The company announces a server designed to transform edge and telecommunications infrastructure: it is, according to its internal analysis, the first single-server solution for Open RAN and Cloud RAN deployments, with up to 72 cores, 24 SFP28 (25G) ports, and precise synchronization functions for demanding networks. It will be available globally in the first quarter of 2026.
Dell Technologies has unveiled the PowerEdge XR8720t, a new compute sled from the XR8000 family designed to address the root causes of bottlenecks hindering Cloud RAN and Open RAN adoption. The proposal comes with a bold promise: consolidate what traditionally required multiple systems into a single server, resulting in reductions in costs, operational complexity, space, and energy consumption. Announced on September 30, 2025 from Round Rock (Texas), the XR8720t positions itself as a benchmark for telecommunications and enterprise edge environments.
The strategic perspective is clear. Until now, deployments of Cloud RAN — the paradigm that virtualizes network functions and runs them on general-purpose infrastructure — faced multi-server architectures that increased TCO, complicated maintenance, and required more space and power in often restricted locations. With the XR8720t, Dell introduces a turning point: a single server capable of absorbing workloads that previously required multiple nodes, while maintaining the latency and synchronization demanded by a next-generation mobile network.
Performance and connectivity in a compact format
The PowerEdge XR8720t comes with more than double the processing power compared to the previous generation — based on internal company data — supporting up to 72 cores in a 2U configuration. For traffic handling, the system raises port density with 24 SFP28 (25G) interfaces, enabling a combined bandwidth of up to 600 GbE. Practically, this high-density connectivity reduces additional switching, simplifies cabling, and brings capacity closer to the demand point—the network edge.
The depth of 430 mm is another notable feature. Designed for limited-space locations —such as towers and cell sites— it facilitates installations where every centimeter counts. Its ruggedized and modular design, with front access and easy maintenance and expansion, aims to minimize downtime and speed up field replacements. Additionally, the system operates within a -5 °C to 55 °C thermal range, covering demanding scenarios in telecom, defense, or industry.
Network timing for demanding workloads
One of the key features of the XR8720t is its focus on precise synchronization. The server integrates Intel® Xeon® 6 SoC with Intel vRAN Boost and Intel Ethernet E830-XXVDA8F technology, which includes a hardware “timing follower” to achieve timing accuracy across the network. Protocol-wise, the system supports PTP, PTM, and SyncE, standards that ensure synchronization among antennas, fronthaul, midhaul, and backhaul, maintaining quality of service in Cloud RAN.
This timing block isn’t just a technical detail: in disaggregated mobile networks, every microsecond counts. Poor synchronization can degrade handover, increase interference, or break coordination between cells. By anchoring the timing within the server itself —and avoiding reliance on external equipment— Dell aims for consistency and fewer failure points in complex topologies.
Ready for AI at the edge (and open to GPUs)
The XR8720t positions itself as an AI-ready system for CPU-based workloads, enabling AI agents, real-time analytics, and on-site machine learning. In environments like retail, defense, or manufacturing, this local inference capability reduces latency, data transport costs, and sensitive data exposure. Dell also notes that its flexible configuration supports GPU expansion upon request, providing a pathway for enhanced inference or lightweight training as needed.
NEBS Level 3 certification and durability design
The NEBS (Network Equipment-Building System) Level 3 certification places the XR8720t among equipment suited for extreme environments and telecom-grade continuity requirements. This standard ensures resilience against temperature, vibration, and harsh environmental conditions, which is critical for cell sites, outdoor enclosures, or industrial locations. The 430 mm depth supports installations in shallow racks and outdoor enclosures, expanding deployment options without specialized furniture.
The modular approach completes the design. Front access simplifies interventions, and its smart cooling system optimizes thermal performance in tight spaces where ventilation may be limited. Its mechanical and thermal durability makes it ideal not only for telecom but also for industrial edge and military applications, where mean time between failures (MTBF) and ease of service significantly impact TCO.
A “leap” forward for Open RAN and Cloud RAN
Dennis Hoffman, senior vice president at Dell Technologies, summarizes it as a “significant step” in edge and telecom infrastructure innovation. He emphasizes that, as the “first of its kind” platform, the XR8720t meets the requirements of Open RAN and Cloud RAN while delivering performance and scalability for edge workloads. The consolidation of multiple systems into one simplifies infrastructure and opens the door to more powerful, efficient, and AI-ready networks.
