Google and Intel expand collaboration on AI infrastructure

Google and Intel have announced an expanded multiyear partnership to strengthen AI infrastructure and continue joint processor development. The announcement was made on Thursday, signalling a deeper collaboration between the two companies in the evolving AI ecosystem.

As part of the agreement, Google Cloud will continue to rely on Intel’s Xeon processors for a range of workloads, including artificial intelligence, cloud computing, and inference tasks. This includes adopting Intel’s latest Xeon 6 chips. Google has a long history of using Intel’s Xeon processors, spanning several decades.

The partnership will also extend into the co-development of custom infrastructure processing units, or IPUs. These specialised processors are designed to handle and accelerate data centre operations by offloading certain workloads from traditional CPUs, improving overall efficiency and performance.

This joint chip development initiative originally began in 2021 and will now focus more heavily on custom ASIC-based IPUs. These chips are tailored to meet specific infrastructure needs, particularly in large-scale data centres supporting AI workloads.

Intel did not disclose any financial details related to the expanded agreement.

The move comes at a time when demand for computing hardware continues to rise sharply. While GPUs remain essential for training AI models, CPUs play a critical role in running those models and supporting broader AI infrastructure systems.

Intel CEO Lip-Bu Tan highlighted the importance of balanced computing systems in the company’s announcement. He noted that scaling AI requires more than just accelerators like GPUs, emphasising that CPUs and IPUs are key to delivering the performance, efficiency, and flexibility required for modern AI applications.

In recent months, more companies have been shifting attention toward CPUs due to increasing supply constraints. The growing demand for chips across industries has contributed to a global shortage, pushing companies to explore new solutions and partnerships.

Amid this trend, Arm Holdings, owned by SoftBank, recently introduced the Arm AGI CPU — marking the first time the company has produced its own chip — as part of efforts to address the ongoing demand for processing power.

The expanded collaboration between Google and Intel reflects the broader industry push to build more efficient and scalable infrastructure to support the rapid growth of AI technologies.