Samsung’s focus on AI functionality is set to continue into next year, as indicated by insights from a prominent source, In a move similar to incorporating an AMD-designed GPU into its flagship chips, Samsung plans to integrate a Google-designed Tensor Processing Unit (TPU) into its Exynos 2500 chipset.
The TPU, a core component of Pixel phones for several years, forms an integral part of the Tensor chipsets, jointly developed by Google and Samsung. This specialized processing unit is one among various AI accelerators and finds extensive usage across Google’s products. Larger versions of the TPU are available for rent via Google Cloud, particularly beneficial for AI-related tasks.
This integration will grant devices utilizing the Exynos 2500, such as the Galaxy S25 series, access to Google’s machine learning APIs, enabling the execution of models directly on the device. Presently, Galaxy AI heavily relies on cloud resources, given the inadequacy of AI hardware in the Exynos 2400.
Recent leaks suggest that the Galaxy S25 series will employ both Exynos 2500 and Snapdragon 8 Gen 4 chips, contrary to previous rumors suggesting only Exynos chips would be used. Qualcomm has also been emphasizing AI acceleration, with the 8 Gen 3 capable of handling models with up to 10 billion parameters. However, Qualcomm’s AI Engine differs from Google’s TPU.
Insider information hints that the Exynos 2500 will be manufactured using Samsung’s second-generation 3nm node (3nm GAP/SF3). It will feature a Cortex-X5 clocked at 3.2GHz, three A730 cores running at 2.3-2.5GHz, two additional A730 cores operating at a lower clock speed, and four A520 cores. The GPU will be an Xclipse 950, though specifics remain undisclosed.
