US GPU Cards
Professional Graphics Cards and GPU Accelerators
A professional graphics card adds the compute an edge AI computer, workstation or GPU server needs for inference, vision and visualization, with ECC memory and long availability that consumer cards rarely offer. The range covers passive data centre GPUs, actively cooled workstation GPUs and MXM GPU modules for compact embedded systems.
Options run from low-profile 50 W cards that draw all their power from the slot to 600 W cards with 96 GB of memory, suited to machine vision, medical imaging, video analytics and AI servers where the host's power, cooling and slot space set the limits.
Speak to a GPU specialist to match memory, power and form factor to your host system.
Browse Our GPU Cards
- GPU / AI Accelerator: Discrete GPU (MXM)
- GPU: NVIDIA RTX A2000 Embedded (MXM)
- OS: Linux, Windows 11
- Display Outputs: 4x Display Ports
- Temp: 0°C to 55°C [32°F to 131°F]
- GPU / AI Accelerator: Discrete GPU (MXM)
- GPU: RTX A4500 Embedded (MXM)
- OS: Linux, Windows 11
- Display Outputs: 4x Display Ports
- Temp: 0°C to 55°C [32°F to 131°F]
- GPU / AI Accelerator: Discrete GPU (MXM)
- GPU: NVIDIA RTX 2000 Ada Embedded (MXM)
- AI Performance: 232 TOPS
- OS: Linux, Windows 11
- Display Outputs: 3x Display Ports
- Temp: 0°C to 55°C [32°F to 131°F]
- GPU / AI Accelerator: Discrete GPU (MXM)
- GPU: RTX A500 Embedded (MXM)
- OS: Linux, Windows 11
- Temp: 0°C to 55°C [32°F to 131°F]
- GPU / AI Accelerator: Discrete GPU (MXM)
- GPU: RTX A1000 Embedded (MXM)
- OS: Linux, Windows 11
- Display Outputs: 4x Display Ports
- Temp: 0°C to 55°C [32°F to 131°F]
- GPU / AI Accelerator: Discrete GPU (MXM)
- GPU: NVIDIA RTX 5000 Ada Embedded (MXM)
- AI Performance: 682 TOPS
- OS: Linux, Windows 11
- Display Outputs: 4x Display Ports
- Temp: 0°C to 55°C [32°F to 131°F]
- GPU / AI Accelerator: Discrete GPU (MXM)
- GPU: NVIDIA RTX 3500 Ada Embedded (MXM)
- AI Performance: 369 TOPS
- OS: Linux, Windows 11
- Display Outputs: 4x Display Ports
- Temp: 0°C to 55°C [32°F to 131°F]
- GPU / AI Accelerator: Discrete GPU (MXM)
- GPU: RTX PRO 2000 Blackwell Embedded (MXM)
- AI Performance: 572
- OS: Linux, Windows 11
- Display Outputs: 3x Display Ports
- Temp: 0°C to 55°C [32°F to 131°F]
- GPU / AI Accelerator: Discrete GPU (PCIe)
- GPU: NVIDIA RTX A6000
- AI Performance: 310 TOPS
- Display Outputs: 4x Display Ports
- GPU / AI Accelerator: Discrete GPU (PCIe)
- GPU: RTX A5000
- AI Performance: 222
- Display Outputs: 4x Display Ports
- GPU / AI Accelerator: Discrete GPU (MXM)
- GPU: RTX PRO 500 Blackwell Embedded (MXM)
- AI Performance: 294
- OS: Linux, Windows 11
- Display Outputs: 3x Display Ports
- Temp: 0°C to 55°C [32°F to 131°F]
- GPU / AI Accelerator: Discrete GPU (PCIe)
- GPU: RTX A2000 12GB
- AI Performance: 64
- Display Outputs: 4x Mini Display Port
- GPU / AI Accelerator: Discrete GPU (PCIe)
- GPU: RTX A4000
- AI Performance: 153
- Display Outputs: 4x Display Ports
- GPU / AI Accelerator: Discrete GPU (PCIe)
- GPU: RTX A4000
- AI Performance: 153
- Display Outputs: 4x Display Ports
- GPU / AI Accelerator: Discrete GPU (PCIe)
- GPU: RTX A1000
- AI Performance: 54
- Display Outputs: 4x Mini Display Port
GPU Ranges Configured by Things Embedded
We work with the leading manufacturers of professional graphics cards, so the ranges below are the ones we specify for edge AI computers, workstations and GPU servers. Compare them by cooling, power and form factor, then browse the cards above or ask a GPU specialist.
Data Centre GPUs
Passive PCIe accelerators from the L4 to the H200 NVL, cooled by server airflow, for inference, training and virtual workstations.
Workstation Graphics Cards
Blower and flow-through cards across the Blackwell, Ada and Ampere generations, from low-profile 50 W cards to 96 GB 600 W cards.
MXM GPU Modules
MXM 3.1 Type A and Type B+ modules for embedded boards, with display or headless builds and optional coolers.
Not sure which range fits? Tell us the application and we will recommend the right model.
What to Specify in a GPU Card
Our GPU specialists specify every card against the host system, its power and airflow and the software it will run. These are the choices we work through with you before recommending a GPU.
GPU Memory
Model size and batch size set the memory you need, from 4 GB for display duties to 141 GB for large language models.
Active or Passive Cooling
Blower and flow-through cards cool themselves; passive data centre cards need the strong front-to-back airflow of a server.
Power Budget
Slot-powered cards stay at 75 W or below; larger cards need a 16-pin connector and a power supply sized for them.
Slot and Form Factor
Low profile, single slot, dual slot or MXM decides which chassis a GPU will physically fit.
Display Outputs
Workstation cards drive up to four DisplayPort screens; headless compute cards leave outputs off to save power.
Availability
Embedded and professional GPUs carry long availability, so a design qualified today stays buildable for years.
GPU Card Innovation & Edge AI Insights
Speak to a GPU Specialist.
Tell us the workload, the host system and the power and cooling it can provide. We will recommend the right GPU, then fit, configure and burn-in test it in the system before it ships.