AI inference performance depends on storage architecture more than most enterprise infrastructure teams have accounted for. Analyst Ryan Shrout and Avi Shetty, VP of AI Ecosystem and Market Enablement at Solidigm, breaks down why growing context windows force GPU recompute when storage is under-provisioned, how the three-tier inference storage architecture addresses that constraint, and why Jensen Huang's projection that the context memory tier will consume the entire TAM of storage signals the scale of infrastructure commitment enterprises need to be planning for now.

Ryan Shrout and Mitch Lewis of Signal65 walk through their independent validation of the Lenovo Knowledge Superagent with Sarah Lundgren, Director of Technical Enablement for Hybrid Cloud and AI at Lenovo. The conversation covers why agentic AI addresses knowledge management failures that wikis and enterprise search could not, how Lenovo deployed the platform internally to drive organic adoption, and what Signal65 measured: 30% reduction in retrieval time, 81% employee adoption, 120 hours saved per employee annually, and up to $17 million in potential productivity value at scale.

Ryan Shrout and Russ Fellows speak with Vincent Sheu of HPE about what edge AI infrastructure actually requires in production environments. The conversation focuses on the HPE ProLiant DL145 Gen11, predictable inference performance, and why thermal behavior, acoustics, and remote operations matter as much as raw compute.