Ericsson Launches AI in RAN Software to Boost 5G Performance, Energy Efficiency, and AI-Native Network Evolution

Why It Matters

By turning existing 5G infrastructure into an AI‑powered platform, operators can extract more capacity and energy savings, delivering higher‑quality services while deferring costly hardware upgrades.

Key Takeaways

• AI in RAN adds up to 20% downlink throughput boost.
• Spectral efficiency improves by up to 10% with AI-driven scheduling.
• Energy use drops as AI inference runs on Ericsson Silicon.
• Supports twice the number of high‑traffic users on existing hardware.
• Enables 5x more precise user positioning for location services.

Pulse Analysis

The race to embed artificial intelligence into mobile networks has accelerated as operators seek to meet exploding data demand and the low‑latency requirements of emerging applications such as augmented reality and autonomous vehicles. Ericsson’s AI in RAN software marks the first fully integrated, subscription‑based AI layer that runs directly on baseband and radio hardware, eliminating the need for separate edge servers. Leveraging the company’s proprietary silicon accelerators, the platform delivers microsecond‑scale inference, allowing the network to react instantly to changing radio conditions and traffic patterns.

Early deployments across more than 15 global trials demonstrate tangible performance lifts. Operators reported up to a 20% increase in downlink throughput and a 10% boost in spectral efficiency, while the AI‑native scheduler and beamforming modules support roughly twice the number of high‑traffic users on the same spectrum. Energy consumption also falls, as the AI inference workload is handled by Ericsson’s low‑power silicon rather than power‑hungry general‑purpose processors. The software’s continuous‑learning engine refines models from live network data, ensuring that optimization improves over time without manual tuning.

From a business perspective, the upgrade offers one of the strongest ROI propositions in recent telecom history: a pure software enhancement that extracts additional capacity from existing 5G assets while extending equipment lifespan. Analysts predict that AI‑driven RAN will become a prerequisite for 5G‑Advanced services and a foundation for the forthcoming AI‑native 6G era. As rivals such as Nokia and Samsung roll out competing AI solutions, Ericsson’s early‑move and extensive trial base give it a strategic edge, potentially shaping the next wave of network‑as‑a‑service offerings.