INFINIBAND
Quantum-2 and Quantum-3
NDR 400 Gb/s Quantum-2 and XDR 800 Gb/s Quantum-3 InfiniBand switches — the standard fabric for production NVIDIA HGX and DGX training clusters.
01 / 04
Independent supply VENDOR · AI FABRIC & HIGH-PERFORMANCE NETWORKING
NVIDIA Networking — InfiniBand and Ethernet
Quantum InfiniBand, Spectrum Ethernet, ConnectX, BlueField. Sourced through established procurement channels. Delivered DDP across the Balkans.
What used to be Mellanox is now NVIDIA Networking — the InfiniBand and Ethernet fabric, ConnectX NICs and BlueField DPUs that interconnect production AI clusters around the world. MegaCore supplies the NVIDIA Networking portfolio into Serbia, Bosnia and Herzegovina, and North Macedonia — new product, original packaging, manufacturer serial numbers intact, full provenance documentation confirmed per shipment. The networking tier typically ships alongside the GPU compute tier covered on our NVIDIA AI page.
PORTFOLIO HIGHLIGHTS
ABOUT THE PORTFOLIO
AI cluster fabric under the NVIDIA Networking name
NVIDIA Networking (the former Mellanox) builds the fabric for most production AI clusters worldwide — Quantum InfiniBand switches at NDR/XDR speeds, Spectrum-X Ethernet for AI-optimized 800 GbE, ConnectX SmartNICs and BlueField DPUs. The portfolio interconnects GPU compute at scale.
MegaCore is an independent supplier. We source NVIDIA Networking hardware through Gamma Group's established procurement network and deliver DDP to Serbia, Bosnia and Herzegovina, and North Macedonia. Every shipment is new, sealed, in original packaging, with manufacturer serial numbers intact and full provenance documentation included as part of the standard contract. The networking tier is subject to the same export-control documentation workflow as the GPU compute tier — see the NVIDIA AI page for the full per-deployment documentation process.
WHAT WE SUPPLY
Three tiers, one supply path
Where NVIDIA Networking typically fits in a Balkan integrator build.
InfiniBand fabric
Quantum-2 NDR 400 Gb/s and Quantum-3 XDR 800 Gb/s InfiniBand switches — leaf and spine configurations for production AI training clusters.
- · Quantum-2 — NDR 400 Gb/s
- · Quantum-3 — XDR 800 Gb/s (on request)
- · MQM and MQS managed switches
Lead time: 10–14 weeks DDP
Ethernet fabric
Spectrum-4 800 GbE Ethernet switches and the Spectrum-X reference architecture — an alternative to InfiniBand for AI workloads where Ethernet operations are preferred.
- · Spectrum-4 — 800 GbE AI-optimized
- · Spectrum-X reference architecture
- · Lower-speed Spectrum variants available
Lead time: 8–12 weeks DDP
NICs and DPUs
ConnectX-7 and ConnectX-8 SmartNICs plus BlueField-3 DPUs — host-side fabric and offload for storage, security and multi-tenant AI cloud.
- · ConnectX-7 — 400 Gb/s IB / Ethernet
- · ConnectX-8 — 800 Gb/s IB / Ethernet
- · BlueField-3 — 400 Gb/s DPU
Lead time: 8–12 weeks DDP
WHY MEGACORE FOR NVIDIA NETWORKING
Independent supply, AI-cluster ready
Channel
NVIDIA Networking hardware sourced through Gamma Group's established procurement network — new product, original packaging, serial numbers intact, provenance documentation included as standard
Documentation
Provenance, serial numbers, warranty position confirmed per shipment; export-control documentation per deployment — see NVIDIA AI page
Terms
DDP · EUR or RSD · 30% prepayment standard
Channel respect
We supply you; we do not contact your end-client, propose alternative configurations, or follow up post-sale. The relationship runs through you.
MegaCore is an independent supplier. NVIDIA, Mellanox, Quantum, Spectrum, ConnectX, BlueField and related marks are trademarks of NVIDIA Corporation, used here for nominative descriptive purposes only. Configuration availability and service contract scope vary by SKU and region — we confirm the applicable details per shipment before contract. High-performance networking components for AI cluster deployments may be subject to US export controls — see the NVIDIA AI page for the applicable documentation workflow.
CONTACT
Tell us what you need
Send the configuration, the quantity, and the delivery country. We respond within 24 hours for NVIDIA Networking SKUs with availability path, indicative lead time and partner pricing. For full AI cluster fabric deployments expect a sizing call rather than an instant quote — that's how we keep our timelines honest. See the NVIDIA AI page for the full AI infrastructure supply story.