Tether, the issuer of the world's largest stablecoin USDT, has launched a groundbreaking cross-platform AI training framework. The system, developed through its QVAC Fabric platform, utilizes Microsoft's BitNet architecture and LoRA (Low-Rank Adaptation) fine-tuning techniques to enable the training and inference of large language models (LLMs) with up to one billion parameters on consumer-grade hardware.
The framework is designed to eliminate the traditional dependency on expensive, enterprise-grade NVIDIA infrastructure or cloud services. It supports a wide range of chipsets, including AMD, Intel, and Apple Silicon CPUs and GPUs, as well as mobile GPUs from Qualcomm and Apple. According to benchmarks published by Tether, the BitNet-1B model consumes up to 77.8% less VRAM than comparable 16-bit models like Gemma-3-1B. This efficiency allows for significantly faster performance; inference on mobile GPUs is reported to be between two and eleven times faster than on a CPU.
"When model training depends on centralized infrastructure, innovation stagnates and the balance of the ecosystem becomes fragile," stated Paolo Ardoino, CEO of Tether. He emphasized the company's commitment to allocating resources to ensure artificial intelligence is accessible locally on any device.
The practical implications are substantial. Tether's engineers demonstrated that fine-tuning a 125-million-parameter model on a biomedical dataset of about 300 documents completes in just ten minutes on a Samsung S25 smartphone. A 1-billion-parameter model requires about one hour and eighteen minutes on the same device. The framework even allows scaling up to 13-billion-parameter models on devices like the iPhone 16.
This move by Tether is part of a broader trend of crypto and blockchain companies expanding into AI and high-performance computing (HPC). The news references related industry movements, including Google's investment in Cipher Mining for AI data center capacity, Bitcoin miner IREN's fundraising for AI infrastructure, and record revenues for HIVE Digital Technologies from its AI/HPC operations. The rise of autonomous AI agents in the crypto sector, with recent developments from Coinbase, Alchemy, and World (co-founded by OpenAI's Sam Altman), further contextualizes Tether's strategic pivot into decentralized AI infrastructure.
