QCT backs NVIDIA Aerial RAN Computer portfolio with new AI-RAN servers

QCT said at NVIDIA GTC Berlin 2026 that it now fully supports NVIDIA’s Aerial RAN Computer portfolio with new MGX-based servers for telco AI-RAN deployments. The lineup is aimed at helping operators move from traditional RAN to software-defined networks across cell sites, central offices and mobile switching offices.

Why it matters: - QCT is targeting a fast-growing need for infrastructure that can run both telecom radio workloads and AI applications on the same platform. - The new portfolio is positioned to help operators shift from traditional RAN architectures to software-defined AI-RAN deployments. - The servers are designed to support both distributed and centralized network models, giving operators more flexibility across the edge and core.

What happened: - QCT announced full support for the NVIDIA Aerial RAN Computer portfolio at NVIDIA GTC Berlin 2026. - The announcement came as QCT and NVIDIA showcased next-generation NVIDIA ARC MGX servers in Berlin. - QCT said the platforms are built on the NVIDIA MGX reference design and are intended for AI-RAN requirements across central Mobile Switching Offices and distributed cell sites. - The portfolio is built around NVIDIA ARC MGX and NVIDIA RTX PRO 6000 Blackwell Server Edition and RTX PRO 4500 Blackwell GPUs.

The details: - The QCT portfolio is designed to support both AI and RAN workloads on a software-defined, GPU-accelerated architecture. - QCT said the servers provide carrier-grade connectivity and accelerated computing capabilities for communications service providers. - The lineup supports distributed RAN and centralized RAN deployment models. - QCT said the platforms create a transition path from cloud-native RAN to AI-native 5G and 6G networks. - The platforms are available for ecosystem partners and telecommunications operators to evaluate, validate and deploy AI-RAN use cases from the network core to the edge. - QuantaEdge EGN77C-2U, the NVIDIA ARC MGX 2U Short platform, is designed for space-constrained telco distributed deployments and is optimized for D-RAN cell-site environments. - QuantaEdge EGN77C-2U integrates one NVIDIA Grace CPU, three NVIDIA ConnectX-8 SuperNICs and up to two NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs. - QCT said the QuantaEdge system supports AI inference alongside RAN functions at the network edge. - QuantaGrid D66Q-2U, the NVIDIA ARC MGX 2U platform, is built for centralized AI-RAN environments in MSOs, Central Offices and regional points of presence. - QuantaGrid D66Q-2U includes two NVIDIA Vera CPUs, two NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, two NVIDIA ConnectX-8 SuperNICs and one NVIDIA BlueField DPU in an air-cooled design. - NVIDIA said Vera delivers 80% faster agentic CPU performance for longer-thinking agents, faster-learning RL systems and data-driven AI factories. - QuantaGrid D75E-4U, the NVIDIA ARC MGX 4U platform, is built for operators that need maximum AI acceleration and capacity. - QuantaGrid D75E-4U supports centralized deployments in MSOs, Central Offices and regional PoPs. - The D75E-4U system uses two x86 CPUs, four NVIDIA ConnectX-8 SuperNICs and one NVIDIA BlueField DPU. - The platform can be configured with up to eight NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs. - QCT said the D75E-4U is intended for demanding AI-RAN workloads, large-scale inference services and future intelligent network applications.

Between the lines: - The announcement extends QCT’s AI-RAN push from MWC 2026, when the company introduced the QuantaEdge EGN77C-2U with support for Nokia anyRAN and NVIDIA AI Aerial. - QCT is broadening its role from a single platform launch to a wider AI-RAN portfolio spanning distributed and centralized deployment needs. - The collaboration with Nokia and NVIDIA suggests QCT is aiming to stay inside a growing ecosystem of mobile network infrastructure vendors rather than build a standalone stack.

What's next: - QCT said attendees can see the platforms at the Nokia booth, #7056 in Hall 7, and the NVIDIA booth H949 at NVIDIA GTC Berlin 2026. - The company said the systems will be shown as part of an ongoing collaboration with Nokia and NVIDIA to advance AI-native mobile network infrastructure. - Operators and ecosystem partners can evaluate the platforms for future AI-RAN rollouts across the network core and edge.

The bottom line: - QCT is moving to give telecom operators a full menu of NVIDIA-based AI-RAN servers, from edge deployments to centralized facilities, as the industry races toward AI-native networks.