AI’s next phase is capital discipline, not innovation

Enterprise AI is moving from experimentation toward a sharper focus on business value and compute costs, TechRadar’s Manuel Haug writes. Chief financial officers are increasingly asking what AI has delivered, what value it created, and whether that value justifies its expense.

Haug argues that the next phase will favor organizations that generate the greatest business outcomes from efficient compute use, rather than those deploying the most agents or consuming the most tokens. Businesses often begin with a single AI agent for a specific process, then expand into areas such as finance, customer service, procurement, and supply chain operations. Benefits can grow quickly, but so can costs, because AI usage is measured in tokens consumed by prompts, decisions, workflows, and interactions.

The article says operational context is essential for measuring value. An agent rerouting a shipment or chasing a late payment may matter only when linked to a wider process. Context can also help agents make more targeted decisions with fewer prompts, retries, and human corrections.

Haug describes AI governance as increasingly important for tracking consumption and connecting initiatives to indicators such as customer satisfaction, efficiency, revenue growth, and delivery performance. He writes that reusable context can give future systems a stronger starting point. The next decade, he concludes, will favor repeatable practices that reduce token waste and support measurable outcomes across hundreds or even thousands of AI agents.