Before the October 6 press conference, it was reported that the second-generation Tensor processor on the Pixel 7 series is based on Samsung’s 4nm process. However, a Google spokesperson made it clear that the Tensor G2 uses a 5nm process from Samsung’s factory.
In a statement submitted to Android Authority, a Google spokesperson said: “We built Google Tensor G2 specifically for real-world use cases. Our final architecture (including 5nm) helped us achieve this goal while improving performance and Energy efficiency. This approach also enables us to add new capabilities while taking a step forward in machine learning using our next-generation TPU and G2.”
So Tensor G2 is actually a 5nm process chip. However, Google did not tell the exact node process and manufacturer, but it can be confirmed that Samsung is Google’s main partner. On Samsung’s roadmap, there are two 5nm process nodes, the 5LPE used by the original Tensor and the subsequent 5LPP. It’s just that it’s not yet certain which node process Tensor G2 uses.
Either way, given that companies like to tout their use of smaller nodes, Google’s wording makes it clear that the Tensor G2 is not built on what Samsung (or TSMC) calls 4nm, such as the 4LPE nodes used in the Exynos 2200 and Snapdragon 8 Gen1.
