ASE's 2026 Capex Raised for Second Time to Record $10.5 Billion
Miles Bennett
ASE Technology (TWSE: 2311) has raised its 2026 capital spending to a record $10.5 billion — the second increase this year — with the extra $2 billion split evenly between facilities and equipment. AI-driven advanced packaging demand has outrun prior plans, turning capacity itself into the industry's binding constraint.
Where is the extra $2 billion going?
Capex rose from the original $8.5 billion to $10.5 billion, up nearly a quarter. The $2 billion addition is split evenly between construction and manufacturing equipment.
Of the roughly $6.5 billion earmarked for equipment, packaging tools take more than 50%, test gear about 40%, and electronics manufacturing services (EMS) less than 10%.
This means → ASE is directing nearly all incremental spending at packaging and testing — the two links most constrained in the AI chip supply chain — rather than traditional assembly.
What underpins the LEAP revenue doubling?
CFO Joseph Tung said ASE's $3.5 billion LEAP revenue target for 2026 is on track to be met ahead of schedule. With new capacity coming online within roughly a year, LEAP revenue is expected to double again in 2027.
Growth is driven mainly by compute and communications applications, with momentum likely extending beyond 2028.
In plain terms = LEAP — ASE's advanced packaging business line — is shifting from pilot stage to primary revenue engine. Management is doubling down because order visibility already supports the bet.
How far along is the advanced packaging roadmap?
COO Tien Wu sees the semiconductor industry at the very beginning of an AI-driven paradigm shift. Rising compute density and larger reticle sizes will push customers further toward CoWoS and panel-level packaging (PLP) — a method that uses large flat panels instead of round wafers, cutting cost and boosting throughput.
ASE's first 310 × 310 mm fully automated fan-out panel-level packaging (FOPLP) line targets mass production in Q1 2027. Co-packaged optics (CPO) — integrating optical communication modules directly into the chip package to shorten signal paths — enters initial production in H2 2026.
This means → ASE is running two tracks in parallel: scaling CoWoS capacity for near-term shipments and betting on next-generation PLP and CPO, rather than choosing one over the other.
Twenty-one projects at once — how tight is the capacity bottleneck?
ASE now has 13 greenfield and 8 brownfield expansion projects running simultaneously — 21 in total, up by six in roughly a month since the shareholder meeting.
The company recently spent about NT$12 billion to acquire two factory sites — AUO's Luzhu plant (Kaohsiung) and Kuai Kuai's Zhongli plant (Taoyuan) — anchoring expansion capacity in both southern and northern Taiwan.
Tien Wu was blunt: demand for AI infrastructure is no longer in question. The biggest bottleneck is now hardware infrastructure itself — capacity, automation, and manufacturing innovation — a constraint the industry has not faced in nearly forty years.
Can margins keep up with this level of spending?
ATM (assembly, test, and materials) gross margin is expected to break 30% in Q4, surpassing the business's previous structural ceiling.
This reflects the higher value-add of advanced packaging lifting ASE's legacy profit structure — not a case of "growing revenue without growing profit."
In plain terms = that 30% line is the make-or-break test for this capex cycle. Clear it, and scale effects plus product mix upgrades are delivering. Miss it, and the market will question whether $10.5 billion was well spent.
Content is for reference only, not financial advice.