NVIDIA A16 64GB PCIe GPU (Quad-GPU)

NVIDIA A16 64GB PCIe GPU (Quad-GPU)

Brand: NVIDIA | Category: GPUs

SKU: 900-2G160-0000-000 | Part #: 900-2G160-0000-000 | MPN: 900-2G160-0000-000

Contact for Pricing — Request a Quote

Request a Quote Contact Us

About the NVIDIA A16 64GB PCIe GPU (Quad-GPU)

The NVIDIA A16 is a quad-GPU PCIe card purpose-built for enterprise virtual desktop infrastructure (VDI) and cloud graphics workloads. Each A16 card houses four independent NVIDIA Ampere GA107 GPUs on a single PCIe 4.0 x16 form factor, delivering a combined 64 GB of GDDR6 memory (16 GB per GPU) and 64 second-generation RT Cores and 256 third-generation Tensor Cores across the card. This design enables data center operators to maximize user density per rack unit, supporting up to 64 or more concurrent virtual desktop sessions from a single card depending on vGPU profile configuration.

Built on the NVIDIA Ampere architecture, the A16 leverages NVIDIA Virtual GPU (vGPU) software to partition each physical GPU into multiple virtual instances, each with dedicated frame buffer, graphics engines, and encode/decode engines. The card supports NVIDIA GRID vPC and vApps licensing profiles, enabling secure, isolated GPU resources for knowledge workers, designers, and enterprise application users across industries such as healthcare, financial services, education, and government. Hardware-based AV1 decode acceleration and NVENC encode engines on each GPU ensure smooth delivery of high-resolution virtual desktops and media-rich applications.

The A16 is optimized for passive cooling within standard data center server chassis and draws power through a single 8-pin PCIe connector per GPU block, with a total card TDP of 250 W. Its low-profile-compatible, full-height dual-slot form factor integrates into mainstream rack servers from leading brands, making it a practical choice for organizations scaling VDI deployments across UAE, GCC, EMEA, and APAC data center environments without requiring specialized infrastructure.

Ideal for

  • Enterprise virtual desktop infrastructure (VDI) supporting large populations of knowledge workers requiring GPU-accelerated remote desktops
  • Cloud workstation delivery for architects, engineers, and designers running CAD, BIM, and 3D visualization applications remotely
  • Secure remote access environments in regulated industries such as healthcare and financial services where GPU-isolated virtual sessions are required per compliance policy
  • Education and research institution deployments providing concurrent GPU-enabled virtual labs and student workstations at scale
  • Government and public sector thin-client rollouts demanding high user density, session isolation, and data residency within controlled data centers
  • Media and entertainment post-production VDI environments leveraging hardware AV1 decode and NVENC encode for high-fidelity remote content review workflows

Technical specifications

ManufacturerNVIDIA
Manufacturer Part Number900-2G160-0000-000
GPU ArchitectureNVIDIA Ampere (GA107 × 4)
Number of GPUs per Card4
Total Frame Buffer64 GB GDDR6 (16 GB per GPU)
Memory Interface per GPU128-bit
CUDA Cores (Total)2560 (640 per GPU)
RT Cores64 second-generation (16 per GPU)
Tensor Cores256 third-generation (64 per GPU)
Form FactorFull-height, dual-slot, PCIe
PCIe InterfacePCIe 4.0 x16
Display OutputsNone (headless, virtual display only via vGPU software)
Max Simultaneous Users (vGPU)Up to 64 per card (profile-dependent)
vGPU Software SupportNVIDIA GRID vPC, vApps
Hardware Video EncodeNVENC (1 per GPU, 4 total)
Hardware Video DecodeNVDEC with AV1 decode support (1 per GPU, 4 total)
Total Board TDP250 W
Power ConnectorPCIe 8-pin auxiliary
CoolingPassive (requires adequate server chassis airflow)
vGPU Profiles SupportedA16-1B, A16-2B, A16-1Q, A16-2Q, A16-4Q, A16-8Q, A16-16Q (and additional B/A/Q/C profiles per GRID release)
Operating System SupportVMware vSphere, Citrix Hypervisor, Red Hat Enterprise Virtualization, Microsoft Hyper-V (via NVIDIA vGPU software)
ECC Memory SupportYes

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.

Technical Specifications

BrandNVIDIA
CategoryGPUs
SKU900-2G160-0000-000
Part Number900-2G160-0000-000
ConditionNew
Manufacturer Part Number900-2G160-0000-000
GPU ArchitectureNVIDIA Ampere (GA107 × 4)
Number of GPUs per Card4
Total Frame Buffer64 GB GDDR6 (16 GB per GPU)
Memory Interface per GPU128-bit
CUDA Cores (Total)2560 (640 per GPU)
RT Cores64 second-generation (16 per GPU)
Tensor Cores256 third-generation (64 per GPU)
Form FactorFull-height, dual-slot, PCIe
PCIe InterfacePCIe 4.0 x16
Display OutputsNone (headless, virtual display only via vGPU software)
Max Simultaneous Users (vGPU)Up to 64 per card (profile-dependent)
vGPU Software SupportNVIDIA GRID vPC, vApps
Hardware Video EncodeNVENC (1 per GPU, 4 total)
Hardware Video DecodeNVDEC with AV1 decode support (1 per GPU, 4 total)
Total Board TDP250 W
Power ConnectorPCIe 8-pin auxiliary
CoolingPassive (requires adequate server chassis airflow)
vGPU Profiles SupportedA16-1B, A16-2B, A16-1Q, A16-2Q, A16-4Q, A16-8Q, A16-16Q (and additional B/A/Q/C profiles per GRID release)
Operating System SupportVMware vSphere, Citrix Hypervisor, Red Hat Enterprise Virtualization, Microsoft Hyper-V (via NVIDIA vGPU software)
ECC Memory SupportYes

Frequently Asked Questions about NVIDIA A16 64GB PCIe GPU (Quad-GPU)

What server platforms accept the NVIDIA A16 64GB PCIe GPU (Quad-GPU)?

Reference servers include Dell PowerEdge XE9680 / XE9712, HPE Cray XD670, Lenovo ThinkSystem SR685a / SR675 V3, Supermicro AS-A21GE / SYS-821GE, Gigabyte G593 / G894, ASUS ESC. Share your target platform in the RFQ and we will confirm chassis-to-GPU compatibility and recommended NIC pairing.

How long is the lead time on AI GPUs?

Highly model-dependent. L40S / RTX-class: typically 3-6 weeks. H100/H200/B200 in SXM form factor: 12-16 weeks for whole-platform allocations. We quote genuine-channel ETAs only — no grey-market promises.

Do you supply matched networking (Quantum InfiniBand / Spectrum-X)?

Yes — Omnixon stocks the full NVIDIA networking lineup (Quantum-2 / Quantum-X InfiniBand, Spectrum-X Ethernet, ConnectX NICs, BlueField DPUs) so we can quote a complete training-cluster BOM, not just the GPUs.

Can you help with NVIDIA AI Enterprise licensing?

Yes. We hold genuine-channels for NVIDIA AI Enterprise software subscriptions. Add it to your RFQ and we quote node-aligned licensing along with the hardware.