NVIDIA’s GeForce RTX 3060 is making a comebackscheduled for mid-March. To do this, NVIDIA will once again use Samsung’s 8nm DUV node, as it has done in the past. This has been confirmed by the Korean media Hankyung, which reports that Samsung is restarting its production of 8nm nodes to meet NVIDIA’s needs. These GPUs were originally manufactured by Samsung in 2021. The entire NVIDIA “Ampere” architecture line was produced on the 8nm DUV node and we do not anticipate their return after several years. However, given that NVIDIA transitioned to TSMC to manufacture its “Ada Lovelace” and “Blackwell” GPUs, and became TSMC’s largest customer, using the 5nm node, this move is intriguing.

We are still missing concrete information on which version of the RTX 3060 will be reintroduced, whether it will be the original 12GB model with a 192-bit wide memory bus or the newer 8GB variant with a 128-bit bus. Additionally, the decision to use a two-generation GPU architecture in 2026 is puzzling, as the reason NVIDIA chose the RTX 3060 instead of a newer model like the RTX 4060 is still unclear. Speculatively, it could be because the RTX 4060 is based on the same NVIDIA 4N node (5nm class) at TSMC as the current RTX 5060, while the RTX 3060, along with the rest of the GeForce “Ampere” generation, is based on the Samsung 8N foundry node (8nm DUV), which would leave 5nm capability for “Blackwell” and its business variants. Finally, it’s worth noting that when GPU IP design is done, it’s usually tied to the node it was prepared for, so NVIDIA is sticking with Samsung again to avoid potential upfront costs of porting this GPU for a different node.