Red Hat sets out hybrid AI path for ANZ enterprises

Red Hat has outlined a hybrid platform strategy for Australian and New Zealand enterprises seeking to move artificial intelligence into production while retaining control over cost, security and data sovereignty.

The company is positioning Red Hat Enterprise Linux, OpenShift and Ansible as a common foundation for AI and existing workloads. Its approach is designed to let organisations use different models, accelerators and cloud environments without rebuilding their underlying platforms.

The strategy responds to growing AI consumption within enterprises. Red Hat said individual token costs are generally falling, while aggregate spending is rising as engineering teams and other employees use AI more extensively. The company has introduced token budgets for some of its own engineering teams to manage that consumption.

Platform control

"Enterprises will need many models. Some of those will come locally, within Australia and New Zealand or within your region. Some of those will come from other parts of the world. In many cases, you'll use models with public APIs, and we want to make sure that you're able to take advantage of all of that," said Ashesh Badani, Senior Vice President and Chief Product Officer, Red Hat.

Red Hat's architecture spans hardware accelerators, AI infrastructure, inference services, model services and agentic applications. OpenShift provides network isolation and GPU sharing, while an AI gateway can route requests according to cost, performance or governance requirements.

The company said its validated model programme covers more than 20 model collections and nearly 700 variations. It is also adding AI-driven orchestration to Ansible, alongside task-driven and event-driven automation. This would allow agents to initiate workflows while retaining human approval at designated points.

Red Hat is working with NVIDIA on hardware and software integration. The companies are seeking day-zero compatibility between new NVIDIA hardware and Red Hat's software stack. Their work also covers reference architectures intended to improve the number of tokens produced per watt and shorten the time to the first token.

"Agentic AI is now not just a chatbot. It's potentially interacting with your critical systems. Having the partnership with Red Hat around the secure and open alliance is extremely important," said Sudarshan Ramachandran, Country Manager - Enterprise, Australia/New Zealand, NVIDIA.

Sovereign systems

The platform strategy also addresses operational, data and software supply chain sovereignty. Red Hat's position is that organisations should be able to run workloads across their chosen environments, govern models within regulatory boundaries and use software that has been verified and patched.

These requirements are particularly relevant to the Defence Science and Technology Group, which operates networks ranging from unclassified environments to highly isolated classified systems. Its researchers need adaptable networks that can work across multiple classifications and connect with international partners.

The group began moving away from selected vendor platforms as part of a wider infrastructure change. It also needed to retain compatible systems and capabilities across its Five Eyes relationships. Staff training and experience formed a substantial part of the migration because administrators had spent much of their careers working with another design model.

"It's really about maintaining that flexibility and not getting locked into an ecosystem. We need to be able to work with industry, academia and our international partners. Locking into one style of system is not going to work. We need to be able to integrate with our partners, or we're not going to be very effective," said Shane Bennett, Group Leader Cyber Integration, Defence Science and Technology Group, Department of Defence.

The group has introduced the platform across one or two systems and plans to extend it to the rest of the 10 to 15 networks managed by Bennett's team. New Zealand, Canada and the United Kingdom are also examining aspects of the Australian work, according to Bennett.

Operational use

Auckland Transport has taken a separate open-source approach to computer vision. Its 25-person computer vision team operates an in-house platform that analyses thousands of video streams in real time. The system supports transport safety, network performance, compliance and operational decision-making.

The transport authority moved from third-party software to an internally developed, cloud-native platform. It later adopted Red Hat OpenShift for container orchestration, monitoring and workload visibility. The team plans to extend the platform into data science, machine learning, predictive analytics, generative AI and multimodal AI.

Westpac New Zealand has used Red Hat products as part of a move from tactical infrastructure remediation to a broader modernisation programme. The bank said it completed the change without customer disruption. Its next priorities include automating a growing backlog of operational work while retaining human decisions and improving application delivery.

Microsoft is also working with Red Hat on hybrid AI deployments through Azure Red Hat OpenShift. The companies are promoting a migrate, modernise and extend model. This allows enterprises to move existing applications, update selected systems and expose business logic to agentic platforms without a wholesale rewrite.

Microsoft said enterprises should begin with a high-value business use case, identify the data an agent requires and map the existing business process. The aim is to determine where AI can reduce errors, shorten processing time or improve an outcome while preserving human oversight.

Innovation awards

Red Hat recognised three organisations in its APAC Innovation Awards:

• Hybrid Cloud Infrastructure and Cloud-Native Development: Auckland Transport
• Hybrid Cloud Infrastructure and Virtualisation: Defence Science and Technology Group
• Hybrid Cloud Infrastructure and Automation: Westpac New Zealand

"Hybrid isn't going anywhere. We still have applications that run on-premises. We still have applications that run in a hybrid environment. We do obviously have a lot of cloud-native and AI-first applications that we are building. An agent will need to be able to access all the business logic that is sitting across these multiple applications, regardless of where they sit," said Udaiya Karuppasamy, Senior Director, Enterprise Sales ANZ - Azure Cloud & AI, Microsoft.