How Google’s Tensor G5 Chip Could Deliver Improved Efficiency
Google’s upcoming Pixel smartphone models could offer improved efficiency compared to the company’s existing devices, according to a report. After using Samsung’s Tensor G3 chips for years, recent reports suggest that Google is planning to switch from Samsung to Taiwan Semiconductor Manufacturing Co (TSMC) for future processor requirements. While news of the switch has been reported multiple times, the upcoming Pixel 9 series is expected to feature Samsung’s Tensor G4 SoC, which reportedly uses a different packaging method to offer notable improvements over last year’s Tensor G3 chip. A new report suggests that next year’s Tensor G5 processor could use the 3nm process made available by TSMC.
A business Korea report states that Samsung’s foundry has been struggling with yield and power efficiency issues with its 3nm technology for some time, and concludes that the company will eventually lose customers to TSMC’s 3nm process. These customers are said to include big names such as Nvidia, Intel, Qualcomm, Google and more. The report, citing an industry source, claims that Google will opt for TSMC’s 3nm technology.
The detailed report reveals that Samsung will continue to supply processors to Google until the fourth generation of Tensor processors, but will switch to TSMC for the fifth generation, which could be called the Tensor G5, provided Google continues to use the same name for its processor.
Google also appears to be looking to improve the efficiency of its processor, reportedly switching to TSMC’s 3nm technology for its Tensor G5 processor. This could finally bring its processor and phones up to par with Apple’s iPhone, which currently uses an industry-leading 3nm process.
Samsung was the first in the industry to go with the first-generation 3nm node in 2022, but still struggles to deliver the required level of power efficiency and thermal management. TSMC’s process offers better power efficiency compared to Samsung’s and has therefore emerged as the go-to option for most manufacturers going forward.
The process node has proven useful for thermal management, which has important implications not only for smartphones but also for servers. Adding AI capabilities would further increase the overall load, making thermal and efficiency issues more critical.
Meanwhile, TF International Securities analyst Ming-Chi Kuo recently claimed that Samsung is unlikely to use its Exynos processor for its Galaxy S25 series. If this happens, it would mean Samsung will be going back to Qualcomm for processors in its flagship Galaxy smartphone series, after switching to its own Exynos SoCs for the Galaxy S24 and Galaxy S24+ in select markets (including India) this year.
