Manufacturer profile
Hybrid Electronics sources NVIDIA GPUs, Jetson embedded modules, BlueField DPUs, and ConnectX network adapters — from current-production Blackwell and Hopper data-center accelerators to legacy Ampere, Volta, Turing, and Pascal parts still needed for qualified builds and long-lifecycle systems. Same-day quotes, traceable stock, and OEM-excess sourcing since 1995.
NVIDIA Corporation, founded in 1993 and headquartered in Santa Clara, California, is the world's largest designer of GPUs and AI accelerators. Once known primarily for gaming graphics, NVIDIA is now the dominant supplier of data-center compute for generative AI, high-performance computing, and machine learning training and inference. The company's product portfolio spans data-center accelerators (H100, H200, B200, GB200/GB300 rack-scale systems), professional visualization (RTX PRO, Quadro, and RTX A-series workstation cards), consumer graphics (GeForce RTX 50/40/30/20 series and legacy GTX), autonomous machines (Jetson Orin, Xavier, Nano, and TX modules), automotive AI (DRIVE Thor, Orin, Xavier), and networking (Mellanox-legacy ConnectX NICs, BlueField DPUs, Quantum InfiniBand switches).
Hybrid Electronics has been a channel for legacy and excess NVIDIA product since we began sourcing in 1995. We support customers who need obsolete Jetson TX1/TX2/Nano modules for fielded systems, discontinued Tesla-class datacenter GPUs (V100, P100, K80) still qualified into HPC clusters, Mellanox-branded ConnectX-4/5/6 NICs for legacy fabrics, and hard-to-find developer kits and modules when authorized distributors show zero stock or 40+ week lead times.
Our line card spans NVIDIA's full portfolio — from the newest Blackwell data-center GPUs shipping into hyperscaler fleets to legacy Pascal and Kepler parts keeping industrial and defense systems in service.
Blackwell (B200, B100, GB200, GB300, HGX B200), Hopper (H100 SXM & PCIe, H200 SXM & NVL, H100 NVL, H20), Ampere (A100 40GB & 80GB PCIe/SXM4, A40, A30, A16, A10, A2), and legacy Volta V100, Pascal P100 and P4/P40. For AI training, inference, HPC, and data-center virtualization.
GH200 Grace Hopper Superchip (72-core Grace Arm CPU + H100 GPU with NVLink-C2C), GB200 Grace Blackwell, and Grace-only C1 CPU Superchip. NVL72 and NVL4 rack-scale configurations for the largest AI training clusters.
RTX PRO Blackwell series (RTX PRO 6000, 5000, 4500, 4000 Blackwell), RTX Ada (RTX 6000 Ada, 5000 Ada, 4500 Ada, 4000 Ada), RTX A-series Ampere (A6000, A5500, A5000, A4500, A4000, A2000), and legacy Quadro RTX (RTX 8000, 6000, 5000, 4000). For CAD, DCC, simulation, and content creation.
GeForce RTX 50 series (RTX 5090, 5080, 5070 Ti, 5070), RTX 40 (4090, 4080 Super, 4070 Ti Super, 4060), RTX 30/20 (3090 Ti, 3080, 2080 Ti), plus obsolete GTX 10/9 series (GTX 1080 Ti, 1660 Super, 980) and Founder's Edition cards for legacy build maintenance.
Jetson Orin (AGX Orin 64GB & 32GB, Orin NX 16GB & 8GB, Orin Nano 8GB & 4GB, AGX Orin Industrial), Jetson Xavier (AGX Xavier, Xavier NX 8GB & 16GB, AGX Xavier Industrial), and legacy Jetson TX2/TX1/Nano modules and developer kits. For robotics, autonomous machines, medical imaging, and industrial AI vision.
BlueField-3 DPUs (B3220, B3240, B3140), BlueField-2 (MBF2M345A, MBF2H332A, MBF2M322A), ConnectX-7/6/5 smart NICs (200/400 GbE, HDR/NDR InfiniBand), and legacy Mellanox Quantum/Spectrum switches. Data-center offload, storage acceleration, and zero-trust networking.
DRIVE Thor, DRIVE Orin, DRIVE Xavier, and DRIVE developer platforms. For ADAS, autonomous driving, cockpit compute, and in-vehicle AI infrastructure.
Tegra X1 / X2 / TX2 NX, plus obsolete Tegra 3/4/K1 ARM SoCs still in fielded industrial and set-top applications. Nintendo Switch's Tegra X1 remains a legacy sourcing driver.
Hybrid Electronics quotes and sources the following NVIDIA parts across data-center, embedded, and networking segments. NVIDIA uses 900-XXXXX-XXXX-XXX
SKU numbers for modules and OEM boards and 945-XXXXX-XXXX-XXX for developer kits — contact us for the exact PN, region code, and revision you need.
