When AT&T handed the keys to its 5G core network to Microsoft in 2021, it underscored that building next-generation wireless infrastructure is as much about selecting cloud platforms, virtualization strategies, and spectrum assets as it is about radios and towers. Those choices will shape which mobile operators succeed over the coming decade.
The three largest U.S. carriers have pursued markedly different technology and business strategies for 5G. AT&T has placed its bet on Microsoft’s cloud, Verizon has prioritized owning and virtualizing as much of its stack as possible, and T‑Mobile has leaned on the mid‑band spectrum it gained from Sprint to deliver fast consumer speeds.
Each approach produces distinct outcomes for network performance and economics. As 5G evolves toward 5G Advanced and, later, 6G, these strategies will be tested by real-world demand and technological change.
AT&T’s cloud-based 5G infrastructure: The Microsoft gamble
In June 2021 AT&T made a significant strategic move: it sold its in-house cloud technology to Microsoft and committed to running its 5G core on Microsoft Azure. The agreement transferred roughly 100 AT&T engineers and nearly a decade of carrier-grade cloud intellectual property to the hyperscaler.
The rationale was straightforward. AT&T had pursued a cloud‑native network approach for years but faced the reality that hyperscalers possess unmatched scale and capital to operate cloud platforms efficiently. By partnering with Microsoft, AT&T sought to accelerate deployment, reduce time to market, and take advantage of the public cloud’s automation and scale.
By October 2025 AT&T had launched a nationwide 5G standalone (SA) core on Azure. The carrier describes the result as an “open, virtualized architecture” composed of more than 60 containerized network functions running in a hybrid environment that spans on‑premises data centers and Microsoft’s public cloud.
The commercial benefits include faster service rollouts driven by software updates rather than hardware upgrades, automated scaling during traffic surges, and AI‑driven optimization across the network. AT&T executives emphasize a software‑first mindset and the ability to evolve the network via continuous software pushes.
However, the partnership also shifts critical control to Microsoft. Dependency on a hyperscaler introduces potential risks around integration issues, strategic alignment, and pricing disputes. AT&T exchanged direct control over its infrastructure for speed, scale and the economics of public cloud operations.
Verizon’s virtualization infrastructure: Owning the stack
Verizon chose a different path: rather than outsourcing core network functions to hyperscalers, it has invested in an end‑to‑end virtualized network it owns and operates. Verizon’s “Intelligent Edge Network” focuses on running network functions as software on commercial off‑the‑shelf hardware and integrating virtualization across RAN, core and edge.
By early 2025 Verizon had deployed tens of thousands of virtualized RAN (vRAN) sites and over 170,000 Open RAN‑capable radios. Those systems replace specialized hardware with software‑defined radio solutions that Verizon can configure and upgrade remotely.
Key milestones include containerizing core network functions, virtualizing baseband processing, and deploying a multi‑vendor RAN Intelligent Controller (RIC) to coordinate disparate vendors’ equipment using common interfaces. Projects combining Samsung’s AI energy manager and Qualcomm’s automation tools illustrate progress toward vendor‑agnostic, software‑based control.
Owning the full stack gives Verizon greater operational control, the ability to tune performance to specific needs, and insulation from hyperscaler pricing pressures. The tradeoff is the capital, technical expertise and ongoing investment required to build and maintain that capability internally.
T‑Mobile’s spectrum-driven 5G infrastructure: Efficiency over heavy investment
T‑Mobile has followed a strategy that emphasizes spectrum advantages and targeted deployment rather than matching rivals’ infrastructure spending. The carrier’s Sprint merger brought substantial mid‑band spectrum, which has translated into strong consumer throughput with lower relative CAPEX.
Independent testing in 2025 showed T‑Mobile delivering significantly higher average download speeds than its competitors. Those performance gains stem from effective use of mid‑band spectrum and an algorithmic approach to network expansion that prioritizes capacity where customers need it most.
T‑Mobile’s annual CAPEX has been materially lower than AT&T’s and Verizon’s, yet the company has pursued features such as network slicing within its 5G standalone architecture, enabling differentiated service experiences for enterprise, consumer and IoT use cases.
The primary risk is capacity: if data consumption grows faster than mid‑band spectrum can support, T‑Mobile will need to add infrastructure or broaden its spectrum mix. The carrier is exploring mmWave and other densification strategies, but scaling those approaches economically remains an open question.
How performance diverges
Independent network metrics illustrate how each strategy plays out in user experience. Across different performance categories no single operator dominates:
- Speed — T‑Mobile leads on average download speeds, reflecting its mid‑band advantage.
- Reliability — Verizon and T‑Mobile show strong marks for consistent and reliable service.
- Coverage — Verizon generally leads on overall coverage footprint.
- Consistency — T‑Mobile often scores highly for consistent quality experienced by subscribers.
AT&T’s cloud‑native core has not yet produced speed leadership, Verizon’s heavily virtualized approach emphasizes reliability over peak velocity, and T‑Mobile’s spectrum‑centric strategy delivers speed advantages while raising long‑term capacity questions.
What this means for 5G’s future
As networks evolve toward 5G Advanced, greater edge integration, pervasive AI management, and more sophisticated network slicing, the architectural choices made between 2021 and 2025 will be validated or rejected by real‑world performance and economics. Enterprises and consumers should recognize that today’s speed leader may not be the carrier with the deepest infrastructure investments; the most virtualized operator may not lead in raw throughput; and outsourcing to hyperscalers is a calculated tradeoff between control and scale.
Ultimately, time and continued user experience data will determine which combination of cloud partnerships, virtualization, and spectrum strategy best balances cost, performance and flexibility for the next generation of mobile networks.
(Photo by Boris Misevic)
Interested in how IoT is shaping telecom and connectivity? Industry events and conferences continue to highlight developments in 5G, edge computing and IoT that are influencing network strategies and services.