Databricks CEO Ali Ghodsi dismisses runaway AI risk, says cyber is the real impact zone
a16z · x · 2026-09-19
In an a16z interview, Databricks CEO Ali Ghodsi laid out his take on AI risk and adoption:
- Existential risk: He isn't losing sleep over runaway takeoff — several opposite conditions currently exist, and frontier training runs keep demanding more power, GPUs, and engineers, with some attempts burning huge money. Until that reverses, he doesn't see the loop happening.
- Cybersecurity: The space he anticipates real impact in — most orgs aren't equipped for agentic capabilities, and the time from vulnerability publication to weaponization has dropped from years to hours.
- Adoption: Most companies don't need a smarter model; the gap is context the models don't absorb — the unwritten knowledge inside employees' heads.
More from AGI Musings
- Half my replies are AI bots, so I'm moving to email and voice memos: Allie K Miller — alliekmiller · 2026-09-19
- Databricks CEO Ali Ghodsi: runaway AI risk is overblown, cyber is the real threat — a16z · 2026-09-19
- AI safety researcher publicly disputes MacAskill op-ed over 'double counting' Claude's expressed uncertainty — rgblong · 2026-09-19
- EA veteran: it's a useful mental toolkit, not a source of deep meaning — AndyMasley · 2026-09-19
- Agents are dissolving software moats: UI, workflow, and brand lose leverage — signulll · 2026-09-19
- Simon Willison: ignoring LLMs is like a geneticist snubbing Jurassic Park — Simon Willison · 2026-09-19