AMD and KDDI team up to boost 5G virtual network performance

KDDI and AMD’s new partnership marks a significant step in Japan’s 5G infrastructure development, bringing chiplet-powered processing to mobile networks. The collaboration aims to address the growing demand for 5G capacity while prioritizing energy efficiency—a critical balance as mobile data consumption continues to surge globally. This partnership also highlights the strategic importance of processor architecture in preparing telecom networks for AI integration.

The big picture: Japanese telecom giant KDDI will deploy AMD’s 4th Generation EPYC processors across its 5G virtualized network infrastructure, with nationwide implementation beginning in 2026 following a testing phase in 2025.

Why this matters: The partnership addresses two critical telecommunications challenges simultaneously—scaling network capacity to meet growing 5G demand while reducing energy consumption in data-intensive operations.

• AMD’s EPYC processors utilize advanced chiplet technology that packs more processing cores into each CPU, delivering higher performance with lower power requirements.
• The energy efficiency improvements could serve as a model for other carriers seeking to balance network expansion with sustainability goals.

What they’re saying: KDDI’s president Hiroji Matsuda emphasized the dual benefits of the partnership, focusing on both performance enhancement and environmental responsibility.

• “We’re proud to work with AMD to improve our 5G network’s performance and reduce energy use. We believe AMD’s top-tier chip technology will help us build better data centers for the AI era,” Matsuda stated.
• AMD CEO Lisa Su highlighted the processors’ role in positioning KDDI for future AI integration: “Our EPYC processors will help them deliver faster, more efficient networks and prepare for the next generation of AI-powered data centers.”

Looking ahead: Beyond immediate 5G improvements, the partnership will explore optimizing EPYC processors specifically for AI workloads in telecommunications infrastructure.

• The companies plan to test various CPU implementations, with particular focus on AI-driven data center applications.
• This collaboration signals how telecommunications infrastructure is increasingly being designed with AI capabilities as a core consideration rather than as an afterthought.