As AI-driven services expand, mobile networks are being pushed to deliver higher performance with smarter automation. Ericsson is responding with AI in RAN, a new software subscription that injects telco-grade AI models directly into basebands and radios to improve efficiency, performance, and energy savings.
This is a commercially scalable offering that gives communications service providers immediate benefits for 5G networks while supporting the transition to AI-native RAN without requiring new hardware.
AI in RAN brings three key capabilities:
- Telco-grade AI models engineered to run in real time within the radio access network.
- Continuous learning software powered by scalable, high-quality data.
- Agentic AI support for advanced RAN automation and network operations.
Mårten Lerner, Head of Networks Strategy and Product Management at Ericsson, said the company is redefining what mobile networks can achieve by delivering powerful AI capabilities to service providers. He noted that AI in RAN marks a major step toward AI-native networks, complementing the AI-ready radios Ericsson unveiled in February.
The telco-grade AI models are built for ultra-low latency inference at the microsecond level, delivering high reliability across diverse and dynamic RAN environments. Developed with deep domain expertise and trained on high-quality data, these models set new standards for performance and efficiency.
Nitin Bansal, Managing Director of Ericsson India, emphasized that service providers need networks that are high-performing, secure, and efficient in the AI era. With AI in RAN, Ericsson is bringing AI into networks to elevate operational efficiency, boost network performance, and open new business opportunities.
The solution works with Ericsson 5G Advanced across purpose-built and Cloud RAN platforms, enabling new AI-driven services. It puts the company’s AI for Networks vision into practice by using AI to improve performance and efficiency.
AI in RAN uses the right AI model in the right part of the radio network, powered by Ericsson Silicon for energy-efficient AI inference in radios and the latest RAN Compute generation. Cloud RAN software portability allows AI capabilities to be deployed across partner platforms.
The first AI in RAN features are available in Q2 2026, with more enhancements coming later in the year. Features include AI-native Scheduler for Link Adaptation, AI-powered Macro Positioning, AI-managed Beamforming, AI-powered Multi-layer Coordination, Performance Management Event Schema Files, and Augmented Observability for AI in RAN.
Said to be proven in more than 15 deployments and trials worldwide, Ericsson’s AI in RAN delivers up to 20 percent higher downlink throughput and up to 10 percent better spectral 
efficiency. It also supports up to twice as many high-traffic users, achieves 90 to 95 percent coverage prediction accuracy, and delivers up to five times greater user-positioning precision.
Ericsson has introduced AI functionality across its products since 4G, and in 2021 added AI?ready acceleration in RAN Compute.
In February this year, the company unveiled Neural Network Accelerators in its Massive MIMO radios, increasing AI inference capability by 10 times.
Learn more about Ericsson’s AI in RAN subscription, Here
