Current State and Near-Term Trajectory As of September 2026, AT&T does not offer a qualifying commercial product. Its flagship edge-AI initiative—the March 2026 collaboration with Cisco and NVIDIA—fails the core criteria on multiple fronts. Reporting confirms that the GPUs for this "network-driven edge AI" are located in six regional Cisco data centers, rather than being physically hosted in AT&T's own central offices rcrwireless.com. Furthermore, AT&T has explicitly positioned this effort
Evaluating this telco's asset-light network strategy alongside similar connectivity providers delayed the median launch estimate slightly to mid-2035 and shifted the upper bound to 'never', recognizing the high likelihood the strict facility-ownership criteria are never met.
Current Strategy and Capabilities
As of September 2026, Verizon is actively positioning itself in the AI infrastructure space, but it has not released a product meeting the strict bundle criteria. The "Verizon AI Connect" suite markets GPU-as-a-Service alongside connectivity, but the GPUs are currently supplied and managed by partners. Most notably, Verizon's partnership with Vultr involves Vultr deploying its own GPU assets into Verizon's edge/central-office facilities, with Verizon acting
Comparing this timeline against those of other major telecom operators pushed the median later into 2031 and extended the upper tail to "never," reflecting the shared, enduring structural disincentives for connectivity providers to take on the financial risks of directly owning and billing for AI hardware.
Current Strategy and Status Quo As of September 2026, Lumen has no qualifying GPU-as-a-Service product, and its public strategy explicitly points away from hosting compute. Lumen positions itself as the "trusted network for AI" rather than a compute provider 2 sources. Recent earnings calls, Investor Days, and product materials consistently frame the company's role as providing the connectivity layer—such as Private Connectivity Fabric (PCF) deals, Multi-Cloud Gateway, and Dedicated I
Comparing Lumen's asset-light strategy and strict compute hosting criteria against similar telecommunications operators pushed our estimated median launch date later and shifted the upper percentiles to 'never'.
Status Quo: Centralized Strategy
As of September 2026, Deutsche Telekom (DT) is arguably the most aggressive European telco on AI infrastructure, but its build-out remains heavily centralized. DT and T-Systems launched the Industrial AI Cloud into general availability in February 2026, backed by a ~€1bn investment and nearly 10,000 NVIDIA Blackwell GPUs telekom.com. However, this facility is a repurposed bank vault in Munich's Tucherpark, not a distributed central office (CO) o
Evaluating this timeline alongside other top-tier telcos pushed the median into the early 2030s and shifted the right tail to 'never', reflecting the substantial probability that telecommunications firms maintain centralized or partner-led architectures rather than fully integrating these complex, single-bill edge deployments.
Status Quo and the Nscale Partnership As of September 2026, BT has no generally available GPU-as-a-Service product hosted in its own facilities that meets the strict bundled criteria. BT's recently launched "Sovereign Cloud" is underpinned by Rackspace Technology's UK data centers rather than BT's own central offices 2 sources. The most viable near-term pathway to resolution is an April 2026 announcement in which BT and Nscale revealed plans for up to 14MW of NVIDIA-powered AI data ce
Evaluating BT's strategy alongside comparable telecom infrastructure transitions delays the expected timeline for a fully integrated rollout to 2031 and extends the upper tail to 'never', reflecting a substantial chance the strict in-house and single-bill requirements are permanently missed.
Ask a followup
Sign in to run · $20 free credit, no card · every claim cited