Ericsson has launched a new software offering designed to embed artificial intelligence directly into radio access networks (RAN), enabling mobile operators to improve network performance, efficiency and automation using existing infrastructure.
The solution, called AI in RAN, was introduced on 11 June and is available as a software subscription for Ericsson 5G Advanced networks. According to the company, the platform allows telecom-grade AI models to operate in real time within basebands and radios, supporting the industry’s transition toward AI-native network architectures.
AI Integrated into Existing Network Infrastructure
Ericsson said the software enables communications service providers to deploy advanced AI capabilities without requiring new hardware installations.
The platform combines real-time AI inference, continuous learning software trained on large-scale network data and support for Agentic AI applications designed to automate network operations.
The company noted that the technology is built to run at microsecond-level latency, allowing AI models to respond rapidly to changing network conditions.
“Ericsson is redefining what’s possible in mobile networks by bringing powerful AI capabilities to service providers,” said Mårten Lerner, Head of Networks Strategy and Product Management at Ericsson.
“With AI in RAN software, we are taking a major step toward AI-native networks, alongside the AI-ready radios we unveiled in February.”
Improving Performance and Energy Efficiency
According to Ericsson, AI in RAN is designed to optimize radio resources, increase throughput and improve energy efficiency through intelligent automation.
The software leverages Ericsson Silicon for energy-efficient AI inference within radios and RAN compute platforms while maintaining compatibility across both traditional and cloud-based RAN deployments.
“Service providers need networks that are high-performing, secure and efficient in the AI era,” Lerner said.
“With AI in RAN, Ericsson is bringing AI into networks to elevate 5G performance and efficiency through energy-efficient AI inference at scale.”
Initial Features Already Available
The first set of AI in RAN capabilities became commercially available earlier this year, with additional features scheduled for release later in 2026.
Current features include:
- AI-native scheduling and link adaptation
- AI-powered macro positioning
- AI-managed beamforming
- AI-powered multi-layer coordination
- Enhanced observability for AI-driven network operations
- Performance management event schema support
Ericsson said these functions help operators improve radio performance while reducing operational complexity.
Deployment Momentum Builds
The company reported that AI in RAN has already been deployed or tested in more than 15 projects worldwide.
Ericsson claims the technology has demonstrated measurable improvements, including:
- Up to 20% higher downlink throughput
- Up to 10% greater spectral efficiency
- Support for up to twice as many high-traffic users
- Coverage prediction accuracy of up to 95%
- Up to fivefold improvement in user-positioning precision
Several telecom operators have participated in trials and deployments, including SoftBank Corp, Bell, SK Telecom and Rogers Communications.
Advancing Toward AI-Native Networks
The launch reflects a broader industry trend toward integrating AI directly into telecommunications infrastructure as operators seek to maximize the value of 5G investments and prepare for future network technologies.
By embedding AI capabilities directly within the radio network, Ericsson aims to help operators deliver higher performance, improve user experience and automate increasingly complex network environments.
The company views AI in RAN as a foundational step toward fully AI-native mobile networks capable of continuously adapting to traffic demands, user behavior and service requirements in real time.
Source: DataMagz