Tiiny AI launched a Kickstarter campaign on March 11, 2026, for the Tiiny Pocket Lab — a pocket-sized device the company is calling "the world's first pocket-size AI supercomputer." The premise is straightforward: run AI models and agentic workflows entirely offline, no cloud dependency, no subscription fees, no per-token charges. Early backers can lock in the device at $1,299, a price Tiiny AI says will be the lowest it ever sells for.
What the company hasn't published is what's actually inside it. The campaign offers no chip specifications, no memory configurations, no benchmark figures — the kind of detail that would let anyone independently assess whether "supercomputer" means anything beyond a marketing decision. The "world's first" claim is similarly unverifiable without hardware specifics. Crowdfunded devices have a well-documented history of ambitious specs that arrive revised, delayed, or not at all, and pocket-sized AI hardware is not an empty category: products from companies like AYANEO and various NPU-equipped mini PCs already occupy adjacent space.
The campaign covers eight markets at launch — the United States, Germany, the UK, France, Italy, Spain, the Netherlands, and Singapore — with estimated delivery in August 2026. There's also a parallel deposit system running on Tiiny AI's own website alongside the Kickstarter. Backers need matching email addresses across both platforms to hold the early-backer price, an unusual step that suggests a hybrid go-to-market approach but adds friction to what is already a speculative commitment.
The clearest concrete claim in the campaign is pricing structure: the FAQ confirms that all local AI features, including model downloads and agent usage, are bundled into the purchase price with no ongoing costs. For developers frustrated by cloud API economics, that framing has obvious appeal. Whether the hardware underneath can deliver inference that justifies it is a question Tiiny AI will need to answer with specs — and eventually with units — before August.