←21 Sep 2026 · Vinod Bijlani

An AI decision that arrives too late is not intelligence

Edge AI is the bridge to physical AI. Scaling it safely needs a distributed AI factory: local execution, central control and evidence flowing back.

Edge AIAI factoriesPhysical AI
Part of the series: My Thoughts on AI factory build-outs →

Edge AI is the bridge to physical AI, and investment in that bridge is accelerating. IDC estimates edge computing spending will grow from about US$265 billion in 2026 to nearly US$450 billion by 2029.

AI is moving beyond generating answers. It is starting to see, understand and act in the physical world: a virtual store manager prioritising stock, staffing and customer queues; a production line catching a defect before it becomes waste; a building adjusting its energy use in real time.

These decisions cannot always wait for the cloud. Intelligence has to move closer to where data is created and action is needed.

One example I shared is a virtual store assistant, currently in development. Store managers often switch between point-of-sale, inventory and rostering systems to make routine decisions. An edge AI agent can reason across that local context, stay available when connectivity is limited and keep routine operational data on site. Straightforward decisions are resolved locally. Complex or low-confidence questions are escalated to a larger model. Escalation is a designed control, not a failure.

Scaling this across hundreds or thousands of stores takes more than deploying local models. It requires distributed execution, central governance and clear boundaries around what an AI agent is allowed to do.

I explored this in my keynote on edge AI and real-time decision making at the Sydney Cloud & Datacenter Convention 2026. The best part was the discussion afterwards, and one question captured the real challenge: if decisions happen locally, how do we govern them consistently and reconstruct why an agent acted?

That moves the conversation from individual use cases to a bigger question: how do we scale physical AI safely across hundreds or thousands of locations? That is where the distributed AI factory comes in: local execution, central control and evidence flowing back.

Move intelligence closer to the event. Keep accountability close to the decision.

Invite Vinod to speak, write or advise.

Keynotes, panels and executive briefings on AI factories, sovereign AI and AI economics. Send a message on LinkedIn.