Google and AMD Reportedly Partnering on New Tensor Chip with On-Package Cores

Google and AMD Reportedly Unite on Custom AI Silicon

Google and Advanced Micro Devices (AMD) are reportedly collaborating on a custom Tensor Processing Unit engineered for heavy artificial intelligence workloads. The partnership surfaced via a report from Tom’s Hardware, citing intelligence shared by user Sean on X. Neither corporation has officially confirmed the arrangement.

The alliance bridges Google’s proprietary AI silicon initiatives with AMD’s specialized engineering capabilities. Historically, Google has independently designed and manufactured its TPUs without external hardware partners. These chips currently drive on-device AI tools like Gemini on Pixel smartphones alongside massive data pipelines across Google Cloud data centers.

Seeking External Engineering Expertise

Google appears to be pursuing outside engineering support in specific semiconductor manufacturing and design domains where AMD boasts proven market dominance.

Inside the Rumored Tensor Hardware

The joint venture centers on technical arenas demanding outside proficiency, notably CPU intellectual property and programming logic. Industry reports indicate the rumored chip will feature on-package CPU cores.

That architectural decision aims to maximize power efficiency while simultaneously boosting general-purpose Tensor compute performance.

Hyperscaler Stakes and Pixel Futures

For AMD, working alongside Google unlocks a lucrative revenue stream. Hyperscaler tech giants continue to invest heavily in custom silicon to fuel expanding AI infrastructure.

Even as this joint Tensor chip targets enterprise-grade workloads, Google maintains separate hardware tracks. The tech giant is independently developing the Tensor G7 processor slated for upcoming Pixel devices.