This report is part of a research collaboration between McKinsey, the Mozilla Foundation and the Patrick J. McGovern Foundation.
Open Source Software has long been an essential element of the technological ecosystem. Commercial software generally requires a commercial license or a subscription and restricts access to its basic technology. However, open source tools are developed in collaboration and made available to the public to use, modify and distribute with much less restrictions. The open source model gives developers the possibility of adapting and shaping well -cut solutions to the specific needs of their organizations.
The current age of AI is no different. While more and more companies are building and deploying solutions focused on AI in their business, they turn to an increasing range of open source technologies. These offers include Meta’s Llama family, the Gemma family from Google, the Olmo family of the Allen Institute for Artificial Intelligence, the Nemo family of Nvidia, Deepseek-R1 and the Qwen 2.5-Max by Alibaba Cloud-many of which are the rapid closure of the performance gap compared to the AI Propriement models.
A new survey, which is of its kind, with more than 700 technological leaders and senior developers in 41 countries, led by McKinsey, the Mozilla Foundation and the Patrick J. McGovern Foundation, provides the largest and most detailed analysis of the way in which companies reflect and deploy an open source AI in their organizations. The results suggest that managers adopt open source tools as essential components of their technological batteries, citing advantages such as high performance, ease of use and lower implementation and maintenance costs compared to proprietary tools. The developers, on the other hand, consider the experience of Open Source more and more as an important part of their overall satisfaction at work. Although open source solutions are with concerns about security and value delay, more than three -quarters of survey respondents expect to increase their use of open source AI in the coming years.
Our research shows that companies use open source models more than we could expect. In several fields of the AI technology stack, more than 50% of respondents report that their organizations use open source AI technologies (often alongside tools of players such as Anthropic, Openai and Google). Organizations which grant a high priority to AI are the most likely to use open source technologies: respondents of organizations who consider AI as important for their competitive advantage are more than 40% more likely to report the models and tools of Open Source than respondents from other organizations. The technological industry opens the way, 72% of respondent organizations using an open source AI model, compared to 63% of respondents in all organizations.
The survey results provide several key results:
- Use and open source AI trends:
- The use of open source AI technologies is widespread. More than 50% of respondents declare to take advantage of an open source solution in each of the data, models and tools of the fields of the technological battery.
- The technical maturity and the experience of developers influence the use of open source AI. Such use is the highest in technology, media and telecommunications (70%), and experienced AI developers are 40% more likely than peers to use open source IA solutions.
- Organizations use open source AI tools for familiar players. The most commonly used open source AI tools among companies in January 2025 are those developed by major technology players, such as Meta with her Llama and Google family with his Gemma family.
- Value for organizations and developers:
- Most respondents are satisfied with their Open Source models. The main reasons for satisfied satisfaction are performance and ease of use.
- Open source AI tools lead to cost benefits, while owners’ tools have faster value. Respondents claim that open source AI has a reduction in implementation costs (60%) and a reduction in maintenance costs (46%). But respondents see a faster time to assess owner AI tools (48%).
- Developers appreciate the Open Source tools. Most developers (81%) report that experience with Open Source tools is highly appreciated in their field and that working with such tools is important for their work satisfaction (66%).
- Future perspectives on open source AI: organizations are open to a mixture of open and owners. Almost three -quarters of respondents (more than 70%) say they are open to open source technologies or AI owners in areas of the technological battery.
- Risks and attenuation strategies:
- Open Source tools involve potential challenges. Respondents cite concerns about cybersecurity (62%), regulatory compliance (54%) and intellectual property violation (50%) when they engage with AI tools.
- Organizations implement guarantees to manage the risks associated with Open Source tools. Strategies include strengthening information security frameworks and software supply chain controls, using the third -party model assessment and the implementation of guards to limit the behavior of the model.
Overall, more than three quarters of respondents – 76% – explain their organizations to increase the use of Open Source technologies in the coming years. This can be partly because open source tools have been part of a dynamic software ecosystem in many categories of corporate software, as well as a fundamental resource for developer communities for decades. While AI continues to improve, business leaders and technology should pay particular attention to all the opportunities and innovations that emerge. As in the cloud and software industries, a multimodel approach will probably be widespread for many companies, with open source technologies and owners coexisting in several areas of AI technological battery.
For the complete data and information, download the report here.