In a bold move that could democratize artificial intelligence, Tether, the issuer of the world’s largest stablecoin by market cap, USDT, has launched a groundbreaking AI training framework. The system, part of Tether’s QVAC platform, leverages Microsoft’s BitNet architecture and LoRA techniques to enable large language models to be fine-tuned on consumer hardware, including smartphones and non-Nvidia GPUs.
Breaking Down the Technology
According to Tether’s announcement, the framework significantly reduces memory and compute requirements, making AI development more accessible and cost-effective. It supports cross-platform training and inference across a variety of chips, including AMD, Intel, and Apple Silicon, as well as mobile GPUs from Qualcomm and Apple. This broad support is a significant step forward, as it extends the reach of AI development beyond the realm of high-end, specialized hardware.
Performance and Capabilities
Tether claims that its engineers have successfully fine-tuned models with up to 1 billion parameters on smartphones within two hours. Smaller models can be fine-tuned in mere minutes, and the framework supports models as large as 13 billion parameters on mobile devices. Built on BitNet, a 1-bit model architecture, the framework can cut VRAM requirements by up to 77.8% compared to similar 16-bit models. This reduction in VRAM allows larger models to run on devices with limited hardware capabilities.
Expanding the Ecosystem
The performance gains are not limited to training; the framework also accelerates inference, with mobile GPUs running BitNet models several times faster than CPUs. Tether’s framework supports on-device training and federated learning, enabling models to be updated across distributed devices without the need to send data to centralized servers. This approach can significantly reduce reliance on cloud infrastructure and enhance data privacy and security.
Industry Context and Implications
Tether’s move into AI infrastructure is part of a broader trend of crypto companies expanding into compute and machine learning. In September, Google took a 5.4% stake in Cipher Mining as part of a $3 billion, 10-year deal tied to AI data center capacity. Bitcoin miner IREN announced plans to raise about $3.6 billion to fund AI infrastructure in December. The trend continued into 2026, with HIVE Digital Technologies reporting record revenue of $93.1 million, driven by growth in its AI and high-performance computing (HPC) operations, and Core Scientific securing a $500 million loan facility from Morgan Stanley.
Looking Ahead
The intersection of cryptocurrency and AI is becoming increasingly significant, with applications ranging from mining infrastructure to autonomous agents. Tether’s new AI framework not only lowers the barriers to entry for AI development but also opens up new possibilities for decentralized and privacy-preserving AI solutions. As more crypto companies follow suit, the landscape of AI and blockchain technology is likely to become even more dynamic and innovative.
About the Author
