Large companies plan to use more lower-cost open-weight AI models as rising bills push them to seek cheaper options while keeping closed models for difficult tasks.
Capgemini’s November 12 survey found that nearly six in 10 executives expect their organizations to increase their use of open-weight AI over the coming year. The cost pressure is already clear: 59% of respondents said their AI spending in the past year was significantly above expectations or budgets. On average, spending was 24% above expectations or budgets.
More than half of respondents, 53%, said AI costs worried them, up from 45% the previous year. Capgemini said open-weight models cost an average of 87% less per token than closed models. In September, open-weight models made up 72% of activity on AI provider OpenRouter but only 27% of spending, according to the study.
The survey covered 500 senior executives at organizations with annual revenue above $1 billion across 10 countries. The shift appears to be toward using a mix of models, not replacing closed models entirely. Half of respondents said they would keep closed models for complex tasks when their extra cost was justified.