NVIDIA and six supply chain partners will expand manufacturing capacity across substrates, advanced packaging, testing and optical engines, the company announced at GTC Berlin on October 21. The agreements are meant to lift the physical supply behind NVIDIA’s AI factory platform, with substrate capacity set to rise 2.5 times over the next three years and a $1.1 billion advance funding new chip-testing facilities in the United States.
The partners are Amkor Technology, IBIDEN, King Yuan Electronics (KYEC), Samsung Electro-Mechanics, Siliconware Precision Industries (SPIL) and Unimicron Technology Corporation. “We are in the midst of the largest infrastructure buildout in human history. Demand is extraordinary. Capacity is the constraint,” NVIDIA chief executive Jensen Huang said in the announcement.
Substrate Capacity Rises 2.5x Over Three Years
Substrates — the layered boards that connect a chip package to the rest of a system — are one of the quiet bottlenecks in AI hardware production. NVIDIA said agreements with IBIDEN, Samsung Electro-Mechanics and Unimicron will, in aggregate, drive a 2.5x increase in substrate capacity over the next three years, with more agreements still underway.
Under an agreement signed earlier this year, NVIDIA is providing IBIDEN a multibillion-dollar capacity reservation advance to support its Ono Plant in Japan, targeting three times the plant’s current capacity, with the additional output expected to begin ramping in the first quarter of 2028. Samsung Electro-Mechanics is joining NVIDIA’s substrate supply chain, bringing immediate capacity from its existing operations in South Korea and supporting facility upgrades and new factories in Sejong and Busan; by 2028, production allocated to NVIDIA is projected to account for a significant portion of its total annual package substrate capacity. NVIDIA is also supporting Unimicron’s KF substrate expansion in Taiwan under a capacity reservation agreement, with capacity expected to start arriving in the second quarter of 2027 and reach full capacity by mid-2028.
Packaging, Testing And Optical Engines Scale In Parallel
NVIDIA and Amkor plan to expand advanced packaging capacity in South Korea, targeting an increase of more than 300% in interposer wafer capacity by the end of 2028, alongside new Amkor manufacturing facilities in the country to support higher volumes for current and next-generation platforms.
On testing, NVIDIA expects its total semiconductor testing capacity to double from calendar 2026 to 2027, supported by buildouts in the US, Taiwan and Singapore. The $1.1 billion capacity reservation advance to KYEC will support new US testing facilities and equipment, with high-volume production targeted for April 2027, while expanded testing operations in Singapore are expected to begin high-volume production in February 2027.
For connectivity, NVIDIA and SPIL plan to expand manufacturing capacity in Taiwan for the optical engines used in co-packaged optics, targeting a 15x increase by mid-2027 and aiming for another doubling by the end of 2028. The expansion sits alongside the containerised factory designs NVIDIA detailed in its DSX POD announcement for Vera Rubin AI factories at the same conference.
Why It Matters
NVIDIA’s supply chain expansion matters because the constraint on AI computing has moved from chip design to the factories, substrates and test floors that turn designs into shipped systems. By paying capacity reservation advances — $1.1 billion to KYEC alone, and a multibillion-dollar advance to IBIDEN — NVIDIA is effectively pre-buying manufacturing output years ahead, a model closer to how it already works with the data centre builders behind efforts such as its OpenShell reference architecture for enterprise agents. For the six partners, the deals convert NVIDIA’s demand forecasts into funded factory programmes in Japan, South Korea, Taiwan, Singapore and the US. For customers waiting on Vera Rubin systems, the timelines — new testing volume from early 2027, substrate ramps through 2028 — are the clearest public map yet of when supply should loosen.
Background
NVIDIA’s GTC Berlin keynote day produced a cluster of infrastructure announcements, from the Hyperion platform for European robotaxi fleets to modular AI factories. The supply chain agreements extend a pattern set earlier in 2026, when NVIDIA began signing long-term capacity deals with substrate and packaging suppliers as AI data centre demand outgrew the industry’s existing plants; substrates and advanced packaging, rather than silicon wafers, have repeatedly been the stages where production queues form.
Source: NVIDIA official announcement via GlobeNewswire (Manila Times syndication)
0 Comments