The public sector GenAI push needs a stronger foundation

Australia’s public sector is leaning hard into generative AI (GenAI), and for good reason. From service automation to document processing and citizen engagement, GenAI holds real promise to make the government more responsive, efficient and human-centric. But there’s a catch: many of the systems this technology depends on are nowhere near ready to support it at scale.

According to the Nutanix Public Sector Enterprise Cloud Index, more than 80% of public sector organisations globally say they have a GenAI strategy. That’s an encouraging signal; but it’s quickly undermined by the 76% who admit their infrastructure isn’t yet fit to support cloud-native applications. In today’s world, where secure, scalable AI hinges on containers, robust data pipelines, and resilient hybrid multi-cloud environments, that’s not a small oversight — it’s a strategic risk.

Public sector CIOs are under immense pressure, with heavy scrutiny on budgets and resource gaps that make even ‘keeping the lights on’ a delicate balancing act. When resources are already stretched to capacity just keeping departments online, where is the bandwidth to innovate and invest in AI initiatives going to come from?

Regardless of these issues, the building blocks of digital government are changing fast. Core infrastructure isn’t just a back-end concern anymore: it’s now the make-or-break layer that determines whether GenAI becomes embedded, responsible and impactful, or just another over-hyped pilot that never reaches maturity.

To be fair, there’s no lack of ambition. Most government agencies are chasing GenAI to unlock productivity gains, reduce costs and improve citizen outcomes. What’s notable is that long-term optimism remains strong, with nearly 70% of leaders expecting GenAI projects to deliver gains within three years, even if short-term ROI is shaky. This reflects a growing understanding that transformation is a marathon, not a sprint.

But the optimism must be backed by practical investment — and the most urgent area is capability. Only 51% of public sector teams say they have the skills to support GenAI today, and while 70% are hiring for GenAI roles, demand is outstripping supply, especially in high-pressure domains like cybersecurity, data engineering and platform operations.

Security and governance, in particular, can be major hurdles for further AI adoption. Given the extremely sensitive personally identifiable information many agencies hold, there is a high degree of caution around any AI use that would interact with these data sets. Innovation might be the aim, but not at the cost of cyber risk or crossing data privacy boundaries.

Without the internal skills to manage and secure these tools, many agencies risk becoming overly reliant on external vendors or off-the-shelf models, undermining data sovereignty and long-term flexibility. Deployment platforms with in-built data governance and role-based access controls can help mitigate these concerns.

This is where the infrastructure conversation returns. You don’t need a single-vendor AI stack to succeed, but you do need a consistent, cloud-native foundation that can scale across environments, support complex data governance, and adapt as regulations and organisational requirements evolve.

That’s where hybrid multi-cloud and container orchestration tools like Kubernetes have a critical role to play. They offer public sector teams the flexibility to build once and deploy anywhere — on-premise, in the cloud, or across edge environments — from a single pane of glass while enforcing the security and control that GenAI demands.

We’re already seeing early signs of this shift. Public sector organisations are increasingly containerising their GenAI workloads, modernising legacy platforms and experimenting with multi-environment deployments. The challenge now is moving from experimentation to standardisation, embedding these modern practices across departments so they become the default, not the exception.

If public sector leaders can align GenAI ambitions with infrastructure modernisation and workforce investment, there’s every reason to believe Australia can lead in responsible, high-impact AI adoption. But if we skip the hard parts — modernising legacy systems, upskilling teams, and building in governance by design — we risk creating a new digital divide, between those who can use AI well and deliver valuable outcomes, and those who simply use it.

GenAI is a powerful tool, but like any tool, its impact depends on how well it’s embedded. With the right foundation, Australia’s public sector can move beyond pilots and make GenAI part of everyday public service delivery. The race isn’t to adopt the fastest, but to build the smartest.

Image credit: iStock.com/BlackJack3D