88% of Hospital IT Can't Handle the AI It's Already Using

Australian hospitals are racing to adopt artificial intelligence while running on infrastructure that can't keep up. The 2026 Nutanix Healthcare Enterprise Cloud Index survey quantifies a sector trying to scale AI faster than its technology foundations can safely support. When 88 percent of healthcare IT leaders report their on-premises infrastructure cannot fully support AI workloads, and nearly four in five have already encountered unsanctioned AI tools deployed by clinical staff, the system faces both an innovation crisis and a patient safety risk

Intensive care units now generate up to 20 connected devices per bed and 15 terabytes of data annually per patient room. Latency, once a performance metric, is now a clinical risk. The survey found that 79 percent of healthcare organisations have encountered AI applications deployed outside IT oversight, and 83 percent acknowledge these unsanctioned tools create business risk. This is shadow AI at clinical speed: clinician burnout and legitimate workflow pressure colliding with organisations that lack approved alternatives

Department silos deepen the problem. Eighty-three percent of IT leaders reported that gaps between clinical departments and technology teams make initiatives harder to execute. Both groups must align to ensure bedside AI tools meet privacy, safety, and compliance standards. Australian data sovereignty requirements compound the pressure: 72 percent of organisations now prioritise sovereignty in infrastructure decisions, and 54 percent said regulatory expectations and public trust require keeping infrastructure within the country

Running secure, low-latency AI inference locally demands infrastructure most Australian health organisations do not yet possess. Containerisation offers a practical path forward; 86 percent of surveyed organisations said AI is accelerating container adoption, and 80 percent are building new applications in containerised environments. By packaging workloads in containers, hospitals can run AI closer to where data is generated while keeping sensitive information local. Without coordinated investment in governed, unified infrastructure, the gap between adoption velocity and technology readiness will continue to widen

The case for urgent, coordinated infrastructure investment has never been clearer