CTO Chris Aniszczyk on the CNCF Push for AI Interoperability

Key Concepts

Cloud Native: An approach to building and running scalable systems using technologies like containers, microservices, and Kubernetes. Focuses on resilience, scalability, and dynamic development.
AI Native: Encompasses both AI native infrastructure (leveraging cloud native technologies to support AI workloads) and AI native development (using AI-powered tools throughout the software lifecycle).
CNCF (Cloud Native Computing Foundation): A neutral home for collaboration on cloud native technologies, hosting projects like Kubernetes, Prometheus, and Envoy.
Kubernetes AI Conformance: A new set of standards defining capabilities required for Kubernetes clusters to reliably run AI/ML workloads.
DRA (Dynamic Resource Allocation): A Kubernetes feature supporting different types of accelerators (GPUs, TPUs) for AI workloads.
MCP (Model Configuration Protocol): A protocol enabling agents to communicate with services, foundational for agentic infrastructure.
Agentic AI: AI systems comprised of autonomous agents that interact with each other and services.
AI Gateways: Specialized gateways extending API gateway and service mesh functionalities to handle the unique requirements of AI workloads.

The Future of Cloud Native & AI Native: A 2026 Outlook

This discussion, featuring Chris Anistic, CTO of the CNCF, explores the evolving landscape of cloud native technologies and their intersection with Artificial Intelligence, specifically looking ahead to developments anticipated in 2026. The conversation centers on defining AI native, the role of CNCF in supporting both cloud native and AI native ecosystems, and emerging trends in infrastructure and development.

Defining Cloud Native and AI Native

Chris Anistic differentiates between cloud native and AI native. Cloud native, as defined by the CNCF, is an approach to building and running scalable, resilient systems using technologies like containers, microservices, and Kubernetes. Key characteristics include loose coupling, observability, and scalability. AI native, however, builds upon cloud native.

He further breaks down AI native into two aspects: AI native infrastructure – which relies heavily on cloud native technologies to orchestrate and manage AI agents – and AI native development – which involves developers utilizing AI-powered tools throughout the entire software development lifecycle (planning, design, development, testing, deployment, and maintenance). He posits that agents, in many ways, resemble microservices but with unique scaling and management characteristics. He states, “in order to be AI native you have to be cloudnative by default.”

CNCF’s Role and the Kubernetes AI Conformance Program

The CNCF’s role is to provide the foundational technologies enabling cloud native at scale, including Kubernetes, gRPC, Prometheus, and OpenTelemetry. Anistic details his responsibilities as CTO, overseeing the technical aspects of the 200+ CNCF projects, managing the technical operating committee (TOC), and supporting the technical community. He also highlights the recent launch of the AIF (Agentic AI Foundation), a sister organization to the CNCF, hosting projects like MCP and Goose.

A significant focus of the discussion is the new Kubernetes AI Conformance program. This initiative aims to establish a baseline of compatibility and support for AI/ML workloads across different Kubernetes distributions and cloud providers. Anistic explains that, mirroring the success of the original Kubernetes conformance program, this new effort addresses the inconsistencies and complexities encountered when scaling Kubernetes environments for generative AI.

The program focuses on ensuring support for:

Dynamic Resource Allocation (DRA): Enabling the use of GPUs, TPUs, and other accelerators.
Networking and Traffic Management: Adapting to the unique traffic patterns of AI inference workloads.
Standardized APIs: Preventing fragmentation and ensuring portability.

He emphasizes the importance of vendor neutrality and community collaboration, noting the participation of major cloud providers (Amazon, Google, Microsoft) in the initiative. He states, “Our goal at least within the CNCF is to have basically every major provider out there… to support this baseline of compatibility.”

New Incubating Projects in CNCF

The conversation highlights two recently incubated projects:

Metalcubed: A project originating from Ericsson, designed for managing Kubernetes deployments on bare metal infrastructure. It integrates with OpenStack’s Ironic component. It’s targeted towards telecommunications providers and organizations managing their own data centers.
OpenYear: A control plane for managing Kubernetes deployments at the edge. It utilizes a custom resource definition (CRD) called “node pool” to manage edge clusters, and is relevant for use cases like retail stores and the example of Chick-fil-A managing Kubernetes in each of its locations.

Emerging Trends and Predictions for 2026

Anistic predicts several key trends for 2026:

Growth of AI Infrastructure Technologies: All CNCF projects will evolve to better support AI workloads. Existing projects like OpenTelemetry will be refactored to improve observability of AI systems.
MCPification of Everything: Widespread adoption of the Model Configuration Protocol (MCP) for agent communication.
Rise of Specialized Cloud Providers (“NeoClouds”): Increased use of GPU-focused and smaller cloud providers alongside the major hyperscalers, facilitated by the portability enabled by Kubernetes and the new AI conformance program.
AI Gateways: The emergence of specialized gateways extending API gateway and service mesh functionalities to handle the unique requirements of AI workloads, including indeterministic behavior and intelligent routing. He anticipates consolidation in this space, predicting a reduction in the number of competing solutions. He clarifies that these AI Gateways build upon existing technologies, stating, “it’s like the same technology you have like bytes flow in and then you have something that basically decides what to do with it, where to send it, how to route it.”

Logical Connections & Synthesis

The discussion demonstrates a clear progression from defining the foundational concepts of cloud native and AI native, to outlining CNCF’s role in fostering these ecosystems, and finally, predicting future trends. The Kubernetes AI Conformance program is presented as a crucial step in bridging the gap between cloud native infrastructure and the demands of AI workloads. The emphasis on MCP and AI gateways highlights the growing importance of agentic AI and the need for specialized infrastructure to support it. The prediction of a rise in specialized cloud providers underscores the increasing flexibility and portability enabled by cloud native technologies.

In conclusion, the future of cloud native is inextricably linked to the advancement of AI. CNCF is positioned to play a central role in this evolution, providing the foundational technologies and standards necessary to build and deploy scalable, resilient, and AI-powered systems. The key takeaway is that AI native is not a replacement for cloud native, but rather an extension of it, requiring a robust and standardized cloud native infrastructure as its foundation.