APJ enterprises face infrastructure gap for AI adoption

Enterprises across Asia Pacific and Japan are accelerating artificial intelligence initiatives, though many remain constrained by infrastructure limitations that hinder deployment at scale.

Industry data indicates that between 77% and 87% of organisations across the region report that their on-premise infrastructure is not yet ready to support AI workloads. The shortfall has become a growing concern for IT leaders who face pressure from executives and boards to deliver AI capabilities while existing systems struggle to support emerging demands.

AI pressure

"There's probably not a conversation that I have today with a C-level executive where they don't talk about the pressure that they're feeling from the CEO or from the board that are setting very aggressive targets. And then on the other side, they're essentially not ready to deliver on the AI applications that are required. Oftentimes, it's because the infrastructure that's required to provide it simply isn't there yet," said Jay Tuseth, Vice President and General Manager, APJ, Nutanix.

Artificial intelligence has become the latest phase in a broader shift in enterprise technology. Earlier transitions centred on cloud computing and software delivered through subscription models. The current phase focuses on AI applications that require significant computing resources and complex data architectures.

As organisations experiment with AI use cases, many are discovering that legacy environments were designed for traditional enterprise workloads rather than large-scale model inference, machine learning pipelines, or distributed edge deployments.

Infrastructure design has therefore become a strategic issue for corporate leadership. Systems must support AI workloads across multiple locations, including data centres, edge environments and hyperscale public clouds.

Infrastructure shifts

To support those requirements, organisations are adjusting their infrastructure strategies toward hybrid environments that combine on-premise infrastructure with public cloud services.

Data sovereignty rules across the region are shaping these deployments. Countries including Singapore, India and Australia have introduced regulations or guidelines that affect how organisations manage and store data used in AI systems.

Compliance obligations require enterprises to maintain local control over certain datasets while still enabling distributed computing environments for AI applications.

Infrastructure vendors have also increased focus on storage and performance integration to support AI workloads. Some systems now incorporate flash-based storage designed to deliver very low latency, which is required for real-time AI inference.

Hardware partnerships are also becoming part of the infrastructure stack. Semiconductor vendors and infrastructure platforms are increasingly collaborating to support agent-based AI models and high-performance compute workloads.

Edge deployments

A second trend affecting infrastructure design is the expansion of AI workloads beyond traditional corporate environments.

Many organisations are deploying AI systems in operational environments where computing infrastructure must function outside centralised data centres.

Examples include mining operations in Australia, micro-banking networks in India and retail environments in Southeast Asia.

These deployments introduce new constraints around network connectivity, system management and security oversight. In many cases, there are limited technical staff on site to maintain systems.

"As enterprises push AI workload beyond headquarters they have some unique challenges. And those challenges maybe can be summarised in three key areas: The performance bottleneck, security and sovereignty, and deployment and management complexity that we have to deal now with the AI infrastructure where no IT talent resides," said Daryush Ashjari, CTO and VP of Solutions Engineering, APJ, Nutanix.

Container-based software platforms are becoming a key part of addressing these constraints. Containers allow applications to run consistently across different infrastructure environments and enable simplified deployment at remote locations.

Survey findings indicate that 85% of enterprises globally report that AI initiatives are accelerating container adoption. In India, 97% of organisations expect containerised workloads to increase as AI adoption grows.

Shadow AI

At the same time, many organisations are dealing with the emergence of AI tools deployed without formal oversight from IT departments.

The phenomenon, often described as "shadow AI", occurs when business units adopt AI applications independently in order to meet internal deadlines or deliver new capabilities quickly.

"Nearly four in five IT leaders in our region encounter AI tools implemented outside of IT and oversight. This is a massive challenge when it comes to organisation control and longevity of this solution and maintaining compliance and mandatory requirement where outside of IT decisions are made," said Ashjari.

Unmanaged AI tools raise concerns around data security, regulatory compliance and operational governance. Organisations in Singapore and India report particularly high levels of such activity.

To address the issue, some enterprises are building internal self-service platforms that allow business teams to access approved AI resources through a centralised environment. This approach enables faster deployment while maintaining governance controls.

Market impact

Changes in the infrastructure market are also influencing enterprise technology decisions.

Recent policy shifts affecting VMware customers have prompted some organisations to review alternative platforms for virtualisation and cloud infrastructure management.

"As Broadcom has made changes to the VMware partner ecosystem in addition to some very challenging business policy that's impacting their end customers we're seeing both of them navigate their way to alternatives, including Nutanix, which I think is probably the most viable alternative," said Tuseth.

The broader trend reflects the increasing role of infrastructure in AI adoption strategies. As organisations move from experimentation to operational deployment, computing platforms must support large volumes of data, complex software frameworks and distributed environments.

For many enterprises across Asia Pacific and Japan, the challenge now lies in aligning infrastructure capabilities with executive expectations for AI-driven transformation.