Skip to main content
M

Mellanox

The company provides an integrated interconnect solution comprising 200 GbE/400 GbE network adapters with RDMA and kernel bypass, low‑latency Ethernet switches with cut‑through forwarding, and a unified software stack for management and telemetry. The hardware offloads NVMe‑over‑Fabrics and RoCE v2 to reduce CPU overhead and latency, and it integrates with major cloud hypervisors and orchestration platforms for AI, HPC, and enterprise workloads.

Sunnyvale, United StatesFounded 199986050K+ followers
Updated 3 months ago

Funding

Funding not disclosed

SV
Funding rounds are not available yet.

Founders

Founder details are not available yet.

Product

Problem

Data centers and cloud platforms often experience network bottlenecks when supporting latency‑sensitive, high‑throughput workloads such as artificial‑intelligence training, high‑performance computing, and large‑scale enterprise applications. Existing interconnect hardware and software stacks can add unpredictable latency and limit overall storage and compute performance.

Solution

The company delivers an integrated interconnect portfolio that combines high‑speed network adapters, low‑latency Ethernet switches, and purpose‑built software to provide deterministic, high‑throughput connectivity. Its adapters support 200 GbE/400 GbE with RDMA and kernel bypass to reduce CPU overhead. Switches implement cut‑through forwarding and adaptive congestion control, ensuring minimal packet delay across fabric layers. The software stack offers unified management, telemetry, and firmware orchestration, enabling seamless provisioning and performance tuning. By offloading storage protocols such as NVMe‑over‑Fabrics, the solution accelerates end‑to‑end data movement for AI and HPC pipelines. The end‑to‑end hardware‑software integration is designed for compatibility with major cloud hypervisors and orchestration platforms, allowing operators to scale bandwidth without sacrificing latency.

Target Audience

Primary customers are data‑center operators, cloud service providers, and enterprise IT teams that run latency‑critical AI, HPC, and large‑scale enterprise workloads.

Features

  • 200 GbE and 400 GbE network adapters with RDMA, kernel bypass, and PCIe Gen5 interface
  • Low‑latency Ethernet switches featuring cut‑through forwarding, adaptive congestion control, and sub‑microsecond switching latency
  • Integrated offload engines for NVMe‑over‑Fabrics and RoCE v2 to accelerate storage traffic
  • Programmable firmware and open‑API hooks for custom traffic shaping and QoS policies
  • Unified management console providing real‑time telemetry, firmware updates, and automated topology discovery
  • Compatibility layer for leading cloud hypervisors (KVM, VMware, Hyper‑V) and orchestration tools (Kubernetes, OpenStack)
  • End‑to‑end encryption and role‑based access controls to meet enterprise security standards
  • Compliance with industry standards such as IEEE 802.3bs, RoCE, and NVMe‑OF specifications
This profile is AI-generated and may contain inaccuracies.