SoftBank and Ericsson verified Ericsson's AI in RAN software on a commercial 5G network in Japan.
Spectral efficiency rose up to 25% and downlink throughput up to 50%.
Both metrics improved about 10% on average everywhere.
What happened
SoftBank and Ericsson Japan demonstrated Ericsson's "AI in RAN" software, specifically the AI-native link adaptation scheduler, on SoftBank's 5G commercial network — the first such verification in Japan as of July 31, 2026.
Why it matters
The trial showed spectral efficiency (how much traffic a given frequency bandwidth can carry) improved by up to about 25% and downlink user throughput by up to about 50% compared with conventional technology. Across all evaluation areas, both metrics improved by an average of about 10%, indicating consistent performance gains.
What to watch
The companies say this is a step toward AI-native mobile networks that can handle growing traffic from AI assistants, autonomous agents, and immersive apps. They plan to continue collaborating, combining SoftBank's network insights with Ericsson's AI in RAN expertise to improve performance and evolve the software.
Ask the AI about this article →
The verification is part of a broader collaboration between SoftBank and Ericsson toward AI-native RAN, aiming to prepare networks for the growing use of AI services. The press release notes that traffic growth alone isn't enough — RAN must adapt flexibly to changing wireless conditions and traffic patterns. By applying AI directly to the RAN, the companies showed that existing network resources can be used more efficiently, improving performance without requiring extra spectrum or hardware.
Both companies see this as a milestone for future network generations, including 5G-Advanced and 6G. SoftBank's CNO emphasized that mobile networks must handle larger and more variable traffic as AI adoption spreads, while Ericsson's executives highlighted the role of telecom-grade AI and the use of existing Ericsson silicon. The trial is a concrete step toward the companies' shared goal of building high-performance, secure, and flexible networks that can support the next wave of AI applications.
For example, today's edition would include:
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. Q&As are published on this page for other readers too.
AI system scaling has pushed interconnect requirements inside data centers from chips and boards up to racks…
Chinese large-model developer Z.ai says it can now support large-scale inference using roughly 100,000 domesti…
Analyst Ming-Chi Kuo says Nvidia has revived the Rubin CPX AI accelerator with a substantially redesigned arch…
Palantir Technologies stock has posted multi-year gains, including an 11x return over 3 years
Apple has escalated its legal battle against OpenAI, claiming in a new court filing that OpenAI is actively de…
Samsung Electronics has locked up as much as 70% of its memory production capacity under long-term supply agre…
