CES 2026: We Evaluated a Compact AI Supercomputer, Tiiny AI


AI will be ubiquitous at CES 2026, frequently showing up in the form of wearables. Nevertheless, AI PCs engineered to operate Large Language Models (LLMs) are also capturing interest. A standout product making its debut at CES is the Tiiny AI Pocket Lab, a compact AI supercomputer that can conveniently fit in a pocket, although it’s not classified as a wearable.

The Tiiny AI supercomputer, referred to as the Tiiny AI Pocket Lab or just Tiiny, was presented to the media at the Pepcom event during CES. Mashable had the chance to test Tiiny’s functionalities and examine its small dimensions.

The Tiiny AI booth showcased a Guinness World Records certification, identifying the Tiiny AI Pocket Lab as the “world’s smallest mini PC.” At first look, Tiiny AI resembles a portable battery pack, featuring a rectangular, brick-like design. It easily fits in your hand and is extremely light in weight.

In spite of its compact form, Tiiny AI is equipped with 80GB of RAM and 1TB of SSD storage, allowing it to perform demanding AI computations. This high-performance PC executes AI functions directly on the device, guaranteeing privacy and removing the need for cloud uploads or subscriptions. Tiiny AI seeks to provide the most portable option.

Tiiny AI has additionally created a desktop application to facilitate the operation of AI models on the device, aimed at users who are not AI developers or programmers. The application accommodates various models, including generative AI for producing images and videos.

Evaluating the Tiiny AI Pocket Lab demonstrated its rapid response to inputs and quick content production, on par with well-known cloud AI services. Surprisingly, the device maintained a cool temperature even with extended usage, a significant accomplishment for a mini computer tackling demanding AI processing.

The Tiiny AI Pocket Lab is set to launch on Kickstarter in the upcoming months, priced at $1,399. While it may seem expensive for a mini-computer, it is considered relatively economical for AI computers.