If you’ve ever hosted a festive family meal across Asia — be it a Lunar New Year reunion, Diwali dinner, or Hari Raya gathering — you know the choreography involved. There’s the claypot bubbling in one corner, rice steaming in another, side dishes arriving just in time. It takes planning, coordination, and the right tools to ensure everything’s hot, harmonious, and served at once.
That’s what Kubernetes is doing for AI workloads today: acting as the head chef in the enterprise kitchen. Generative AI and agentic AI may be the headline acts, but behind the scenes, Kubernetes is turning innovation into something scalable, consistent, and cost-efficient.
So where’s the bottleneck? Too many organisations are betting big on AI without modernising the infrastructure to support it. The result? Pilots stall, costs balloon, and potential is left on the table.
To truly operationalise AI at scale, we need to ask: Is our foundation ready?
Increasingly, that foundation starts with a comprehensive Kubernetes platform that includes rich data services.
To scale AI, start with Kubernetes
Most boardrooms I sit in are doubling down on AI. But when it comes to deployment, they hit the same wall — infrastructure that wasn’t built to support AI’s complexity.
The numbers are just as telling. Nutanix’s latest Enterprise Cloud Index found that more than 70% of Singapore organisations continue to grapple with implementation challenges as they actively explore generative AI’s potential, highlighting infrastructure as a key concern.
The ambition is there, but the foundation isn’t keeping up.
That’s where Kubernetes comes into the picture. While it’s no magic fix, it gives organisations a way to manage complex, containerised workloads without adding complexity. Kubernetes makes AI workable across environments — whether that’s public cloud, on-premises, or edge. It helps AI move from lab to live without breaking things in the process.
According to Edge Delta, more than three in five enterprises now use Kubernetes. It’s fast becoming the foundation of AI infrastructure. So what’s stopping more organisations from using it to scale AI successfully?
What’s Holding AI Back
Moving from idea to impact is where most AI projects falter. Use cases are clear. Teams are energised. But execution often stalls. It’s not for lack of ambition, it’s about readiness.
Operational complexity is often the first hurdle. Fragmented legacy systems, disparate datasets and outdated processes struggle to coordinate AI workloads across diverse environments, hindering automation, visibility, and consistency.
Infrastructure readiness comes next. AI’s resource demands require scalable systems without added complexity or cost. Yet nearly 80% of Singapore organisations admit their current systems need major AI-ready upgrades, highlighting a significant gap, according to the same Nutanix report.
Scaling from pilot to production is where momentum often stalls. Around 98% of global organisations face challenges when scaling AI workloads, many struggling with integration. Promising proofs of concept may falter with live data and governance. Manual processes and slow iteration hinder the translation of promising ideas into scalable impact.
Enterprise readiness also includes the ability to manage updates and patches systematically, meet compliance requirements, and offer enterprise-grade backup, disaster recovery, and business continuity solutions.
One city government in Southeast Asia used Kubernetes to integrate public services into a unified digital platform, resulting in greater efficiency and improved citizen access.
These aren’t new challenges, but AI magnifies them. Unless they’re addressed head-on, more projects will remain stuck in the gap between potential and production.
A Strategic Approach
Singapore is catching up, but there’s still a gap. According to the Nutanix report, just over 60% of organisations here have containerised some of their applications, compared to more than 80% globally. This matters, as containerisation is often the first step towards building infrastructure that can handle the scale and unpredictability of real-world AI.
So, how should business leaders move forward?
Here are some approaches I’m seeing work across the region, from fast-moving start-ups to highly regulated incumbents.
- Embrace a multi-cloud mindset
Some of the most effective teams I’ve seen are designing for multi-cloud from day one. They’re training AI models in the public cloud, then deploying them closer to their data — on-premises or at the edge — depending on performance, cost, or compliance needs. Kubernetes makes that shift possible without added friction or headcount pressure.
- Build for platform flexibility
AI doesn’t work well in silos, but too often, infrastructure does. Teams lose momentum when they need to rebuild or refactor systems just to modernise. Infrastructure that supports both virtual machines and containers allows teams to adopt Kubernetes incrementally, without disruption. This gives leaders consistency in governing and scaling workloads, without having to start from scratch or hire specialist skills.
- Prioritise data security and lifecycle management
As AI scales, it becomes more important to know where your data is stored, how it moves, and who has access. More than that, it’s what gives leaders confidence that their AI systems are doing the right thing for the right reasons.
Getting AI to the table
The hardest part of scaling AI isn’t about choosing the right model. It’s aligning the moving parts — data, infrastructure, security, governance — so they work together at the right time, in the right sequence. Kubernetes plays a key role here by helping organisations manage that complexity across environments.
To support AI at scale, organisations need infrastructure that enables:
- Efficient Kubernetes operations and lifecycle management.
- Flexibility to adapt to evolving requirements.
- Support for both stateless and stateful applications.
- Rapid development and iteration.
- Consistency across deployment environments.
- Visibility into system performance and operations.
- Robust security and compliance enforcement.
- Reliable access to data storage and services.
Kubernetes may not be the flashiest part of the AI stack, but it brings discipline to complexity. With the right approach, it helps organisations move from pilot to production with greater confidence and consistency.
For leaders operationalising AI, Kubernetes might not be the main course. Yet, more often than not, it’s what brings the table together.
