Tether, the issuer of the world’s largest stablecoin by market capitalization, USDT, has released a modern AI training environment that it says enables the tuning of vast language models on consumer hardware, including smartphones and non-Nvidia GPUs.
According to Tuesday announcementThe system, part of the QVAC platform, leverages Microsoft’s BitNet architecture and LoRA techniques to reduce memory and computation requirements, potentially lowering costs and hardware barriers to developing AI models.
The platform supports cross-platform training and inference on a range of chips, including AMD, Intel and Apple Silicon, as well as mobile GPUs from Qualcomm and Apple.
Tether claims its engineers were able to tune models with up to 1 billion parameters on smartphones in under two hours, and smaller models in minutes, with support extending to models with up to 13 billion parameters on mobile devices.
Built on the 1-bit BitNet architecture, the platform can reduce VRAM requirements by up to 77.8% compared to similar 16-bit models, according to the company, enabling larger models to run on narrow hardware. It also enables LoRA tuning on non-Nvidia hardware for 1-bit models, extending support beyond GPUs typically used for AI training.
The company says the performance gains extend to inference, as mobile GPUs run BitNet models several times faster than CPUs. Use cases such as on-device training and federated learning are also highlighted, where models can be updated across distributed devices without sending data to centralized servers, potentially reducing dependency on cloud infrastructure.
Related: Messari’s modern CEO is increasing its utilize of artificial intelligence as the company reduces staff
Crypto companies are expanding into artificial intelligence, from mining infrastructure to autonomous agents
Tether’s move into artificial intelligence infrastructure comes as crypto companies expand into computation and machine learning, as bitcoin mining activity accelerates and the number of artificial intelligence agents increases.
In September, Google acquired a 5.4% stake in Cipher Mining as part of a 10-year, $3 billion deal related to AI data center performance. In December, bitcoin miner IREN said it planned to raise about $3.6 billion to fund artificial intelligence infrastructure.
This trend will continue in 2026. HIVE Digital Technologies reported record revenue of $93.1 million in February, driven by growth in its artificial intelligence and high-performance computing (HPC) business, while Core Scientific secured a $500 million loan from Morgan Stanley in March with an option to expand to $1 billion.
The mining sector has turned to AI and HPC as AI agents – autonomous programs that can conduct transactions, interact with services, and perform tasks – gain popularity across the crypto sector.
In October, Coinbase introduced a wallet infrastructure that allows AI agents to conduct onchain transactions. Last month, Alchemy launched a system that allows agents to access blockchain data services using USDC on Base. Also in February, Pantera and Franklin Templeton joined Arena, Sentient’s enterprise AI agent testing platform.
On Tuesday, World, the identity network co-founded by OpenAI’s Sam Altman, launched AgentKit, a toolkit that allows AI agents to verify that they are associated with a unique human using the World ID feature when making payments via the x402 micropayments protocol.
Warehouse: Fully 21 Million Bitcoins Are at Risk from Quantum Computers
