乌鸦传媒

In 1952, computer science pioneer , famously said, 鈥淭he most dangerous phrase in the language is, 鈥榃e鈥檝e always done it this way.鈥欌�� For telecom leaders, her words resonate as the sector faces major transformation.

AI is now an operational reality, and 5G networks are evolving into platforms for advanced AI-driven applications. Initiatives like the integrate AI within Radio Access Networks (RAN), enabling telcos to create new revenue streams, offer new services, and enhance network performance.

The AI RAN Alliance is based on three distinct concepts that are gaining significant traction among communications service providers (CSPs) and 5G operators that are eager to maximize returns on their 5G investments: 

1. AI and RAN: Aims to run AI and RAN workloads on a shared accelerated infrastructure to maximize utilization while generating new revenue streams from AI-driven services. 
2. AI on RAN: Focuses on deploying AI services directly at the network edge through RAN, enhancing operational efficiencies and creating new service opportunities for mobile users. 
3. AI for RAN: Focuses on incorporating AI algorithms within radio signal processing to improve network performance and maximize spectral efficiency. The O-RAN community is also working on this concept with the definition of . 

AI applications running at the edge are a major driver for enterprise markets, as the ability to provide AI and gen-AI-as-an-edge-service from shared RAN infrastructure can maximize the return on investment for telcos and enterprises alike. The question is, how can companies jumpstart this journey?   

乌鸦传媒 + NVIDIA: Enabling 5G AI-RAN for the edge 

乌鸦传媒, using NVIDIA AI Aerial Platform, developed an innovative 5G AI-RAN reference architecture that supports these goals. The architecture will allow CSPs to deploy vRAN software as another workload together with AI workloads, allowing both 5G services and AI/gen AI-as-a-service to be orchestrated on a common accelerated compute infrastructure. These co-located services will allow end consumers faster access to AI and gen AI agents, without compromising 5G quality of service. This shift is poised to unlock a new generation of services for CSPs.

At Mobile World Congress (MWC) 2025, 乌鸦传媒 is showcasing its collaboration with NVIDIA through a wind farm demo. We will show how core technology from NVIDIA and 乌鸦传媒, and advanced data capabilities, AI, and digital twin technology, can help teams better manage resources and proactively identify issues. The demo showcases several unique technology components and capabilities, including:  

• A 5G vRAN stack running on the , powered by NVIDIA AI Aerial Layer-1 software libraries and 乌鸦传媒鈥檚 5G RAN Layer-2 and Layer-3 software. The NVIDIA   transforms the way large datasets and compute tasks are co-located and co-processed using acceleration鈥攗nlocking new possibilities for telecommunications. 
• The same NVIDIA GH200 also hosts AI-based application services for a 5G-connected wind farm. In this use case, wind turbines stream sensor data in near real time, enabling AI models hosted on the local 5G AI-RAN server, to process and infer outcomes from hundreds of sensors across the wind farm, and take actions such as modifying the speed or direction of the wind turbine via the 5G network. 
• The wind farm aggregates data from hundreds of wind turbines, each with hundreds of sensors. These sensors generate streams of data vectors that are processed by AI models to infer actionable insights. Importantly, this system implements the concept of federated learning, which allows the aggregation of AI models into a global model. This method delivers accuracy comparable to traditional centralized approaches while significantly reducing bandwidth usage and maintaining data privacy. The AI model developed by 乌鸦传媒 can process sensor inputs like sound from acoustic sensors and use it to reliably detect 鈥渕etal fatigue鈥� and 鈥渧ibration鈥� which helps intervene before a catastrophic failure of the fan-blades. 
• A digital twin of the windfarm provides a visual reference to the streams of data and AI insights being generated. The digital twin combines data from the sensors on the wind turbine as well as insights generated by the AI models and allows operations personnel to navigate and visualize the data. This will help energy companies鈥� predict when they need adjustments to the wind turbine or schedule maintenance or replacement. Windmills connected over carrier 5G/wireless networks will make energy management more intelligent by leveraging telco AI-RAN edge for digital twins of the managed windmills. 
• The dual use of the NVIDIA GH200鈥攆or both RAN connectivity and AI applications鈥攊s orchestrated by 乌鸦传媒鈥檚 EIRAx platform. EIRAx forecasts 5G network-compute demand and intelligently manages GPU resources. 
• When network demand spikes, EIRAx, in combination with 乌鸦传媒鈥檚 Intelligent Edge Application Platform (IEAP), can relocate lower-priority AI applications from one AI-RAN server to another edge server. This ensures uninterrupted service levels for both the 5G network and the wind farm AI applications. 

鈥淚nferencing for AI applications will place unique demands on mobile networks, making AI-ready distributed infrastructure an essential part of their network architecture,鈥� said Soma Velayutham, Vice President, Telecoms, NVIDIA. 鈥淎I-RAN allows this infrastructure to serve both 5G and AI needs, benefiting operators and customers alike 鈥� and 乌鸦传媒鈥檚 full-stack AI-RAN showcase using NVIDIA AI Aerial is a significant step toward advancing the AI-RAN ecosystem and commercialization.鈥� 

3 steps to tap the potential of AI-RAN for CSPs  

The NVIDIA GH200 is an example of how innovation can challenge ways of thinking and open new horizons. By fusing AI with RAN, telcos can move beyond connectivity and become providers of next-gen AI services. This isn’t just about technology鈥攊t’s ushering in a new era for the telecommunications industry, where AI is not just a tool but the very foundation for building and managing networks.

To seize this opportunity, operators should focus on three key actions:

1. Investing in edge infrastructure that supports both network and AI applications. Platforms like 乌鸦传媒鈥檚 IEAP and EIRAx ensure critical workloads are optimized in real time, enhancing customer experiences and operational efficiency.
2. Integrating AI as a core function of their network architecture. With platforms like the NVIDIA GH200 Superchip and 乌鸦传媒鈥檚 orchestration solutions, CSPs can stay competitive and deliver intelligent, agile services to their customers.
3. Adopting federated learning to manage AI workloads efficiently, reduce bandwidth usage, and maintain data privacy, unlocking new AI-driven revenue streams without compromising network performance.

Meet us at Mobile World Congress 2025 (March 3-6) at booth 2K21 in Hall 2 to experience the demo.

Telco乌鸦传媒 is a series of posts about the latest trends and opportunities in the telecommunications industry 鈥� powered by a community of global industry experts and thought leaders.