NVIDIA Will Use Intel Foveros Tech by 2028
The potential partnership could alleviate significant production pressure on TSMC's advanced manufacturing capacity.
Updated on Oct. 9, 2026 in Semiconductors
Live Poll
Will closer manufacturing partnerships between major tech giants result in better electronics for you?
Analyst reports suggest NVIDIA may utilize Intel's Foveros 3D packaging technology starting in the second half of 2028. This move coincides with the planned production rollout of Intel's 14A manufacturing node.
Why it matters
By diversifying its packaging supply chain, NVIDIA could reduce its heavy reliance on TSMC, which currently faces high demand for advanced chip production. Utilizing Intel's infrastructure may offer a strategic buffer against capacity bottlenecks for future high-performance computing needs.
Intel Foveros enables 3D stacking of computing tiles on a base die, effectively shrinking the horizontal package footprint compared to standard 2D or side-by-side configurations. This approach stands in contrast to Intel's existing EMIB-T technology, which uses horizontal chipset placement for interconnection.
The players
NVIDIA
A dominant designer of graphics processing units and AI accelerators that relies heavily on advanced foundry services.
Intel
A semiconductor developer and foundry operator currently scaling its 14A manufacturing process and Foveros packaging stack.
TSMC
The world's largest dedicated independent semiconductor foundry that currently manages the majority of high-end AI chip production.
Jeff Pu
A securities analyst tracking semiconductor supply chain shifts and manufacturing roadmaps.
The details
Foveros technology uses a vertical stacking architecture, layering specialized computing tiles directly atop a base die to minimize signal latency and board space. Backend chip production—the final stage where processed wafers are cut into individual chips and enclosed in a protective housing—requires specific packaging interconnections to bridge these layers. By adopting this stackable architecture, NVIDIA could theoretically achieve higher logic density within a standard server form factor.
Timeline
January 2026: Analyst reports linked NVIDIA to future use of Intel 14A technology.
H2 2028: Intel 14A production commences and anticipated window for NVIDIA's use of Foveros.
The Tech Race
This move represents a departure from NVIDIA's long-standing reliance on TSMC-exclusive workflows for its flagship products. It forces Intel to prove its foundry capacity against current industry leaders by integrating third-party designs into its proprietary 14A production pipeline.
For data centers and enterprise users, this shift could lead to more stable supply chains for high-performance AI hardware by 2028. End-users will not interact with the packaging directly, but the technical transition aims to prevent the component shortages that have recently constrained hardware availability.
The takeaway
NVIDIA's potential pivot toward Intel's packaging architecture suggests a broader industry move to de-risk production by spreading high-end manufacturing across multiple foundries. Watch for further technical disclosures on 14A node performance benchmarks to see if this integration delivers the promised density gains.
What happens next
Monitor updates regarding the Intel 14A manufacturing node as it approaches its planned production window in the second half of 2028.
Further reading
For more on the evolving fabrication market, visit the Semiconductors section.
Live Poll
Will closer manufacturing partnerships between major tech giants result in better electronics for you?
