DRACO 8× GB200 NVL72 AI SuperCluster

Overview

The DRACO 8× GB200 NVL72 AI SuperCluster is a turnkey, liquid-cooled NVL72 SuperPOD delivering ~110 TB HBM3e of aggregate GPU memory across 8× NVL72 racks. It arrives as a complete, validated system — power, cooling, fabric, storage and software — ready to train frontier models on-premises, in INR, on a GST invoice.

Engineered for sovereign-AI programmes, neoclouds and large enterprises building data-hall-scale capacity, it is delivered as a single engagement: RDP designs the reference architecture, integrates and burns it in, and hands over one validated SuperPOD with one warranty and one support contract — removing multi-vendor integration risk.

Key highlights

• 8× GB200 NVL72 · ~110 TB HBM3e aggregate — data-hall-scale GPU memory for training and serving the largest models.
• Unified NVLink domains + non-blocking InfiniBand spine — each NVL72 rack is a unified NVLink domain of 72 Grace-Blackwell GPUs; the 8× racks are joined by a non-blocking spine-leaf InfiniBand fabric.
• 288× NVIDIA Grace (ARM) (coherent) + Grace LPDDR5X coherent memory — host/CPU compute matched to 576 Blackwell GPUs.
• 8 PB parallel NVMe parallel filesystem — high-throughput training data and checkpoint storage at cluster scale.
• 8-Rack SuperPOD, liquid-cooled, turnkey — delivered racked, cabled, cooled, validated; one SKU, one warranty.
• On-prem data sovereignty — training data and weights stay in-house; DPDP-friendly, air-gappable.
• Make-in-India OEM — predictable INR pricing, GST tax invoice (HSN 8471), pan-India onsite support, GeM-procurable.
• Scale path — extend the fabric to larger SuperPODs and multi-pod data halls.

AI workload fit (what it actually runs — honestly)

• Frontier training: distributed pre-training of large foundation models across the 576 GPUs with 3D parallelism.
• Large-scale fine-tuning & serving: fine-tune and serve many large models in parallel, or shard the very largest across the NVLink domains.
• RAG, multimodal & agentic platforms: organisation-wide production AI on the cluster’s storage and fabric.
• Engineering note: an NVL SuperPOD combines several unified NVLink domains (72 GPUs each) over a non-blocking InfiniBand spine — within a rack the GPUs act as one accelerator, across racks the spine carries collectives. This is the architecture frontier-model training actually uses; we validate real scaling for your workload, not a peak number.

AI workload positioning

This sits at the data-hall / foundation-model stage: a complete AI supercomputer. With ~110 TB HBM3e of GPU memory on a non-blocking fabric, it is sized to sustain frontier-model training and organisation-wide serving on-prem — the sovereign, owned alternative to renting a hyperscale cloud cluster.

Industry use cases

• Government & sovereign AI — a national foundation-model SuperPOD on GeM-procurable infrastructure.
• Neocloud / AI providers — a competitive GPU-cloud region built and validated end-to-end.
• BFSI & conglomerates — private large-model training under data-residency rules.
• Healthcare & life sciences — large-scale research and imaging programmes, data in-house.
• Research & national labs — an institutional AI supercomputer.
• Telecom — in-house foundation-model development at scale.

Performance — and how to be sure

We don’t publish inflated peak numbers. The honest picture: 576 Blackwell GPUs (~110 TB HBM3e) on a non-blocking fabric are sized for frontier-scale distributed training and organisation-wide serving. Want certainty? Request a free scaling benchmark of your model and dataset on a representative configuration before you commit; we’ll send back real tokens/sec, scaling efficiency and projected time-to-train.

Series & upgrade path

• DRACO (flagship supercluster tier) — this.
• Scale ladder: SXM pods (256–512 GPU) and GB200/GB300 NVL72 SuperPODs (hundreds–thousands of GPUs).
• When to step up: extend to larger GB300 SuperPODs or multi-pod data halls — talk to an architect about the fabric and facility.

On-prem vs cloud — the TCO case

At supercluster scale, owning is a strategic and economic decision: an always-on hyperscale cloud cluster dominates any AI budget over a multi-year horizon, while on-prem removes egress and keeps sovereign data and weights in-house. RDP pricing is fixed in INR with a GST input-credit-eligible invoice — ask for a multi-year TCO and financing model.

Software & day-one readiness

Ships pre-integrated and validated: NVIDIA driver, CUDA, cuDNN, NCCL, the InfiniBand stack, Slurm and/or Kubernetes, container registry, PyTorch and vLLM / Triton / TensorRT-LLM on Ubuntu LTS, with monitoring and a scheduler. Optional managed cluster operations and an MLOps platform.

Power, cooling & rack integration

A 8-rack superpod liquid-cooled SuperPOD with high power density — RDP scopes facility power, CDU/manifold and water, the InfiniBand spine, floor layout and redundancy in the reference-architecture design. (Exact power, BTU, flow and facility figures confirmed in the design package.) Full out-of-band management.

Deployment, warranty & support

• Made to order, integrated, racked, cabled, cooled and burned-in in India; project timeline confirmed at quote.
• Delivered as one system: NVL72 racks, leaf/spine switches, cabling, PDUs, cooling, parallel storage, and the full cluster software stack.
• Onsite warranty + AMC with pan-India coverage, cluster-level SLAs and an RMA/escalation path (exact terms confirmed at quote).

Why RDP

14 years of Make-in-India infrastructure and 300,000+ devices shipped. Indian OEM, INR pricing, GST tax invoice (HSN 8471), pan-India onsite engineers, GeM availability, and DPDP / sovereign-AI-ready deployment.

Buy with confidence

This is a turnkey AI SuperPOD, made to order — talk to an RDP solution architect, co-design the reference architecture, fabric and data hall, get a multi-year TCO and financing plan, and run a scaling benchmark before you commit. Request a quote to begin.

Specifications