Google confirms that the Tensor G2 is actually a 5nm chip
During its Made by Google October event, the tech giant revealed the Tensor G2. For quite some time now, the Internet assumed that under the Pixel 7's hood beats a 4nm silicon due to rumors and Google not confirming the Tensor G2's process technology.
However, it appears that the Internet was wrong. The Tensor G2 is actually a 5nm SoC, not a 4nm one. This has been confirmed to Android Authority by a Google spokesperson. The representative stated that the Tensor G2 has been made for real-world use cases and that the 5nm architecture has helped reach that goal while also improving the chipset's performance and power efficiency.
But Google's answer still doesn't reveal the manufacturer of the Tensor G2. If we had to guess, we would say that the chipset was made by Samsung. Both companies have partnered before in creating the first generation Tensor chip, so Google might have used Samsung's technology again for its next-gen in-house SoC. But which of the two Samsung 5nm nodes has been used remains a mystery.
As you probably know, a smaller nm process usually means better power efficiency. And at the moment, all flagship SoCs like Apple's A16 Bionic, Qualcomm's Snapdragon 8+ Gen 1, and Samsung's Exynos 2200 are built on 4nm nodes.
However, it appears that the Internet was wrong. The Tensor G2 is actually a 5nm SoC, not a 4nm one. This has been confirmed to Android Authority by a Google spokesperson. The representative stated that the Tensor G2 has been made for real-world use cases and that the 5nm architecture has helped reach that goal while also improving the chipset's performance and power efficiency.
As you probably know, a smaller nm process usually means better power efficiency. And at the moment, all flagship SoCs like Apple's A16 Bionic, Qualcomm's Snapdragon 8+ Gen 1, and Samsung's Exynos 2200 are built on 4nm nodes.
Google, however, appears to be betting more on software than hardware. As Google's spokesperson told Android Authority, the Tensor G2 uses a next-gen TPU (Tensor Processing Unit), which advances even further the chipset's machine learning capabilities.
Things that are NOT allowed: