Lenovo 4XC7A95695 Thinksystem Broadcom 57608 2X200/1X400GBE QSFP112 OCP Ethern

Lenovo 4XC7A95695 Broadcom 57608 2x200GbE / 1x400GbE QSFP112 OCP Ethernet Adapter

Overview

The Lenovo 4XC7A95695 is a high-density OCP 3.0 Ethernet adapter built on the Broadcom 57608 ASIC, delivering 2x200GbE or 1x400GbE connectivity over QSFP112 — the port configuration that matters when spine-to-server bandwidth is the bottleneck in AI compute clusters, high-throughput NVR backbones, or dense virtualization hosts. As an OCP (Open Compute Project) form-factor card, it fits the tool-less, hot-swap mezzanine bay on OCP-compliant ThinkSystem platforms, keeping cabling tidy and swap time low in rack-dense environments. Manufactured in Mexico and sourced factory-new, the 4XC7A95695 is a genuine Lenovo option part validated for ThinkSystem server platforms.

Key Features

• Broadcom 57608 Silicon: The 57608 is a production-proven, data-center-class ASIC with RDMA (RoCEv2) capability — critical when you need low-latency, CPU-bypass fabric for clustered NVR storage, distributed AI inference, or iSCSI/NVMe-oF storage fabrics. Hardware offload means the host CPU handles application workloads rather than network stack overhead.
• 2x200GbE or 1x400GbE via QSFP112: QSFP112 connectors support 400G-SR4 (short reach over OM4), 400G-LR4, and compatible 2x200G breakout cables — giving you flexibility to connect to a 400GbE spine switch as a single fat pipe or break to two 200GbE leaf uplinks without changing the card. That breakout flexibility reduces the number of spine ports consumed per server.
• OCP 3.0 Form Factor: OCP mezzanine adapters install into a dedicated bay on OCP-capable ThinkSystem servers — they consume no PCIe slot, simplify cable routing, and are removed without tools. In high-density deployments (40+ nodes per rack), this form factor difference meaningfully reduces installation time and airflow obstructions compared to standard FHHL PCIe cards.
• QSFP112 Interface: QSFP112 is the 400G-generation module standard, backward compatible with QSFP28 in lower-speed configurations on capable switches. Verify your switch or director's QSFP112 transceiver compatibility before ordering — not all 400G switches that accept QSFP-DD also have QSFP112 ports.
• ThinkSystem Platform Integration: As a Lenovo option part (prefix 4XC7A), the 4XC7A95695 is designed to clear Lenovo's system compatibility validation, firmware co-development, and XClarity Administrator management integration — the difference between a generic NIC that works and one that shows up correctly in your server management console with health telemetry and firmware lifecycle management.
• Factory-New, Genuine Hardware: Sourced direct from distribution — no refurbished, no grey-market, no parallel imports. For servers running critical workloads, the provenance matters: valid manufacturer warranty coverage, clean firmware baseline, no tampered hardware.

Integration and Compatibility

The 4XC7A95695 is designed for OCP 3.0 mezzanine bays on supported Lenovo ThinkSystem servers — confirm your specific server model against Lenovo's ServerProven compatibility list before placing an order, as OCP NIC compatibility is validated per platform generation. On the fabric side, the QSFP112 ports connect to 400GbE-capable network switches and directors; breakout to 2x200G requires a compatible QSFP112-to-2xQSFP56 breakout cable, which is not included. For high-speed NIC selection guidance, match the adapter's bandwidth to both your switch fabric headroom and storage back-end throughput — a 400GbE uplink to a 100GbE storage array delivers no benefit. In AI and data center server environments using RoCEv2 fabrics, ensure your top-of-rack switch supports Priority Flow Control (PFC) and ECN — the Broadcom 57608 supports these but the network must be configured end-to-end.

Frequently Asked Questions

Q: What servers is the Lenovo 4XC7A95695 compatible with?

A: The 4XC7A95695 is an OCP 3.0 mezzanine adapter designed for Lenovo ThinkSystem servers with an OCP 3.0 mezzanine slot. Confirm your specific server model against Lenovo's ServerProven compatibility database before ordering — OCP NIC compatibility is validated per server generation and is not universal across all ThinkSystem platforms.

Q: Does the 4XC7A95695 support RDMA / RoCEv2?

A: The Broadcom 57608 ASIC, which powers the 4XC7A95695, is a data-center-class silicon known for RDMA over Converged Ethernet (RoCEv2) capability. However, confirm RoCEv2 support and required driver versions in Lenovo's official documentation for this specific option part, as feature enablement depends on firmware and OS driver combinations.

Q: What cables or transceivers does the QSFP112 interface require?

A: QSFP112 ports support 400GbE single-port operation or 2x200GbE breakout configurations depending on the cable and switch port type. Compatible options typically include 400G-SR4 (short reach over OM4 fiber), 400G-LR4, and QSFP112-to-2xQSFP56 breakout cables. Cables and transceivers are not included with the adapter — source Lenovo-validated optics for warranty and compatibility assurance.

Q: Is the 4XC7A95695 a standard PCIe card or an OCP adapter?

A: It is an OCP 3.0 mezzanine form-factor adapter, not a standard PCIe add-in card. It installs into the dedicated OCP mezzanine bay on a compatible ThinkSystem server and does not occupy a PCIe slot. It cannot be installed in a standard PCIe x16 slot.

Q: What is the country of origin for the 4XC7A95695?

A: The 4XC7A95695 is manufactured in Mexico (MX), per the product's distribution data.

Jerry Tildsen

The 4XC7A95695 sits at the intersection of bandwidth density and platform integration — it delivers 400GbE throughput in a QSFP112 OCP mezzanine form factor, which means it drops into a ThinkSystem OCP bay without touching your PCIe slot allocation. That matters when you're building a server with multiple GPU cards, NVMe controllers, and need the NIC to not compete for PCIe lanes or slot real estate.

Technical Highlights:

• QSFP112 at 400G: QSFP112 is the 400G-generation connector that supports both single-mode and multimode optics at full 400GbE, plus breakout to 2x200GbE — one card covers both your spine uplink and dual-leaf configurations without swapping hardware.
• OCP 3.0 Mezzanine: No PCIe slot consumed, tool-less swap, integrates with Lenovo XClarity Administrator for firmware and health management — the operational overhead difference in a 20-node cluster is measurable, especially at firmware update time.
• Broadcom 57608 ASIC: Production-proven in hyperscale environments with RoCEv2 offload capability — CPU stays on application workloads rather than packet processing when the fabric is correctly configured with PFC and ECN end-to-end.

Deployment Considerations:

• Verify server model compatibility against Lenovo ServerProven before ordering — OCP 3.0 NIC validation is platform-specific and not all ThinkSystem generations with an OCP bay support every OCP NIC option part.
• QSFP112 transceivers and breakout cables are not included and must be sourced separately — budget for Lenovo-validated optics to maintain warranty coverage and avoid firmware compatibility issues with third-party modules.

This adapter is the right fit for ThinkSystem-based AI inference nodes, high-throughput distributed storage clusters (NVMe-oF or iSCSI), or any deployment where 400GbE fabric bandwidth is required and PCIe slot count is already constrained by GPU or storage controller cards.