More cloud capacity won’t mean greater AI readiness for Australia

Investments in cloud infrastructure are climbing at a steady pace, and Australia is attracting a significant share of that spending: AWS plans to invest more than AU$20 billion in local data centre infrastructure through 2029, while Microsoft has committed over AU$25 billion to Australian AI and cloud capacity over the same period. These investments are important, but capacity and readiness aren’t the same thing. Compute capacity alone will not determine which Australian organisations succeed with AI.

The real constraint is becoming data. IDC forecasts that Asia-Pacific will generate 160.9 zettabytes of data in 2028, with regional investment in AI rising to US$175 billion (over AU$250 billion). Growth of this scale is already reshaping how cloud infrastructure in Australia is designed.

Australia’s next phase of cloud maturity will be less about how much compute is available: it will be measured by whether data can move, perform and remain governed across the environments businesses increasingly depend on.

Australia’s AI debate is too focused on compute

AI workloads place different pressures on the cloud. Agentic systems consume and create massive datasets. Inference runs continuously, and performance hiccups can quickly flow through to the user experience. By 2027, most new enterprise applications may include embedded AI inference as a default feature. This means the data layer no longer exists quietly in the background: its performance and availability shape the success of the product itself.

Yet much of the AI infrastructure conversation remains centred on models, GPUs and data centre capacity. Those investments matter, but they solve only one part of the problem. A powerful model connected to fragmented, poorly governed or inaccessible data will produce limited business value, no matter how much compute sits behind it.

AI has amplified data’s gravity: storage, snapshots, replication and recovery matter more, and they matter constantly. For organisations responsible for patient records, citizen information or essential public services, weak data controls don’t just create a compliance risk — they erode public trust and interrupt critical operations.

Multi-cloud wasn’t a strategy, it accumulated

Many Australian organisations became multi-cloud by accumulation rather than design. Different teams made sensible calls at different times and multi-cloud was the result. AI is accelerating that pattern: teams now pick environments based on model availability, performance, geography or cost, and fragmentation follows whenever data isn’t managed consistently across those environments.

That’s why the conversation with Australian technology leaders is changing. It is shifting from which cloud to choose, to how data can behave consistently across all of them. Cloud choice only becomes strategically useful when data can follow the workload without forcing teams to rebuild governance and recovery each time.

Users don’t want to assemble a data stack anymore: they want enterprise-grade storage native to the cloud they’re already using — with the same console, APIs, billing and operational model — and they want to move or rebalance workloads between hyperscalers with far less data management friction when requirements change.

In Australia, there is the added dimension of sovereignty: government agencies and highly regulated industries like healthcare require sensitive information to remain within controlled environments while still supporting AI-driven decision making. Future infrastructure may need to span isolated sovereign environments, on-premises systems and public cloud services without fragmenting how data is managed. This is the practical role of intelligent data platforms: giving organisations a consistent way to access, protect and govern data across on-premises environments and the public cloud, without weakening the controls surrounding their most sensitive information.

The economics follow the data

Data services can no longer be treated as an add-on to compute: they increasingly determine application performance, resilience, and operating costs. In AI-driven applications, where data sits can influence the latency users experience, the cost of moving or duplicating it, how quickly systems recover and how much operational overhead teams carry. AI raises the stakes on all four, because AI fails fast when data isn’t fast, consistent or reliable.

It’s why Australian organisations running mission-critical workloads are increasingly looking for predictable throughput, low latency, and built-in protection, without redesigning applications. These services run inside the hyperscalers’ Australian regions, keeping data onshore, while built-in storage efficiencies and tiering help keep the economics contained as workloads scale.

Where this leaves Australia

The next wave of cloud infrastructure decisions in Australia won’t be led solely by compute. As hyperscaler capacity expands, compute will remain essential — but capacity alone will be less differentiating. What will meaningfully drive decisions is data: where it accumulates, the agentic workloads it powers, the regulations that apply to it and the costs associated with it.

This shift in gravity towards data will reshape budgets, as spending will extend beyond compute into capabilities for the governing, protecting, moving and recovering of data. It will redirect teams, with developers building where data already lives, because that’s where AI features ship fastest. And it will narrow strategic decision making to a single question: can our data move, meet compliance and perform wherever the business’s AI workloads need it next?

Organisations that can answer yes are already positioned for what comes next. Getting there starts with treating the data layer as the strategy, not the afterthought.

Image credit: iStock.com/luchezar