We have come across multiple rumors in the past claiming that Google will switch to TSMC for its Tensor chips. Previously leaked documents have already revealed that Google will switch to TSMC for the fabrication of the Pixel 10 series’ Tensor G5 chipset.
Pixel devices aren’t the best when it comes to power efficiency
While Pixel devices are among the smartest smartphones on the market, they are not the most power-efficient. Power efficiency is crucial for a device that relies on limited battery power. A less efficient device requires better cooling solutions or must reduce performance to operate safely. We’ve seen numerous reports in the past about Pixel devices getting hotter than expected under certain use cases.
The higher-than-usual power consumption and occasional overheating are attributed to the process nodes on which Tensor chips are fabricated. For those unfamiliar, Google’s Tensor chips are produced by Samsung Foundry.
Things could change with the Tensor G5 powered Pixel 10 series
Now a report out of Taiwan claims that Google has partnered with TSMC and has already made some progress in the development of the Tensor G5. Notably, the Tensor G5 will use TSMC’s 3nm process. Google is also reportedly investing in independent research and development of its chips.
Tensor G5 is going to be Google’s first independently designed smartphone chip. All the Tensor chips Google has used in Pixel devices so far were customized by Samsung. In contrast, Google Tensor G5 will be the first to adopt the company’s own architecture design, which is expected to significantly improve the performance of the chip. It not only holds the potential to solve some of the glaring issues of Tensor chips but also offers Google more control over its chips.
According to the report, Google will not only benefit from a higher power efficiency on the Tensor chips but will also be able to offer a more capable AI experience on Pixel devices with this development. The integration between hardware and software is also expected to improve with the use of Google‘s own architecture.
(Source)
Comments