Product images are provided for reference and may not represent the exact model, configuration, or included components.

Overview

SKU: 4X67A99729
UPC: 889488758142
Condition: New
Write a Review

Lenovo 4X67A99729 NVD L4 24GB PCIE4 GPU CSP-C1

Lenovo 4X67A99729 PCIe Gen4 GPU AcceleratorOverviewThe Lenovo 4X67A99729 is a PCIe Gen 4 GPU accelerator carrying 24GB of on-board memory, designed fo…

$9,199.99
Ships same business day
In stock

Quantity:

Adding to cart… The item has been added
Compatibility guidance available for your deployment
Senior specialists for pre and post-sales support
Authorized sourcing and documentation support
Shipping and lead-time confirmation before install

Laura Bennett, IPSD Senior Specialist

Talk to Laura

200+ hrs training • U.S - based

Senior Specialist • 877-277-7147

Lenovo 4X67A99729 NVD L4 24GB PCIE4 GPU CSP-C1

$9,199.99

Overview

SKU: 4X67A99729
UPC: 889488758142
Condition: New

No Bots, Just Experts

Questions about this product? Free pre-sales support from a senior specialist — product questions, compatibility checks, BOM quotes, price confirmation — typically answered within one business day. Need camera placement or system design work? Engineering time is $175 per hour (qty 1 = 1 hour). Hardware buyers get up to one hour ($175) credited back on their order.

Description

Lenovo 4X67A99729 PCIe Gen4 GPU Accelerator

Overview

The Lenovo 4X67A99729 is a PCIe Gen 4 GPU accelerator carrying 24GB of on-board memory, designed for compute-intensive server workloads including AI inference, video analytics, and high-throughput data processing. Built in Mexico and catalogued under UNSPSC 43201401 (data processing hardware), this card ships as a factory-new, genuine Lenovo option intended for CSP and enterprise rack deployments where density and per-slot throughput matter.

The dual-interface design — PCIe for host connectivity and Ethernet for out-of-band or direct-attached network paths — gives architects flexibility in how the card integrates into existing Lenovo server infrastructure or broader data center compute platforms.

Key Features

  • 24GB GPU Memory: 24GB of on-board VRAM is enough headroom to run large inference models or multi-stream AI video analytics workloads without host memory bottlenecks — a meaningful threshold for production deployments that outgrow smaller 8GB or 16GB cards.
  • PCIe Gen 4 Interface: PCIe 4.0 host connectivity doubles the per-lane bandwidth of PCIe 3.0, reducing the transfer bottleneck between CPU and GPU in data-heavy pipelines. Verify your target server platform supports PCIe Gen 4 slots before ordering.
  • Ethernet Interface: The secondary Ethernet interface supports out-of-band management or direct network attach configurations, reducing the need for a separate management NIC in tight slot budgets.
  • 10 lb Card Weight: At 10 lb, this is a substantial card — confirm your server chassis has adequate slot reinforcement and that your rack's weight budget accommodates it, particularly in high-density multi-GPU configurations.
  • CSP-C1 Form Factor: The CSP-C1 designation targets cloud service provider and enterprise rack environments, meaning the card is validated for high-availability, always-on duty cycles rather than workstation or desktop use cases.
  • Mexico Origin, UNSPSC 43201401: Country of origin and UNSPSC coding are relevant for procurement teams managing TAA compliance or category spend reporting — confirm TAA requirements with your contracting officer, as Mexico origin does not automatically satisfy TAA.

Integration & Compatibility

The 4X67A99729 is a Lenovo option card, meaning it is validated and supported within specific Lenovo ThinkSystem and ThinkAgile server platforms. Before deploying, confirm the target server's PCIe Gen 4 slot availability, power delivery capacity for a high-TDP GPU, and chassis airflow design. Pairing with high-bandwidth network switches on the Ethernet interface side ensures the secondary interface is not a bottleneck in throughput-sensitive configurations. Consult Lenovo's server compatibility matrix for the specific system board and firmware revision required.

Frequently Asked Questions

Q: What server platforms is the Lenovo 4X67A99729 compatible with?

A: The 4X67A99729 is a Lenovo option card validated for specific ThinkSystem and ThinkAgile platforms. Consult the Lenovo compatibility matrix for your specific server model and firmware version before ordering.

Q: Does the 4X67A99729 satisfy TAA compliance requirements?

A: The card is manufactured in Mexico. Mexico origin does not automatically confer TAA compliance — confirm with your contracting officer and review applicable procurement regulations for your program.

Q: What is the PCIe slot requirement for the 4X67A99729?

A: The card uses a PCIe Gen 4 interface. Your host server must have an available PCIe Gen 4 slot with sufficient power delivery and physical clearance for a 10 lb, full-height card.

Q: Can the Ethernet interface on the 4X67A99729 be used for standard LAN connectivity?

A: The Ethernet interface supports out-of-band management or direct-attach configurations. Its specific use cases depend on the server platform and driver configuration — verify with Lenovo documentation for your deployment scenario.

Q: What workloads is the 4X67A99729 intended for?

A: Based on its 24GB GPU memory, PCIe Gen 4 interface, and CSP-C1 designation, the card is suited for AI inference, video analytics acceleration, and compute-dense data center workloads in enterprise or cloud service provider environments.

Marty Allison
Marty Allison

The 4X67A99729 stands out primarily because of its 24GB on-board GPU memory — that is the number I look at first when scoping multi-stream inference or analytics pipelines, and 24GB gives you room to run without paging to host RAM. The PCIe Gen 4 interface is the other spec worth flagging: if you drop this card into a Gen 3 slot, you lose half the available bandwidth, which matters when you are streaming high-frame-rate video data to the GPU continuously.

Technical Highlights:

  • 24GB GPU Memory: Sufficient headroom for large model inference without host RAM spill — a practical threshold for production multi-stream analytics deployments.
  • PCIe Gen 4 Interface: Doubles per-lane bandwidth versus Gen 3; host platform compatibility is a hard requirement, not an assumption to make at install time.
  • 10 lb Card Weight: Heavier than most single-width GPU options — chassis slot reinforcement and rack weight budgeting are real pre-install checks, not optional.

Deployment Considerations:

  • Validate the target Lenovo server model against the official option compatibility matrix before purchase — CSP-C1 designation means this card is validated for specific platforms, not universally compatible across the ThinkSystem line.
  • Mexico country of origin requires explicit TAA compliance verification for federal procurement; do not assume compliance without checking your contract vehicle requirements.

This card is the right fit for a CSP or enterprise data center operator running Lenovo ThinkSystem platforms who needs validated, high-memory GPU compute for sustained AI inference or multi-stream video analytics — not a card to drop into a workstation or mixed-vendor chassis without compatibility confirmation.

Specifications
Weight: 10.00 lb
Country Origin: MX
Interface: PCIe, Ethernet
Country Of Origin: MX
Unspsc Code: 43201401
Q&A
Reviews
Have Questions?

RELATED PRODUCTS

System Design, Deployment & Technical Support

Support services and planning resources for commercial surveillance, access control, and infrastructure deployments.

Fixed scope • Fixed price

System Design Assistance

  • Get help validating product compatibility
  • Coverage requirements
  • Storage planning and deployment architecture before you buy.
Request Design Help

Deployment & Configuration Support

  • Access fixed-scope support for rollout planning
  • User setup guidance
  • Migration and system standardization across single-site or multi-site deployments
View Support Services

Guides, Tools & Calculators

  • PoE requirements
  • Storage retention
  • Camera selection and deployment methodology
Open Technical Resources