What are the hardware recommendations for use with the TBT5-AI?

Last Update: October 2nd, 2026
Article ID: 3641512

Host System Specifications

Best Match

A Windows 11 notebook or desktop computer with Thunderbolt 5 that includes a 100W or lower rated USB Type-C power supply. Systems without discrete NVIDIA graphics controllers are easier to get started.

Also Compatible

A Windows 11 notebook or desktop computer with Thunderbolt 4 or USB4. If the system includes a power supply rated over 100W we recommend connecting this to the computer while using the TBT5-AI enclosure.

A discrete NVIDIA graphics controller may conflict with the NVIDIA desktop graphics controller installed into the enclosure. This is generally resolved by installing the latest NVIDIA graphics drivers following our guide: How-To: Install the latest NVIDIA Graphics Drivers.

This includes systems with:

  • Intel 12th generation i-Series processors and newer
  • AMD Ryzen 6000 series processors and newer
  • USB4, Thunderbolt 4, and Thunderbolt 5 host ports

Graphics Card Compatibility

We expect all NVIDIA RTX (GeForce and Quadro) cards with up to 600W power requirements to function well in this enclosure. This includes the NVIDIA RTX 5080 (850W recommended) and RTX 5090 (1000W recommended) cards, the recommended wattage is most often full-system wattage, both models have a GPU only TDP lower than the 600W available in the enclosure.

AMD Radeon 6000 series and newer graphics cards should also be compatible with this enclosure, however we do not officially support these for machine learning/LLM workflows.

Physical Dimensions

  • Maximum GPU dimensions are 346mm x 170mm x 77mm
  • Maximum GPU TDP (power) is 600W. The enclosure's 850W ATX PSU includes two 6+2 pin and 1 x 12VHPWR power connectors.

If you have any questions please feel free to contact us at support@plugable.com and we'll be happy to help!