Renen Hallak, co-founder and CEO of VAST Data, identifies the core blocker in enterprise AI: companies cannot hand sensitive data to external models, and AI labs cannot hand over model weights. The result is a stalemate that has kept serious AI deployment out of regulated industries.
VAST Data's answer is to bring proprietary models into customer-controlled environments, including air-gapped systems. This is not about cloud connectivity or API calls. It is about physically deploying capable models inside the security perimeter of banks, defense contractors, and healthcare systems that cannot tolerate data leaving their walls.
The conversation is worth reading in full for how Hallak frames this as a structural market opportunity, not a compliance checkbox. The mechanics of how model weights get deployed without exposure, and what VAST's infrastructure actually does to make that possible, are the details that will change how you think about enterprise AI timelines.
[WATCH ON YOUTUBE →]