PARIS — November 12, 2026 — Major companies are reconsidering how they deploy artificial intelligence as rising costs challenge corporate budgets, prompting increased interest in lower-cost models and more flexible technology strategies.
A new study by the Capgemini Research Institute, based on a survey of 500 senior executives at large organizations across 10 countries, found that 59% of respondents said their companies had spent significantly more on AI over the past year than originally expected or budgeted.
Among organizations reporting higher-than-expected spending, costs averaged 24% above expectations, according to the study.
The findings highlight a growing challenge for businesses that have moved quickly to introduce AI tools into their operations but are still working to establish how much those systems cost and what measurable value they deliver.
Companies reconsider their AI spending
The study found that concern about AI expenses is increasing. More than half of the surveyed executives said AI costs were a significant concern, while only 56% expressed confidence in their ability to forecast future spending as adoption expands.
The financial consequences can extend beyond the direct cost of using AI models.
Companies must also account for infrastructure, employee oversight, cybersecurity, governance, maintenance and the additional work required to check or correct AI-generated results.
These indirect expenses can make it difficult for organizations to calculate the true cost of deploying AI across multiple departments.
For companies under pressure to demonstrate returns on technology investments, the issue is becoming increasingly important.
Growing interest in open-weight AI models
One response has been a shift toward open-weight AI models, which allow organizations to access model parameters and deploy or adapt systems within their own technical environments, subject to the applicable licenses.
Unlike many closed commercial systems accessed primarily through cloud-based services, open-weight models can give businesses greater flexibility over deployment, infrastructure and operating costs.
Examples include models developed by companies such as DeepSeek and Mistral.
According to Reuters’ reporting on the Capgemini study, 59% of executives at large companies expect their organizations to increase their use of open-weight models over the next year.
The attraction is not necessarily a complete replacement of established commercial AI services. Instead, companies may use lower-cost models for routine or specialized tasks while retaining more advanced systems for complex applications.
The rise of multi-model AI strategies
The research also points to growing interest in multi-model strategies.
Rather than relying on a single AI provider, businesses are increasingly considering systems that select different models depending on the task, cost, performance requirements, speed and security considerations.
Such an approach could allow organizations to reserve more expensive models for work that requires their advanced capabilities while using smaller or less costly systems for simpler operations.
For technology departments and financial executives, this creates a new challenge: determining which model is appropriate for each business process without compromising reliability or security.
Measuring value becomes as important as controlling costs
The study suggests that many organizations still struggle to connect AI spending with measurable business results.
While companies may track the number of AI users, subscriptions or generated outputs, these figures do not necessarily demonstrate whether the technology is improving productivity, reducing operating expenses or increasing revenue.
Capgemini reported that only one in five organizations surveyed calculates what it calls AI Yield, a measure intended to assess the verified value of AI-assisted outcomes after accounting for the full cost of delivering them safely.
This gap could become increasingly significant as companies move from experimental AI projects to larger deployments involving employees, customers and critical business operations.
A more disciplined phase of AI adoption
The findings suggest that corporate AI strategies are entering a more financially disciplined phase.
Businesses are not necessarily abandoning artificial intelligence. Instead, they are becoming more selective about which systems they use, how they deploy them and whether the investment produces meaningful results.
The next stage of competition may therefore depend not only on developing more powerful models, but also on making AI more affordable, efficient and commercially sustainable.
For the global technology industry, the growing focus on cost management could reshape competition between major AI developers, cloud providers and companies building alternative models.
World Media Feed Editorial Team
Sources
- Reuters — Most large firms plan higher open-weight AI use to cap bills, study shows
- Capgemini Research Institute — AI costs and multi-model strategies
About the Author
