F5 released its latest Application Strategy Report and found that 78% of organizations worldwide run AI inference in-house.
In Australia and New Zealand, this figure is 48%, demonstrating a significant gap with the broader market as companies move artificial intelligence workloads into daily operations.
Based on responses from enterprise IT and security leaders, the report suggests that AI is moving beyond pilot projects and into everyday production environments. Across our global sample, organizations currently manage an average of seven AI models in production, and 77% said inference has become their primary AI activity, ahead of model building and training.
In contrast, organizations in Australia and New Zealand reported on average about four AI models in operation. Still, 72% said they already use AI for automated operational decision-making.
The findings also demonstrate how AI adoption is closely tied to broader changes in enterprise infrastructure. Globally, 93% of enterprises use a multicloud setup, and 86% say they run applications on-premises, in the public cloud, or in colocation environments.
The same pattern is seen in Australia and New Zealand. Three-quarters of organizations in the region use more than one cloud provider, and 73% say they operate from multiple on-premises data centers.
business shift
The report argues that AI inference now stands alongside other business-critical systems and needs to be managed with the same discipline. This marks a shift from the early stages of enterprise AI adoption, which focused on experimentation, proof-of-concept work, and model training.
Research shows that only 8% of organizations rely solely on public AI services. Most enterprises use a combination of models and environments, increasing the need to direct workloads across systems and set policies around cost, accuracy, and availability.
Kunal Anand, chief product officer at F5, said the data shows a clear shift in the way companies are handling artificial intelligence.
“AI has moved from experimentation to production. The question now is not whether companies will use AI, but whether they can run it reliably, securely, and at scale,” Anand said.
He said the impact extends beyond infrastructure.
“This year’s data shows a clear shift. AI inference is becoming core to business. This means delivering AI has become a traffic management challenge, and AI security has become a governance and control challenge. Companies that understand this shift early will move faster and more securely,” Anand said.
security pressure
Security also emerged as another major theme in the study. Globally, 88% of organizations say they are experiencing AI-related security challenges, and 98% say they are preparing for agent AI (autonomous systems that require identity, permissions, and control).
In Australia and New Zealand, the main barriers were cost and skills. Approximately 44% of organizations in the region identified skills gaps and the high cost of AI workloads as their biggest hurdles, while 97% said they were still preparing for agent AI.
The report also suggests that control points for AI systems are moving from just infrastructure to prompts, tokens, and application programming interfaces. In Australia and New Zealand, 31 percent of organizations identified the prompt layer as their primary delivery mechanism, and 24 percent said the token layer was their priority for delivery and security.
This shift is important because it changes where companies need to apply surveillance. As AI systems become embedded in customer service, internal tools, and automated decision-making, companies face tremendous pressure to control who has access to their systems, how requests are processed, and what safeguards are in place.
regional lag
Low levels of self-performing reasoning in Australia and New Zealand may reflect a combination of cost constraints, skills shortages, and slow implementation cycles. At the same time, the data shows that local organizations are not standing still, given their advanced readiness for agent AI and strong adoption of hybrid and multicloud infrastructure.
Regional numbers show that while the market is embracing AI, it is still not moving at the same pace as its global peers in directly operationalizing inference. For vendors, cloud providers, and enterprise technology teams, the focus is likely to be on deployment models that reduce complexity while meeting governance and security requirements.
For business leaders, the broader message is that AI is increasingly being treated as part of core operations rather than a separate innovation track. As organizations distribute applications across public clouds, on-premises systems, and colocation sites, AI deployments are becoming more closely tied to network management, policy controls, and cyber security.
Based on our findings, a combination of AI adoption and hybrid infrastructure appears to be the way to go. Globally, only 8% of organizations currently rely solely on public AI services.
