Tensorwave owns and operates advanced GPUs and data center infrastructure as a service for AI companies who need to rent time on GPUs to train and use their models. The company is built around AMD’s MI300x GPU. In addition to a better GPU, Tensorwave has proprietary software and interconnect technology that allows for massive nodes and clusters with hundreds or thousands of GPUs. These large clusters allow faster LLM training and enable models larger than the current generation of LLMs.