AI Token Costs Spiral Out of Control as Enterprises Scramble for Solutions

Enterprise AI costs are spiraling out of control as token bills from frontier models like ChatGPT, Claude, and Gemini blow past budgets, forcing companies to adopt smarter model routing and hybrid compute strategies while vendors like Databricks race to offer cost-control solutions.

• Enterprise executives are sounding alarms over skyrocketing AI inference costs, with token bills from coding agents and frontier models like ChatGPT, Claude, and Gemini running far over budget and creating a financial crisis for organizations.
• Two key solutions are emerging to combat token costs: smarter model selection by routing simple tasks to smaller or open-source models instead of expensive frontier models, and hybrid compute strategies that run open models locally on company hardware to reduce per-token cloud expenses.
• Databricks is capitalizing on the crisis by launching Unity AI Gateway for token visibility and cost control, along with Omnigent and Genie One, positioning itself among a growing list of vendors racing to help enterprises rein in out-of-control AI agent spending.