en.Wedoany.com Reported - AT&T recently stated that for the first time in a decade, enterprise customers are beginning to inquire about edge computing services, a demand primarily driven by the need for localized deployment of AI inference. Andy Foerstner, Managing Director of AT&T Strategic Managed Services, noted that edge computing has long been viewed as technically feasible but lacking a clear commercial market opportunity, a situation that is now changing.

Foerstner said AT&T has received requests from customers for local instances of large language models, but the company has not yet launched corresponding productized services. AT&T is actively listening to market and customer needs, aiming to integrate edge computing with the company's overall value proposition—namely, converged connectivity—to deliver optimal reliability. One unresolved core issue is whether enterprise AI inference should run on customer-owned premises or adopt a GPU-as-a-service model (new cloud) similar to CoreWeave. Foerstner noted that the manufacturing, retail, and transportation industries have varying demands for edge computing, and there is currently no perfect universal use case.

At this week's Cisco conference, company executives pointed out that AI represents a significant opportunity for telecom operators. Cisco CEO Chuck Robbins predicted that network capacity demand could triple within three years and believes that central offices are ideal locations for inference services. Chief Product Officer Jeetu Patel added that inference needs will extend beyond data centers to campuses, branches, and even desktops.

AT&T is collaborating with Cisco on the Smart Edge initiative, aiming to integrate AT&T's converged access and connectivity into a single device, providing a flexible ecosystem of network functions on that device. This initiative is seen as a migration path where customers can start with Layer 3 routing, gradually add SD-WAN, security features through AT&T's SASE partnerships with Cisco and Palo Alto Networks, and network observability via Cisco ThousandEyes, all without needing to return to physical sites. Foerstner stated that his department's services cover SD-WAN, universal CPE with network functions virtualization, SASE, and Wi-Fi (provided through Cisco Meraki and HPE Aruba Central).

In March of this year, AT&T and Cisco demonstrated the Cisco AI Grid implementation in collaboration with Nvidia at the AT&T Dallas Discovery Zone. The demo used Linker Vision cameras and Nvidia computing to monitor public areas for spills, obstacles, and perimeter anomalies in near real-time. AT&T also described a commercial pilot with TanMar Companies, an oilfield services firm in Louisiana, where the platform was deployed for license plate recognition and perimeter monitoring at industrial sites.

Additionally, post-quantum cryptography (PQC) is a priority for AT&T. The company is venturing into this field through the Cisco 8000 series routers. Foerstner believes the Cisco 8000 series is the most advanced hardware solution on the market to address PQC challenges, with no other company currently offering its onboard hardware acceleration capabilities. The U.S. federal government has set the FIPS 203 compliance deadline for 2030, which will be a major driver for government and defense industrial base customers. Cisco plans to release software version 26.2 in August to support PQC on deployed 8000 series routers.

AT&T has made significant investments in infrastructure, spending $22 billion on fiber last year and $23 billion on spectrum acquisition from EchoStar. Foerstner said these investments are being deployed in new ways, but the specific form of AT&T's edge computing products is still being explored.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com