Nvidia Shows Prototype of Future Co-Packaged Optics
The OOI solution is a research-stage design slated for integration into the Feynman architecture expected in 2028.
Updated on Oct. 31, 2026 in Semiconductors
Live Poll
Do you believe rapid advancements in proprietary hardware will meaningfully improve your daily digital experience?
Nvidia has showcased a prototype of its new OOI co-packaged optics solution at the SEMICON West 2026 event in San Francisco. The technology remains in the research and prototype stage, with mass production not expected until after 2030.
Why it matters
Co-packaged optics represent a shift in data center architecture intended to address throughput and power limitations in high-performance computing. By integrating optical interconnects directly with silicon, Nvidia aims to prepare its future hardware platforms for the scaling demands of next-generation workloads.
Nvidia demonstrated its OOI prototype at SEMICON West 2026, positioning the technology for inclusion in its future product roadmap. This development precedes the slated 2028 release of the company's Feynman architecture.
The players
Nvidia
A designer of graphics processing units, AI accelerators, and high-performance networking hardware for data centers.
The details
Co-packaged optics move optical components, such as lasers and photodetectors, closer to the processor chip to reduce the distance data must travel over copper traces. This approach replaces traditional electrical interconnects with light-based signals to improve bandwidth and energy efficiency. Nvidia plans to incorporate this method into its upcoming Feynman architecture, which is currently scheduled for release in 2028.
Timeline
Nvidia unveiled the OOI prototype at SEMICON West 2026 in San Francisco.
The Feynman architecture is currently scheduled for a 2028 release.
Mass production of OOI hardware is not anticipated until after 2030.
The Tech Race
Nvidia is aligning its hardware roadmap with the industry-wide push to transition away from traditional electrical signaling in high-density data centers. The OOI prototype serves as a research milestone leading toward the company's anticipated 2028 Feynman architecture release.
This technology will not affect current data center deployments as it remains in the prototyping phase. Interested parties should watch for the 2028 debut of the Feynman architecture as a bellwether for the eventual commercialization of these optical solutions.
The takeaway
Nvidia is signaling a long-term shift toward optical interconnects to overcome the physical constraints of copper in AI data centers. Watch the 2028 rollout of the Feynman architecture to see how these optical solutions are integrated into production-grade server systems.
Further reading
For more on the industry's shift toward new data transport methods, visit the Semiconductors section.
Source note: This article includes information reported by Metal.
Live Poll
Do you believe rapid advancements in proprietary hardware will meaningfully improve your daily digital experience?
