On Thursday, Google and Intel revealed an extended multi-year collaboration, wherein Google Cloud will persist in leveraging Intel’s AI infrastructure and the companies will continue joint processor development.
Google Cloud is set to deploy Intel’s Xeon processors, encompassing their newest Xeon 6 chips, across AI, cloud, and inference operations. This continues a long-standing practice, as Google has relied on Intel’s Xeon processors for many decades.
Furthermore, both companies are set to broaden their joint efforts in creating bespoke infrastructure processing units (IPUs), designed to enhance the speed and oversight of data center operations by shifting workloads away from central processing units.
Commencing in 2021, this chip development alliance will now primarily concentrate on developing custom IPUs based on Application-Specific Integrated Circuits (ASICs).
Intel chose not to disclose any financial details concerning this agreement.
This expanded collaboration arises amidst a significant industry demand for CPUs. Although Graphics Processing Units (GPUs) are utilized for creating and training AI models, Central Processing Units (CPUs) remain vital for executing AI models and foundational AI infrastructure.
In a company press release, Intel’s chief executive, Lip-Bu Tan, stated, “AI is transforming infrastructure design and scalability.” He added, “Effectively scaling AI necessitates not just accelerators, but well-balanced systems. CPUs and IPUs are fundamental to providing the necessary performance, efficiency, and adaptability for today’s AI workloads.”
Lately, an increasing number of enterprises have shifted their attention towards CPUs, prompted by an escalating scarcity of these critical components.
Amidst a global scarcity of CPUs, SoftBank-controlled Arm Holdings recently unveiled the Arm AGI CPU, marking the semiconductor powerhouse’s inaugural self-produced chip.