AT&T and Cisco pilot edge AI inference for surveillance and security

Additional Context

AT&T’s move toward edge AI coincides with a massive expansion of its underlying infrastructure. Per AP News and Bloomberg (August 2025), AT&T secured a $23 billion all-cash deal to acquire 50 MHz of nationwide low- and mid-band spectrum from EchoStar. This acquisition, expected to close by mid-2026, focuses on 3.45 GHz and 600 MHz licenses to bolster 5G capacity and home internet growth. The carrier also pledged in late 2024 to extend its fiber footprint to 50 million locations by 2029, creating a high-capacity baseline for the low-latency services required by edge AI applications. The competitive landscape is following a similar trajectory. According to RCR Wireless and Fierce Network (March 2026), Comcast, Spectrum, and T-Mobile are also building Nvidia-powered "AI grids" to transform their networks into distributed compute platforms. This momentum is driven by a surge in the AI video surveillance market, which Grand View Research estimated at $6.51 billion in 2024 and projected to grow at a 30.6% CAGR through 2030. Industry reporting suggests that hardware currently accounts for over 40% of this revenue, as enterprises prioritize specialized GPUs and AI-capable cameras for local processing. Security remains a primary friction point for these distributed architectures. Per Cisco and Orange Business (February 2026), service providers are increasingly adopting post-quantum cryptography to defend against "harvest now, decrypt later" threats. Cisco’s roadmap commits to embedding quantum-safe communications across its core portfolio by December 2026, aligning with U.S. federal mandates for FIPS 203 compliance by 2030. This security layer is becoming a standard requirement for industrial and government clients who refuse to route sensitive video feeds or proprietary data through centralized public clouds.