Large Firms Shift to Open-Weight AI Models to Control Costs

As artificial intelligence (AI) continues to evolve, large firms are increasingly turning to open-weight AI models to manage rising costs. According to a recent study by Capgemini, nearly 6 in 10 senior executives at large companies plan to increase their use of these more affordable AI models over the next year. This shift comes as AI expenditures have significantly exceeded budgets, with executives reporting an average overspend of 24% over the past year.

The Capgemini survey, which included 500 executives from enterprises with annual revenues exceeding $1 billion across 10 countries, highlights a growing trend among companies to seek cost-effective alternatives in their AI strategies. The financial pressures are evident, with 53% of executives expressing concerns about AI costs, up from 45% the previous year. This indicates a rising urgency to find solutions that can help mitigate financial strain while still leveraging the power of AI.

Open-weight AI models, such as those developed by DeepSeek and Mistral, are gaining traction due to their accessibility and cost-effectiveness. Unlike closed models like Claude, ChatGPT, or Gemini, which require cloud access, open-weight models can be downloaded and run locally, providing companies with greater flexibility and control over their AI applications.

The Capgemini study reveals that open-weight models accounted for 72% of traffic on OpenRouter, a popular multi-AI provider, in September. However, they only represented 27% of total AI spending, indicating that while these models are widely used, they are not yet the predominant choice in terms of financial investment. This discrepancy suggests that many firms are still weighing the benefits of open-weight models against the performance and capabilities of closed options.

One of the most compelling reasons for the shift to open-weight models is the significant cost difference. Capgemini reports that open-weight tokens are, on average, 87% cheaper than their closed counterparts. This substantial price gap is prompting companies to reconsider their AI expenditures, especially as budgets tighten and the demand for AI solutions continues to grow.

Despite the cost advantages, closed models are not being completely phased out. Half of the executives surveyed indicated that they would retain closed models for tasks where the complexity justified the additional expense. This suggests that while cost is a critical factor, performance and task requirements still play a significant role in AI model selection.

The trend towards open-weight AI models is not isolated to the findings of the Capgemini study. Industry leaders are also making similar moves. For instance, SAP's finance chief recently announced plans to replace some of their most expensive AI models with more affordable options, provided that performance remains satisfactory. This reflects a broader industry recognition of the need to balance AI capabilities with budgetary constraints.

Additionally, Anthropic, a company that recently launched its IPO roadshow, has warned in its prospectus that customers may increasingly shift towards cheaper AI options, including open-weight models. This sentiment underscores the growing consensus among industry players that cost control is becoming a paramount concern in AI deployment strategies.

As the landscape of AI continues to evolve, the implications of this shift towards open-weight models could be profound. Companies that successfully integrate these cost-effective solutions may find themselves at a competitive advantage, particularly in industries where margins are tight and operational efficiency is crucial.

The Capgemini study serves as a wake-up call for organizations to reassess their AI strategies in light of rising costs. With the potential for significant savings and the ability to maintain operational flexibility, open-weight models are likely to play an increasingly central role in the AI strategies of large firms moving forward.

The findings from Capgemini not only highlight a significant trend in AI adoption but also signal a potential shift in how companies approach technology investments in the future. As organizations navigate the complexities of AI implementation, the focus will increasingly be on balancing performance, cost, and operational efficiency.