Enterprise AI Adoption Stumbles on Context Gaps and On‑Prem Security Demands

A new MIT study finds that 95 % of commercial‑real‑estate firms see little return from generic generative‑AI tools, citing a lack of contextual knowledge and fragmented "data debt" that prevents models from capturing the nuanced decisions made by experienced staff. The research highlights that successful AI in the sector requires a persistent, firm‑specific "company brain" embedded within workflows rather than large, off‑the‑shelf models.

Meanwhile, JPMorgan Chase has selected SambaNova Systems as its on‑premises AI inference partner, underscoring a growing split in enterprise AI infrastructure. The bank’s move signals that regulated institutions such as banks, hospitals and defense contractors are unwilling to expose sensitive data to hyperscaler clouds and are instead investing in secure, in‑house hardware solutions. Analysts anticipate that this on‑premise demand will create a distinct revenue stream separate from the dominant cloud‑first AI narrative.