By Hassan Mujtaba
Publication Date: 2026-05-03 18:45:00
NVIDIA’s Feynman GPUs will be the first to feature Co-Packaged Optics, but this wasn’t always the case until the AI giant decided to switch gears.
Co-Packaged Optics Were Many Years Away, But NVIDIA Decided To Move Ahead With Its Feynman GPUs
CPO or Co-Packaged Optics (Silicon Photonics) is the next-generation solution that reduces reliance on copper and harnesses light to transfer signals. These CPOs are packaged alongside hardware accelerators such as GPUs and will be a key solution for next-gen AI factories, offering improved interconnect latency and creating high-bandwidth connections between CPU and GPU.
If we go by the original plans, CPO was originally slated for commercialization by 2033, but NVIDIA fast-forwarded the optics roadmap by 5 years, bringing it forward in 2028 with its next-gen Feynman GPUs. The report from Nikkei Xtech states that as AI firms scale out, distances between installed platforms can reach north of 10km, and data will need to be transmitted fast at several hundred Gigabits per second or more. Traditional methods make this difficult, & that’s where optics replace copper cables as the primary means to transfer data.
As such, the Optical Compute Interconnect Multi-Source Agreement (OCI-MSA) announced the establishment back in March, involving major AI firms such as NVIDIA, Broadcom, AMD, Meta, OpenAI, and Microsoft. NVIDIA, being the largest of the bunch, is bringing…
