DRACO ORv3 480 kW AI Factory Rack

Overview

The density class AI factories are being built at. Engineered to the ORv3 480 kW rack class with fanless direct-to-chip liquid cooling and near-total heat capture to water — the architecture that lets a single rack do what a row used to, and lets the building reuse the heat instead of fighting it.

Key highlights

• Up to 480 kW per rack {VERIFY final config} — the current top of the ORv3 density class.
• 21″ OCP Open Rack v3, ~48 OU, 48 V DC busbar with blind-mate power.
• Fanless direct-to-chip liquid — approaching 100% heat capture to the water loop.
• Removing server fans removes fan power outright — a real, measurable efficiency win.
• Heat reuse ready — warm-water return usable for district or process heat.
• Multi-rack pods with shared facility CDU and coordinated power smoothing for AI load transients.
• Designed, integrated and commissioned in India by RDP.

AI workload fit

• AI factory and supercluster builds at multi-megawatt scale.
• Sovereign AI national datacenter programmes.
• LLM training at frontier scale across many racks.
• High-density inference estates for large-scale services.
• HPC & AI convergence in new-build facilities.
• Neocloud capacity build-outs where kW per square metre decides the economics.

AI workload positioning

This is a new-build or major-retrofit product, not something that drops into an existing hall. At this density the rack, the cooling plant, the electrical infrastructure and the building are one system, and RDP scopes them together. Open Rack v3 (ORv3) is the Open Compute Project’s high-density rack standard: a 21-inch rack aperture instead of the legacy 19-inch EIA, a 48 V DC busbar running the full height of the rack, and blind-mate power connectors so servers slide in without individual power cords. Capacity is measured in OU (OpenU, 48 mm) rather than RU. The 48 V DC architecture raises usable power density substantially over legacy per-server AC distribution, which is what makes 100 kW-plus racks practical at all. The honest constraint: very few Indian facilities can take a 480 kW rack today, and the correct answer for many buyers is the 140 kW ORv3 rack instead. We will say so rather than sell you a rack your building cannot feed.

Industry use cases

• Neocloud: maximum revenue-generating compute per square metre of expensive floor.
• Public sector & sovereign: national AI-factory infrastructure built to open standards.
• Research & education: next-generation national supercomputing facilities.
• Telecom: large-scale AI capacity co-located with core network infrastructure.
• Enterprise & GCCs: a dedicated AI facility rather than space in a shared hall.

Performance & how to be sure

Near-total heat capture to liquid is the number that matters: it eliminates server fan power and removes the room-cooling ceiling that caps air-cooled halls at roughly 30–40 kW per rack. Rather than quote a tokens/sec figure that will not match your workload, RDP offers a “benchmark your model” session: bring the model, precision and context length you actually intend to run, and we will validate it on this exact configuration before you commit.

Series & upgrade path

Below: the DRACO ORv3 140 kW NVL72-Class AI Rack — the right answer for most buildings today. Pairs with RDP facility CDUs, the DRACO 250 kW Power & UPS Block, and Rack Integration & Deployment. Populate with rack-scale AI systems or 8-GPU SXM nodes.

On-prem vs cloud (TCO)

For sustained daily AI work this configuration removes per-GPU-hour billing, queueing for scarce instances, and egress charges on your own data — and keeps everything on-premises for DPDP and data-residency obligations. Cloud still wins for burst capacity and one-off very large training runs; on-prem wins on sustained utilisation, control and predictable INR capital cost. We model the crossover with your actual usage rather than assert it.

Software & day-one readiness

Full rack and plant telemetry: per-shelf DC load, busbar current, coolant flow, inlet/outlet delta-T, leak detection and heat-capture accounting, exported to DCIM. Supports the energy-storage and power-smoothing controls that synchronized GPU clusters now require to avoid stressing upstream supply.

Power, thermal & acoustics

Medium-voltage or 480 V 3-phase input feeding ORv3 rectifier shelves onto a high-current 48 V DC busbar {VERIFY final current rating per config}. Facility-scale CDU, primary and secondary loops, and coordinated power smoothing for training-step load transients. A full facility engineering study is mandatory — power, water, structure, heat rejection and reuse. Exact wattage, BTU and dB(A) figures come from RDP bench measurement rather than estimates — ask for the site-readiness sheet with your quote.

Deployment, warranty & support

Built to order by RDP Technologies. Supplied with GST invoice (HSN 8471), pan-India onsite support, and availability through GeM for public-sector procurement. Built to order — typically 16–24 weeks including facility engineering. Confirmed at quotation.

Why RDP

RDP Technologies is a Make-in-India OEM with 14+ years and 300,000+ units shipped, supplying AI infrastructure from desk-side systems to rack-scale AI factories — with predictable INR pricing, GST invoicing and pan-India onsite service.

Buy with confidence

Use Request a Quote to reach an RDP solution architect for sizing, a benchmark-your-model session, financing options and a delivery plan. No obligation.

Specifications