Alon Horev

Cofounder, Vast Data

Search every verified Alon Horev interview, podcast appearance, and on-the-record quote — each transcript cross-checked by AI and human review to confirm speaker identity. Alon Horev, co-founder and chief technology officer of VAST Data, appeared in an interview at CoreWeave's Fully Connected 2026 event, where he discussed data movement as a key challenge in the agentic AI era. He described VAST Data's origins in 2016, noting that the company aimed to bring high-performance computing-grade storage to enterprise systems before the AI boom. Horev attributed the practicality of disaggregated architectures to declining flash memory prices and faster RDMA networking over commodity Ethernet, which he said enabled compute and storage separation at exabyte scale. In the interview, Horev also addressed the concept of memory in AI agents, distinguishing between long-term memory, which allows agents to reference past interactions, and short-term memory, which relates to active sessions. He suggested that such memory does not necessarily require RAM or SRAM, and he discussed the role of communal knowledge bases for enterprise agents. Additionally, he referenced KV cache in the context of multi-turn conversations and agent sessions, noting that these interactions can generate data useful for fine-tuning or training purpose-built models.