AT&T uses Microsoft Foundry and AMD to scale Open Telco AI models for telecom workloads

Telecom AI Deployment

AT&T created Open Telco (OTel) 2.0 models to integrate telecommunications expertise into AI systems. These models address the limitations of generic AI in understanding specialized telecom networks and operations. The development of OTel 2.0 required a platform capable of supporting large-scale AI development while managing costs and operational complexity.

Scaling with Microsoft Foundry

AT&T utilized Microsoft Foundry Managed Compute to scale its OTel 2.0 development. This platform provides streamlined access to dedicated GPU capacity, enabling AT&T to experiment with multiple open models and optimize workloads across different GPU architectures. The result is an AI development environment that supports trillions of tokens, offering flexibility for iteration and optimization.

Multi-Model Strategy

AT&T adopted a multi-model strategy for OTel 2.0, deploying several models from the Hugging Face collection, including Phi-4, OSS-120B, and Gemma-4, through Microsoft Foundry. This approach allows AT&T to work with approved telecom data, tailor workflows for domain-specific model development, and conduct large-scale experimentation with control over cost and deployment.