AT&T Pushes 'Connected AI' for Factories

AT&T's "Connected AI for Manufacturing" platform is a collaboration with Microsoft, Nvidia, and the edge AI specialist MicroAI. The solution leverages Microsoft Azure's OpenAI Service to power generative AI capabilities at the edge, allowing operators to interact with machinery using natural language. This enables workers to query equipment, diagnose issues, and receive immediate, context-aware recommendations for action. The platform incorporates Nvidia's Metropolis for Factories framework, specifically using its Video Search and Summarization (VSS) blueprint. This allows manufacturers to apply vision-language models to their camera feeds, making it possible to search for and analyze physical events, such as product defects or safety incidents, in near real-time without manually reviewing hours of footage. This is all powered by Nvidia's accelerated computing hardware on-site. For on-premise edge computing hardware, AT&T's Multi-Access Edge Computing (MEC) offerings include solutions built with Dell and Hewlett Packard Enterprise (HPE). The "AT&T MEC with Dell Apex" provides an as-a-service private cloud using Dell's hardware and VMware's virtualization software. Separately, AT&T also utilizes HPE's Edgeline Converged Edge Systems, which are powered by Intel processors and designed for rugged environments. While AT&T has not yet named public customers for this specific manufacturing platform, early pilot deployments have yielded significant results. These include a reduction of up to 80% in waste on injection molding lines, a 35% improvement in fulfillment center efficiency, and the ability to detect potential equipment failures 2.5 to 4 hours in advance. This strategy of integrating with major tech players extends to the broader cloud infrastructure. AT&T is using Microsoft Azure for its enterprise edge solutions like "Connected Spaces" while partnering with Amazon Web Services (AWS) for direct, low-latency fiber and 5G connections into AWS data centers via a service called AWS Interconnect. This dual-hyperscaler approach aims to provide enterprises with flexible and optimized connectivity for various AI workloads.