AI in the Pocket

Unleash your AI potential on-the-go with NVIDIA RTX GPU

Pocket AI offers the ultimate in flexibility and reliability on the move. This portable, plug and play AI accelerator delivers a perfect power/performance balance from the NVIDIA RTX GPU. Pocket AI is the perfect device for AI developers, professional graphics users and embedded industrial applications for boosting productivity by improve the work efficiency.

Product Description


·2048 NVIDIA CUDA® cores / 64 NVIDIA Tensor Cores / 16 NVIDIA RT Cores
·100 TOPS DENSE INT8 in inference
·6.54 TFLOPS Peak FP32 performance
·NVIDIA CUDA X, RTX Software Enhancements


Thunderbolt™ interface used on Pocket AI is an universal standard popularized on laptops, thin clients and compact PCs. With its advance to version 4.0 (backward compatible to 3.0) and the adoption of USB Type-C connections, the peripherals or devices built on the technology are proliferating. Pocket AI takes the advantages of Thunderbolt™ on its lighting transfer speed (up to 40Gb/s) and general availability in modern hosts, creating an intuitive plug-n-play U/X with hyper boost in productivity for AI developers, professional graphic users or combined usages with visual and data analytics in a future proof manner.

Application Field




Experience Unmatched Speed and Efficiency
NVIDIA RTX™ technology is one of NVIDIA’s most important advances in computer graphics, ushering in a new generation of applications that simulate the physical world at unprecedented speeds. Enhanced with the latest enhancements in AI, ray tracing, and simulation, RTX technology enables incredible 3D designs, photorealistic simulations, and stunning visual effects—faster than ever.


Ultimate portability: compact and lightweight

Pocket AI is designed with mobility in mind. Measuring around 106 x 72 x 25mm and weighing only 250 grams, it is truly portable and ready to go. Power supply is easy via USB PD.


GPU Architecture NVIDIA Ampere GA107
GPU clock Base clock: 435 MHz Boost clock: 1335 MHz
CUDA Cores 2,048
NVIDIA Tensor Cores 64
NVIDIA RT Cores 16
Single-Precision Floating Point  (TFLOPS) 6.54 TFLOPS
GPU Memory 4 GB GDDR6
Memory Clock 6001 MHz
Memory Interface Width 64-bit
Memory Bandwidth (GB/sec) 96 GB/s
TGP 25 W*
Interface Thunderbolt™ 3.0 (PCI Express 3.0 x 4)
Dimension 106mm x 72mm x25mm
Weight 250g
NVENC | NVDEC 1x | 2x
Operating System Windows 10 / Window 11 / Linux**
Operating Temperature 0~40°C
(*) Require Adapter/Charger or Power Bank with USB Power Delivery 3.0+ via Type-C connection on 15V and 45W+ supports.
(**) Hot Plug function is not supported under Linux OS

For more products and compatibility issues, please refer to the ADLINK official website.

Contact Us | Purchasing Information|