20 August 2026

Enterprise AI spending grows smarter, not just bigger

• Companies are moving past simple adoption metrics like token volume, after examples like Amazon and Uber found that maximizing AI usage without measuring actual business value wastes money.
• Organizations now route different tasks to different models based on cost and performance, using cheaper open-source models for basic work like summarization and expensive premium models for complex reasoning.
• According to Vercel's AI Gateway data from June, token volume and spending both grew 27-29%, but the average price per token stayed flat, showing strategic deployment rather than wasteful consumption.
• AI agents (systems that reason, call tools and iterate through multiple steps) now drive most workloads and consume far more tokens than simple chatbots, making rising token usage a sign of capability growth, not inefficiency.
• Companies increasingly need fallback systems that switch between multiple AI providers mid-task when one fails, since no single provider dominates all use cases as models and pricing shift rapidly.

Reported by TechRadar