AT&T and Ericsson demonstrated AI-native link adaptation running on a Cloud RAN platform powered by Intel’s Xeon 6 SoC, highlighting how portable AI software can improve radio network performance on commercial off-the-shelf hardware. The test validated Ericsson’s AI-native Link Adaptation running on AT&T’s target Cloud RAN configuration and confirmed measurable gains in throughput and spectral efficiency compared with conventional rule-based algorithms.
The demonstration used Ericsson’s hardware-agnostic Cloud RAN stack running on Intel Xeon 6 SoC infrastructure and tested performance using AT&T’s frequency bands and propagation characteristics. The companies reported throughput gains of up to 20 percent compared with traditional link adaptation methods. Ericsson’s AI models dynamically adjust transmission parameters in real time based on channel conditions and interference, allowing the system to determine how much data to transmit at any given moment.
The collaboration also demonstrated the portability of Ericsson’s commercial RAN software across cloud infrastructure platforms. Ericsson and Intel are benchmarking AI models on commercial server hardware, positioning the approach as a scalable path for deploying AI functionality across AT&T’s RAN environment as the operator expands its Cloud RAN architecture.
• AI-native Link Adaptation uses real-time machine learning models to adjust transmission rates based on changing radio conditions.
• Tests showed up to a 20% improvement in throughput and corresponding gains in spectral efficiency.
• Ericsson’s Cloud RAN software runs on commercial off-the-shelf infrastructure, including Intel Xeon 6 SoC processors.
• Intel Xeon 6 integrates AVX and AMX acceleration to support AI inference workloads for RAN processing.
• The demonstration validated software portability across hardware platforms within AT&T’s open Cloud RAN architecture.
“AT&T is leading the charge toward an open, intelligent, and scalable network future by advancing Open RAN and Cloud RAN with AI-native capabilities at their core,” said Rob Soni, Vice President of RAN Technology at AT&T.
🌐 Analysis: Operators increasingly deploy AI inside the radio stack to improve spectral efficiency and automate network optimization. Ericsson has expanded AI capabilities across its RAN portfolio while Intel positions Xeon 6 as a general-purpose platform capable of supporting AI-driven RAN workloads without discrete accelerators. AT&T has been among the most aggressive Tier-1 operators moving toward Open RAN and Cloud RAN architectures, alongside similar initiatives at Vodafone, Deutsche Telekom, and NTT DOCOMO.
