ARM’s Latest CPUs and GPUs Are Built for AI and Gaming

Why Trust Techopedia
Key Takeaways

  • ARM has unveiled its latest CPU and GPU designs.
  • The Cortex-X925 and Immortalis-G925 deliver especially large boosts for AI and gaming.
  • There are also updated mainstream cores with an emphasis on efficiency.

ARM has introduced its next-generation CPU and GPU design lineups, and the company is clearly targeting the mobile chip range at AI tasks and games.

The lead CPU, the Cortex-X925, is billed as the biggest performance improvement in the X-series’ history. The ARMV9-based hardware is estimated to be 36% faster in single-core performance than the Cortex-X4 it replaces, with a 46% improvement in AI speeds. You should see significant boosts for particularly demanding apps, including generative AI tools and intensive games.

The Cortex-A725 is a successor to the A720, ARM and promises large strides in sustained chip performance. The company claimed a 35% jump in performance efficiency, and a 25% increase in power efficiency.

The Cortex-A520 is a refresh, but a new implementation should offer a 15% energy savings, with power usage cut by as much as half during a typical workload.

The GPUs are revisions of last year’s fifth-generation architecture, but should help with AI and gaming in everything from smartwatches to high-end phones.

The headlining GPU design, the 14-core Immortalis-G925, reportedly manages 37% higher frame rates in mobile games than the 12-core G720. When it matches the older part’s performance, it uses 30% less power. ARM also touted 36% faster inference for AI and a 52% increase in ray-traced lighting performance.

The Mali-G725 has between six and nine cores, and is meant for “premium” smartphones where the Immortalis isn’t an option. A Mali-G625 with one to five cores is built for smartwatches and basic smartphones.

It will take awhile before manufacturers use these newest ARM platforms in chip designs of their own. There’s also no certainty you’ll see these exact cores in your next device. Apple, Qualcomm, and others frequently develop custom chips that may only use ARM’s work on a basic level.

The emphasis on AI is well-timed. On-device AI is quickly becoming a major selling point, with Copilot+ Windows PCs and Apple’s iPad Pro M4 both being aimed at AI users. Google is also rushing to offer Gemini Nano on  Android phones. ARM chip upgrades could both enable more advanced AI and lower the impact on battery life.