| Part number | Family | Description | Quote |
|---|---|---|---|
| 900-21001-0040-100 | A100 SXM4 | A100 80GB HBM2e SXM4 datacenter GPU | Quote |
| 900-21010-0000-000 | A100 PCIe | A100 40GB HBM2 PCIe datacenter GPU | Quote |
| 900-2G133-0000-100 | A40 | A40 48GB GDDR6 pro-viz & VDI GPU | Quote |
| 900-2G133-0020-000 | A10 | A10 24GB GDDR6 mainstream datacenter GPU | Quote |
| 900-2G171-0000-100 | A16 | A16 64GB (4× 16GB) VDI-optimized GPU | Quote |
| 900-21010-0020-000 | H100 PCIe | H100 80GB HBM3 PCIe (Hopper) | Quote |
| 900-21010-0040-000 | H100 NVL | H100 NVL 94GB HBM3 dual-slot PCIe LLM inference | Quote |
| 900-13701-0050-000 | Jetson AGX Orin | Jetson AGX Orin 64GB production module | Quote |
| 900-13701-0040-000 | Jetson AGX Orin | Jetson AGX Orin 32GB production module | Quote |
| 900-13701-0080-000 | Jetson AGX Orin Ind. | AGX Orin Industrial (extended temp, US origin) | Quote |
| 900-13767-0000-001 | Jetson Orin NX | Orin NX 16GB production module | Quote |
| 900-13767-0010-001 | Jetson Orin NX | Orin NX 8GB production module | Quote |
| 900-13767-0030-000 | Jetson Orin Nano | Orin Nano 8GB production module | Quote |
| 900-13767-0040-000 | Jetson Orin Nano | Orin Nano 4GB production module | Quote |
| 945-13766-0000-000 | Orin Nano Dev Kit | Jetson Orin Nano Super Developer Kit | Quote |
| 945-13730-0050-000 | AGX Orin Dev Kit | Jetson AGX Orin 64GB Developer Kit | Quote |
| 900-13448-0020-000 | Jetson Nano | Jetson Nano production module (16GB eMMC) — legacy | Quote |
| 945-83518-0005-000 | Xavier NX Dev Kit | Jetson Xavier NX Developer Kit — legacy | Quote |
| 945-82972-0006-000 | AGX Xavier Dev Kit | Jetson AGX Xavier Developer Kit — legacy | Quote |
| 900-82888-0040-000 | AGX Xavier | Jetson AGX Xavier 32GB production module | Quote |
| MBF2M345A-HECOT | BlueField-2 DPU | 200GbE / NDR200 IB E-Series DPU, crypto, secure boot | Quote |
| MBF2H332A-AECOT | BlueField-2 DPU | 25GbE P-Series dual-port SmartNIC DPU | Quote |
| 900-9D3B6-00CN-AB0 | BlueField-3 DPU | B3240 400GbE / NDR IB dual-port DPU | Quote |
| MBF1M332A-ASCAT | BlueField-1 DPU | Legacy 25GbE dual-port SmartNIC DPU | Quote |
Data-center Blackwell (B100, B200, GB200, H200 NVL, HGX B200), professional RTX PRO Blackwell, GeForce RTX 50/40 series, ConnectX-7/8 adapters, and full Quantum/Spectrum switch line cards also quoted on request. Regional SKU variants (US, EU, JP, CN) available — ITAR- and EAR-controlled parts subject to export screening.
NVIDIA moves fast, and legacy parts leave authorized channels within 1–2 years of a new generation shipping. Systems in defense, medical, industrial automation, broadcast, and long-lifecycle infrastructure often need continued access to obsolete GPUs and modules. Hybrid Electronics sources these end-of-life NVIDIA products from OEM excess, contract-manufacturer surplus, and traceable secondary channels:
Send an RFQ with quantity, target date code range, and any traceability requirements (COO, MSL, factory sealed) — we return same-day availability from stock or an ETA from our sourcing network.
Data-center accelerators (A100, H100, H200), Jetson Orin production modules for embedded builds, and Mellanox-legacy ConnectX / BlueField NICs are our highest-volume NVIDIA quote categories. Legacy V100, P100, and obsolete Jetson TX2/Nano modules are steady low-volume sourcing lines for defense, medical, and industrial customers.
Yes — on an inquiry basis. Blackwell allocation is tight and priced dynamically, so we quote per-unit on demand against real-time supply from OEM excess and channel partners rather than holding stock. Include your target quantity, region, and required SXM/PCIe form factor, and we will return an availability window and quoted price.
Yes when available — most of our OEM-excess and CEM-surplus NVIDIA stock is factory sealed with original date codes and full traceability documentation. For downgraded stock we disclose condition (open box, tested-pull, refurbished) on the quote. Customers requiring MSL data, COO, or DPAS-rated delivery should note that on the RFQ.
Advanced NVIDIA datacenter accelerators (H100, H200, B200, and derivatives) are subject to U.S. export controls including ECCN 4A090 and country-specific restrictions. Hybrid Electronics screens every export order for end-use, end-user, and destination compliance under EAR §744. We do not sell restricted parts into embargoed destinations or to denied parties. AES/ITN filing is included on qualified export shipments.
Same day for parts in stock — typically within 2 hours during business hours (9am–6pm ET). Sourced quotes on out-of-stock parts return within 24–48 hours with an availability window. Email sales@hybridelectronics.com or call (407) 542-3995 to open an RFQ.
No — Hybrid Electronics is an independent distributor and excess-inventory buyer, not an NVIDIA-authorized distributor. We source NVIDIA parts from OEM excess, CEM surplus, and traceable secondary channels. For factory-warranted current-production orders, work with an NVIDIA Elite or Preferred Partner. For obsolete, hard-to-find, or allocated parts where authorized channels show no stock, Hybrid Electronics is a proven independent source since 1995.
NEED NVIDIA?
Send us your NVIDIA RFQ — part number, quantity, target date code — and we'll come back same day with availability from stock or an ETA from our sourcing network.
Or call (407) 542-3995
Hybrid Electronics is an independent distributor — not an authorized NVIDIA distributor. All trademarks belong to NVIDIA Corporation.