By Enterprises want to move at software speed, and now increasingly at AI speed, but the infrastructure underneath many organizations still depends on humans to provision, adjust, and operate critical systems. That mismatch creates friction for platform teams, slows delivery, and makes it harder to scale cloud-native and AI workloads consistently across environments.
In a conversation with TFiR at KubeCon + CloudNativeCon Europe in Amsterdam, Bassam Tabbara, Maintainer at Crossplane argued that the declarative control model popularized by Kubernetes is now expanding into a broader infrastructure layer. His focus: Crossplane, the open source project that allows teams to define and manage cloud and AI infrastructure through Kubernetes-style APIs and controllers.
The Guest: Bassam Tabbara, Maintainer at Crossplane
Key Takeaways
- Crossplane brings a declarative, controller-based model to cloud and AI infrastructure management.
- Its CNCF graduation reflects ecosystem maturity, broad integration coverage, and enterprise-scale usage.
- Platform engineering teams are increasingly using the same operational model for both cloud-native and AI workloads.
- Agentic AI systems may depend on declarative APIs and control planes to deploy and manage infrastructure safely.
- The next phase of infrastructure automation includes both deterministic and probabilistic controllers.
***
In a recent TFiR interview, Swapnil Bhartiya spoke with Bassam Tabbara, Maintainer at Crossplane, about Crossplane’s CNCF graduation, the rise of platform engineering, and why control planes are becoming foundational to cloud and AI infrastructure operations.
Crossplane Extends the Kubernetes Model to Cloud Infrastructure
Tabbara described Crossplane as an effort to bring the declarative control model of Kubernetes to a much broader layer of infrastructure. Instead of limiting the model to containers and cluster resources, Crossplane applies it to cloud infrastructure and emerging AI infrastructure across multiple vendors.
Q: What is Crossplane, and why is it becoming so important for platform teams?
Bassam Tabbara: “Crossplane essentially brings the declarative control model to all of cloud and AI infrastructure. If you know how to use Kubernetes, you declare that you want to run a container, and Kubernetes manages deployment, scheduling, and failover. Crossplane does a similar thing for cloud infrastructure across different vendors. It has a declarative API for cloud and a set of controllers that manage the infrastructure wherever they are deployed.”
He connected that model directly to the demands of modern automation, especially as enterprises begin looking beyond conventional cloud-native operations and toward AI-driven systems.
Q: Why does this model matter even more now?
Bassam Tabbara: “The beauty of the declarative control model is that you declare your intent as a human operator or an agent, and then the control planes manage the rest autonomously. In the age of AI, everybody wants to move at AI speeds. If your backend infrastructure requires a human in the loop to do everything, then you go at human speeds. The only way to get to AI speeds is for every part of your system, from code generation to deployment, management, and operations, to run itself.”
Why CNCF Graduation Matters for Crossplane
Crossplane’s CNCF graduation is not just a symbolic milestone. Tabbara framed it as a reflection of the complexity of building an ecosystem around a project that sits at the center of many infrastructure integrations, including cloud providers, functions, and providers maintained across a large community.
Q: What did it take for Crossplane to reach graduation?
Bassam Tabbara: “Getting to a graduated state at the CNCF is a big deal. It takes a long time, a lot of patience, and a lot of commitment from maintainers. Crossplane is a unique project because it has many integration points. You cannot do anything with Crossplane without connecting it to a cloud provider like AWS, GCP, or Azure, and you also need providers and functions around it. Building that ecosystem takes time because of the diversity of vendors around it.”
He emphasized that the project now sits at a scale where the broader cloud-native community sees it as central infrastructure rather than a niche tool.
Q: What does graduation mean for the community and users?
Bassam Tabbara: “It is really awesome to get to the stage where the CNCF looks at this as a de facto project and gets it to a graduated state where it is being used widely, managing hundreds of millions of infrastructure on behalf of different companies at scale. It is not the end of the journey. It is actually the beginning of a new one.”
Crossplane, Agentic AI, and the Next Generation of Control Planes
One of the strongest themes in the discussion was the intersection of control planes and agentic AI. Tabbara argued that if agents are going to generate code, deploy systems, and manage infrastructure, then they will need a declarative interface and governance model rather than direct, ad hoc operational access.
Q: How does Crossplane fit into the emerging agentic AI story?
Bassam Tabbara: “It has become a really important substrate for the agentic AI story. If agents are getting to the stage of wanting to deploy and manage infrastructure, then a declarative API and controller-based model is the way to go, because it balances autonomy and brings the governance and compliance layer as well. We think control planes are critical plumbing for the agentic AI story.”
He also noted that AI is not only a workload to be managed by control planes, but also a force changing how control planes themselves may operate.
Q: How is AI changing the control plane itself?
Bassam Tabbara: “Control planes run controllers that do the work. Those controllers have been mostly deterministic in the past. We are now running probabilistic, LLM-based controllers as well. So even the control plane technology itself is benefiting from LLMs, and both are super exciting.”
How Platform Teams Are Using Crossplane for AI Infrastructure
Tabbara explained that AI workloads increasingly run on cloud-native foundations, especially Kubernetes, which means platform teams are becoming responsible for both traditional application infrastructure and newer AI infrastructure such as inference and training environments.
Q: What does this look like in practice for platform engineering teams?
Bassam Tabbara: “AI is being looked at now as a workload that runs on what looks like cloud-native infrastructure. Kubernetes is a great example. There is a lot of investment in making sure Kubernetes runs GPU workloads really well. The people who are going to run these AI workloads are also the people running Kubernetes in organizations. These platform teams are now being tasked with running more AI workloads on Kubernetes.”
That convergence matters because it brings AI operations into the same operational and organizational model that enterprises already use for cloud-native systems.
Q: Why is that convergence important?
Bassam Tabbara: “Everything is converging on a similar stack and converging on similar organizational principles within these organizations. It starts to look like a workload, not a completely independent system. That is exciting because it means that we are building on top of all the infrastructure and all the work that this community has been doing for years.”
Adoption, Gaps, and What Comes Next
Although Crossplane has reached a major maturity milestone, Tabbara was clear that the work is not finished. Developer experience, ecosystem coverage, AI operations, and the balance between deterministic and probabilistic controllers remain important investment areas.
Q: Where are the biggest gaps or pain points now?
Bassam Tabbara: “We still continue to invest in developer experience, how easy it is to build these control planes, the ecosystem, and the coverage of providers. There is a new thing every day. The AI workloads themselves, running on top of control planes or AI operations, are a critical part of this. How do you balance both deterministic and probabilistic controllers in the same control plane? Those are all areas of investment that we still continue to pursue.”
On adoption, he pointed to serious enterprise use, especially among large global organizations, while also acknowledging that there is still room for growth.
Q: Are you satisfied with Crossplane adoption today?
Bassam Tabbara: “Within the Global 2000 and the Global 5000, there is really serious adoption of Crossplane. It is hard to measure it exactly because it is an open source project, but Fortune 10, Fortune 50, and Fortune 100 companies are deploying it at scale and managing hundreds of millions of dollars of infrastructure. I am happy with the adoption, but I am also eager to see even more adoption.”
