Background
AI data centers depend on a long chain of suppliers. Chips must be packaged, connected to circuit boards through package substrates, and assembled into complete systems before they reach customers. A shortage or delay at any step can limit how quickly AI infrastructure gets built.
What happened
On November 17, 2026, NVIDIA announced expanded commitments with its supply chain partners to support the buildout of AI infrastructure.[1] The partners named are Amkor Technology, Foxconn, Kinsus Interconnect Technology, Samsung Electro-Mechanics, Siliconware Precision Industries (SPIL) and TSMC.[1]
Key details
Under a new agreement, Kinsus Interconnect Technology plans to expand ABF package substrate capacity dedicated to NVIDIA by more than 3.5 times over current levels. The expansion will run from Kinsus's Youth campus in Taoyuan, Taiwan. Production is expected to begin in the second half of 2027, and the added capacity is targeted for the third quarter of 2028.[1]
NVIDIA also amended its substrate supply agreement with Samsung Electro-Mechanics to improve operational capabilities at existing factories.[1] The change concerns the factories already in use.
NVIDIA said Foxconn is deepening its collaboration to support next-generation GPU and networking platforms.[1]
NVIDIA also said that Vera Rubin NVL72 and Vera CPUs began volume shipments in the third quarter, using TSMC 5nm and 3nm processes and CoWoS advanced packaging.[1] Advanced packaging is the step that combines several chips and memory into one module, so these processes matter for how much finished hardware can be delivered.
Each partner named in the announcement works at a different stage. Some handle chip manufacturing, some package chips, and some make the substrates and boards that connect them to the rest of a system. Because these steps depend on one another, a delay at one partner can hold up the others, which is why supply commitments are made across the whole chain.
Impact
The announcement covers several companies at different stages of the chain, from chip manufacturing at TSMC to packaging, substrates and system assembly. Capacity commitments with specific dates, such as the 2027 start and the 2028 target for Kinsus, show that supply planning for AI hardware runs years ahead of delivery.
Sources
- 1
Cite this page: NVIDIA expands supply chain commitments for AI infrastructure. TellWorld, https://www.tellworld.net/news/nvidia-supply-chain-commitments-2026. Updated 2026-11-18