Stablecoin issuer's AI arm pushes local, privacy-first intelligence as an alternative to centralized cloud services

Tether, the company behind the USDT stablecoin, is arguing that AI running directly on personal devices is ready to compete with the cloud-based systems that dominate the industry today. Through its technology arm, Tether Data, the company has been building a platform called QVAC aimed at making that shift.

The pitch: local over cloud

Tether argues that cloud AI forces users to rent intelligence from centralized, high-cost data centers. Its alternative is local, edge-first AI, where inference and training happen on everyday hardware, so that AI is something users own. Company CEO Paolo Ardoino has said AI should not be controlled only by large cloud platforms.

What Tether has built

  • QVAC Fabric LLM. A runtime and fine-tuning framework that lets users train and personalize large language models on consumer GPUs, laptops and smartphones, without high-end cloud servers or specialized NVIDIA systems.

  • QVAC SDK. An open-source, cross-platform toolkit for building, running and fine-tuning AI directly on devices, including local speech-to-text and translation engines. Tether says the AI keeps working if the internet goes down.

  • QVAC Workbench. An app for running many AI models locally, where Tether says chats stay on the device and remain private.

  • Applications and data. QVAC Health, which analyzes biometric data on-device, and QVAC MedPsy, medical language models that Tether says can run on phones. The company has also released a large synthetic dataset for training STEM-focused AI models.

Why it matters

Tether says on-device processing keeps sensitive data off external servers, lowers latency and reduces dependence on costly cloud GPUs. Its longer-term vision is a peer-to-peer "AI internet" where billions of users and AI agents interact without centralized control, with Bitcoin and USDT eventually letting agents transact on their own.

A note of caution

Many of the performance and privacy claims come from Tether itself and have not been independently verified. Cloud providers still lead in raw compute for the largest models, and Tether's push into AI is part of a broader diversification beyond its stablecoin business.