runs two test-and-R&D centers in California and is planning a third and fourth site; subsidiary SPIL's potential collaboration with TSMC in Arizona is still under evaluation.
How intense is the AI-driven demand?
Tien Wu said AI demand has exceeded expectations; data-center investment shows no sign of slowing and is extending into autonomous driving and humanoid robotics.
This means → the packaging-capacity gap is not cyclical — it is structural, with demand outlets multiplying faster than capacity can follow.
ASE currently has 13 greenfield projects under way (ASE 6, SPIL 7), plus roughly 2 brownfield acquisitions (existing facilities being converted), totaling about 15 new plants advancing through 2026.
Is $8.5 billion in capex still not enough?
Tien Wu signaled that ASE may raise 2026 capex further, exceeding the already-record $8.5 billion disclosed earlier.
He flagged "more surprises" in H2, including fan-out panel-level packaging — FOPLP, a technique that uses large panels instead of traditional wafers to cut unit costs — reaching mass production before year-end.
This reflects an investment horizon stretching well past 2026–2027 orders, targeting 2029, 2030, and beyond; Taiwan's broader supply chain is lengthening its capex cycle in parallel.
What decides who wins this packaging war?
The key test for the TSMC-Amkor alliance: can Arizona deliver true high-volume packaging capacity, not just a demonstration line?
The key test for ASE's global-replication model: can overseas plants match Taiwan-headquarter cost and speed?
In plain terms = the race is not about who ships first — it is about who achieves scalable, repeatable cost and yield first. That is the real deciding factor in the packaging landscape reshaping.
Content is for reference only, not financial advice.