The Next Enterprise AI Bottleneck Isn't Compute—It's Context

Vivek Ahuja, VP of Technology at rSTAR, focused on enterprise AI, architecture, and measurable business outcomes.

Over the past year, I have been in a number of conversations about scaling enterprise AI. The discussion usually starts with models. Which model should we use? Which one has the best reasoning? Should we use multiple models?

Those are important questions. But increasingly, I don’t think they are the questions holding enterprise AI back.

I’ve seen teams build impressive AI solutions only to discover that the model doesn’t have access to the information it needs to make a useful decision. Customer information is in the CRM. Transactions are somewhere else. Policies live in documents. Operational data sits in other systems.

The model may be capable. It just doesn’t understand enough about what is happening inside the enterprise.

That is why I believe the next enterprise AI bottleneck isn’t compute. It’s context.

Intelligence without context has limited value.

Foundation models have become remarkably capable, but enterprise decisions rarely depend on general knowledge alone. They depend on knowing the customer, transaction, asset, policy, history, user’s permissions and what happened five minutes ago.

Think about a customer asking, “Why is my bill higher this month?” The AI may need billing history, the customer’s rate plan, recent usage, account changes and applicable policies across multiple systems. Without that context, even a capable model can provide an answer that sounds reasonable but is wrong for that customer.

This becomes even more important with AI agents. An assistant that generates an inaccurate answer creates a problem. An agent that acts on incomplete context can create a much bigger one.

Gartner’s 2026 research makes a similar point: “Agentic AI outcomes depend on context including semantic representations of data.” Without understanding the relationships and rules within an organization’s data, agents are more likely to produce inaccurate results.