Back Issues This Week → Current Issue → Popular →

All issuesVolume 339, Issue 3IT NewsAI

How Companies Are Racing to Solve the AI Token Problem

Computerworld, Thursday, June 18th, 2026

Companies seek to cut expensive AI token consumption via cheaper models, hardware optimization, and prompt efficiency.

As generative AI adoption accelerates, token costs are rising sharply, forcing enterprises to find solutions.

Companies are exploring multiple cost-reduction strategies: switching to lower-cost models like Google's Gemini Flash, implementing caching layers to reduce redundant processing, and improving prompt efficiency.

Hardware approaches include deploying local AI solutions such as NVIDIA's RTX Spark and on-premise servers. Forward-deployed engineers are architecting systems with cost awareness, while analysts anticipate a shift from token-based pricing toward outcome-based models that measure success by results rather than computational units.

more →  ·  More from AI →