Sources checked:
What changed
On 21 October, NVIDIA announced capacity agreements and planned facilities across several stages of its supply chain. It says substrate agreements with IBIDEN, Samsung Electro-Mechanics and Unimicron should raise aggregate capacity 2.5 times over three years. Amkor and NVIDIA target more than a 300% increase in interposer-wafer capacity by the end of 2028. NVIDIA expects total semiconductor-testing capacity to double from 2026 to 2027; a $1.1 billion reservation advance is intended to support KYEC testing facilities in the United States, with high-volume production targeted for April 2027. SPIL targets a 15-fold increase in optical-engine capacity by mid-2027 and aims to double it again by the end of 2028. These are company targets and forward-looking statements. They are not completed capacity, independently audited output, guaranteed allocation, delivery dates or customer pricing.
Business implications · analysis
The announcement makes the constraint map more useful than a single headline about chip fabrication. Substrates connect the package, advanced packaging assembles complex devices, testing qualifies output and optical engines move data between large systems. A delay or low yield at any one stage can limit deliverable systems even when another stage expands. Buyers should therefore separate announced input capacity from usable system supply. For each critical component, record the qualified supplier, plant, ramp date, reservation or allocation, yield evidence and the system configuration it supports. A capacity advance can secure priority and help fund a plant, but it can also concentrate demand around one customer's roadmap. Procurement decisions should be tied to accepted systems and service levels, not to nominal multipliers announced years before full ramp.
What to watch next
Watch the milestones that convert plans into supply: equipment installation, customer qualification, production yield, high-volume start, actual shipments, allocation terms and system-level availability. Compare each stated ramp date with later partner disclosures rather than treating the original announcement as completion. One scenario is that parallel expansion across several stages reduces a bottleneck and gives system builders more predictable delivery windows. Another is that one late stage, weak yields or a changed product mix moves the constraint elsewhere. The most useful buyer metric is not the announced percentage increase; it is the quantity of qualified, configured systems accepted on time, with transparent lead times and failure responsibility.
Primary sources
Company statements are attributed to the publisher. Our analysis is separate; forward-looking points are questions to monitor, not promised outcomes.