Beyond corporate messaging, the solution aims for concrete metrics: less footprint, less power consumption, less cabling, fewer failure points, and less operational hours invested in integration and support. Given the industry’s focus on capex/opex per cell, replacing “three or four” systems with one is an easy concept to grasp.
Cross-sector applications: from operators to enterprises
While the core market for the XR8720t is telecommunications, Dell also positions it for enterprise edge in retail, defense, and manufacturing. In a store, it could process video for loss prevention, optimize merchandising, or orchestrate AI agents to automate tasks. In factories, it enables visual inspection, predictive maintenance, and digital twins with ultra-low latency. In defense environments, it supports edge processing, sensor fusion, and synchronized communications in operational theaters.
In all cases, the combination of powerful CPU, dense 25G network, precise synchronization, and local AI capability brings decision layers closer to the field, with less reliance on remote clouds. This shift in focus — from cloud to edge — responds to a reality: every millisecond and every watt count when deploying intelligence near data.
Availability and roadmap
Dell projects the global availability of the PowerEdge XR8720t in the first quarter of 2026. Although the announcement does not specify detailed configurations — beyond up to 72 cores and the 24× SFP28 (25G) matrix — all signs point toward adaptable form factors suitable for different cell scenarios and synchronization requirements. Compatibility with the XR8000 platform ensures continuity for existing users within that modular ecosystem.
Market-wise, the XR8720t launch aligns with a maturity cycle for Open RAN / Cloud RAN: pilots have advanced to more aggressive deployments, and operators are seeking to standardize building blocks with fewer parts and more automation. The emergence of operational AI agents — capable of resolving issues, optimizing radio parameters, or predicting congestion — creates an urgent need for on-site computing capacity. The XR8720t aims to fill this niche.
What changes for operators?
In terms of TCO, the equation combines capex (fewer devices, switches, chassis) and opex (fewer onsite visits, reduced integration hours, lower power and cooling needs). In performance, the jump in compute power and port density enables handling peaks and supporting new services without forced upgrades. For resilience, the NEBS Level 3 design and precise synchronization provide operational margin in harsh environments. The potential GPU support offers an expansion valve for more intensive AI workloads.
A piece for the edge that anticipates more autonomous networks
The core message of the XR8720t lies in convergence: virtualized radio, edge computing, and operational AI within a single unit. With AI agents — called agentic AI — capable of reasoning, planning, and executing based on telemetry and network policies, operators aim for more autonomous and efficient networks. But making this a reality requires more than just software: it demands hardware that syncs, computes, and endures in the field. Dell positions its XR8720t to fill this role.
Frequently Asked Questions (targeted to sector-specific inquiries)
What practical difference is there between Open RAN and Cloud RAN, and how does the Dell PowerEdge XR8720t fit in?
Open RAN promotes hardware/software disaggregation with open interfaces, while Cloud RAN virtualizes radio functions to run on standard servers. The XR8720t is positioned as a single-server solution that consolidates typical Cloud RAN and Open RAN workloads, reducing number of nodes, cabling, and failure points, with 24× SFP28 (25G), PTP/PTM/SyncE synchronization, and up to 72 cores to host virtualized BBU/DU/CU.
What does the support for PTP, PTM, and SyncE with a “timing follower” hardware bring to Cloud RAN deployments?
It ensures that all disaggregated functions — from fronthaul to backhaul — share the same timing reference, which reduces interference, improves handover, and maintains quality in 5G/5G-Advanced. Hardware-accelerated via Intel Ethernet E830-XXVDA8F, the timing is more stable and less dependent on OS variations or load fluctuations.
Why is it important that the server meets NEBS Level 3 and has a depth of 430 mm?
NEBS Level 3 certifies the equipment’s ability to withstand extreme conditions such as temperature, vibration, and humidity, critical for cell sites, outdoor cabinets, or industrial locations. The 430 mm depth allows mounting in shallow racks and outdoor enclosures, broadening deployment options without requiring special furniture.
Is it prepared for AI use cases at the edge beyond telecommunications?
Yes. The XR8720t enables agentic AI, real-time analytics, and CPU-based machine learning for retail, defense, or manufacturing, with GPU support upon request for more intensive inference or light training tasks. Its front I/O, modular design, and -5 °C to 55 °C thermal range reduce intervention times and support continuous operation in edge environments.
Notes: Performance figures and “industry-first” claims based on Dell’s internal analysis (September 2025). Intel, the Intel logo, and other trademarks are Intel Corporation or its subsidiaries trademarks. Estimated global availability: Q1 2026.