QCT Showcases Advanced AI Infrastructure at OCP Global Summit

SAN JOSE, CA, UNITED STATES, October 12, 2026 /CourierPR/ -- Quanta Cloud Technology (QCT) showcased its latest advancements in artificial intelligence (AI) infrastructure at the 2026 OCP Global Summit, highlighting a co-designed ecosystem that accelerates AI factory and distributed AI deployments.

QCT's offerings include the QuantaPlex S26F-2U, a server optimized for context memory storage, and the QuantaEdge EGN77C-2U, tailored for AI-RAN. These products are part of a broader suite of solutions that leverage NVIDIA's latest technologies, including the NVIDIA Vera Rubin POD, which is built on a modular, co-designed architecture.

At the heart of QCT’s multi-rack NVIDIA Vera Rubin POD is the NVIDIA Vera Rubin NVL72 platform. This platform features 18 QuantaGrid D76V-1U compute trays, combining 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs. The system delivers exceptional compute density and up to 20.7 TB of HBM4 memory, making it ideal for large-scale AI training and inference.

For CPU-intensive workloads, QCT offers the liquid-cooled QuantaPlex T46X-1U, which features dual-socket NVIDIA Vera CPUs and can scale up to 256 CPUs per rack. Its air-cooled counterpart, the QuantaGrid D66Q-2U, enhances throughput and responsiveness for AI workloads. Together, these servers form a robust AI factory infrastructure.

QCT also showcased a range of NVIDIA MGX architecture-based servers that operate alongside the NVIDIA Vera Rubin NVL72 platform. These servers, including the QuantaPlex T46X-1U, are designed to support CPU-intensive tasks and provide higher throughput and improved responsiveness.

In the realm of AI-native storage, the QuantaGrid S26F-2U, accelerated by NVIDIA BlueField-4 STX Storage Processor, enables KV-cache reuse to boost throughput, improve GPU utilization, and increase power efficiency for agentic AI applications. QCT collaborates with leading storage providers to deliver comprehensive, AI-native storage infrastructure built on NVIDIA CMX context memory storage platform, optimizing performance for AI inference, training, and analytics.

To further accelerate inference compute, QCT's upcoming NVIDIA Groq 3 LPX compute tray is designed to pair with the NVIDIA Vera Rubin NVL72, delivering exceptional performance for trillion-parameter models. This setup supports ultra-low-latency, real-time AI experiences.

Additionally, QCT introduced the QuantaEdge EGN77C-2U, a carrier-grade system based on the NVIDIA Aerial RAN Computer Pro (ARC-Pro) platform. This platform is designed to run 5G and future 6G software-defined radio access network (RAN) applications, enabling coexistence with AI workloads. The system integrates NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs and NVIDIA Grace CPUs, supporting inference-intensive applications and enabling telecom operators to deploy AI services and unlock new revenue opportunities at the network edge.

Leveraging the NVIDIA DSX ecosystem and SimReady assets, QCT constructs digital twins that simulate AI factories from the system to data center level. These digital twins support the AI factory journey from planning and construction to operations, validating scenarios prior to physical deployment to reduce deployment risks and accelerate time-to-market.

QCT's innovations at the 2026 OCP Global Summit underscore its commitment to advancing AI infrastructure and supporting the broader AI ecosystem.