Everpure expands AI infrastructure strategy with OneTouch acquisition
Everpure acquired OneTouch to add data intelligence capabilities, aiming to reduce unnecessary token usage in LLMs. The company targets enterprises, neoclouds, and hyperscalers with a new memory hierarchy.
Rob Lee, Chief Technology and Growth Officer at Everpure, outlined the company's dual strategy for AI infrastructure and data readiness. The infrastructure focus targets faster system performance, improved utilization, KV cache offloading, and support for large-scale training and enterprise inference. The data readiness focus aims to organize enterprise data for efficient recall, reducing unnecessary token usage in large language models. Everpure recently acquired OneTouch to add data intelligence capabilities, helping customers understand data meaning, origin, and correlations. Lee noted that AI creates the impetus to work with all enterprise data sources in unison, transcending silos like SAP and Salesforce. The company serves enterprises, neoclouds, and hyperscalers, addressing the need for a new memory hierarchy that includes GPU HBM, local DRAM, and remote storage to avoid recomputing context in long agent sessions. Everpure argues that AI requires moving compute closer to transactional data and that data intelligence is essential to prevent inefficient LLM processing.