Linux Foundation Research, supported by the Cloud Native Computing Foundation (CNCF) and Broadcom, has released a new research report highlighting the reality of modern infrastructure and how organizations use virtual machines (VMs) alongside containers, with a particular focus on the role of virtualization in supporting emerging AI and Agentic AI workloads.
The report, titled "Virtual Machines, Agentic AI, and Workload Mix: 2026 Cloud Workload Survey Results," was based on responses from 492 leaders and practitioners in cloud infrastructure fields during September 2026. The results showed that virtual machines still constitute an average of 53% of production applications across the various organizations that participated in the survey.
Continued Presence of Virtual Machines in Production Environments
Contrary to the prevailing belief that containers and Kubernetes technologies might completely replace traditional infrastructures, statistics confirm that virtual machines still represent a key pillar; 51% of participants reported that at least half of their production applications run as virtual machines. Additionally, 71% of those surveyed indicated expectations that the share of virtual machines would remain within the same range over the next 12 months.
Organization size plays a notable role in this reliance; companies with 10,000 or more employees recorded an average virtual machine share of 64%, compared to 45% in companies with fewer than 500 employees. Regarding the management of these environments, 61% of participants indicated the existence of full collaboration or integration between container and virtual machine management teams, while teams work independently and in isolation in only 14% of organizations. Furthermore, the usage rate of the CNCF project KubeVirt for managing virtual machines within Kubernetes platforms reached approximately 8%.
The Role of AI and Workload Isolation Security
As organizations expand the deployment of AI applications and Agentic AI that can generate and execute code and interact with external tools, the process of workload isolation becomes a paramount security necessity to contain any unexpected code execution.
34% of participants who are expanding their use of virtual machines confirmed that AI workloads were a primary reason for this growth, while 47% cited security and isolation requirements, 48% cited improved hardware resource utilization, and 53% cited general growth in virtual machine-based applications.
What Does This Mean for Infrastructure and Cloud Teams?
These results demonstrate that the future of cloud infrastructure does not necessarily depend on a single deployment model, but rather reflects an integrated environment where different models coexist according to application and operational requirements. The report emphasizes the importance of exploring several paths for platform engineers and security practitioners, including:
- Coordinating operational processes across various types of workloads.
- Defining appropriate isolation boundaries for AI environments.
- Managing infrastructure resources with high efficiency.
- Evaluating emerging protection technologies such as Confidential Computing and MicroVMs.