Large Companies Turn to Open-Weight AI to Cut Costs

According to research conducted by Capgemini, nearly 60% of senior executives at large companies plan to increase their use of cheaper open-weight artificial intelligence models next year due to AI spending exceeding budgets.

Fifty-nine percent of executives participating in the survey stated that their AI spending last year exceeded expectations or budgets by an average of 24 percent. While 53 percent of executives expressed concern over AI costs, it was noted that this rate was 45 percent last year.

While open-weight models such as DeepSeek or Mistral can be downloaded and run anywhere, closed models like Claude, ChatGPT, or Gemini are only accessible via the cloud. IT consulting firm Capgemini reported that open-weight tokens are an average of 87 percent cheaper compared to closed models.

On the other hand, closed models are not being completely abandoned; half of the participants stated they would continue using closed models in cases where task complexity justifies the additional cost. Meanwhile, SAP's chief financial officer announced last month that they had replaced some expensive AI models with cheaper alternatives while maintaining performance.

While Capgemini conducted the research with 500 enterprise executives across 10 countries with annual revenues exceeding $1 billion, Anthropic, preparing for an IPO, had warned in its prospectus that customers might turn to cheaper options, including open-weight models.