New Zealand requires additional technological infrastructure to enable the full operational capabilities of the Heidi medical AI scribe tool [1, 2].

This shortage creates a barrier for healthcare providers seeking to integrate advanced artificial intelligence into clinical workflows. Without the necessary backend support, the ability to process and deliver real-time medical data at scale remains limited, potentially delaying the efficiency gains promised by AI scribes.

Heidi, the company behind the medical AI scribe tool, said that some capabilities required to run the system at full capacity are not yet available [1, 2]. The tool is designed to assist clinicians by documenting patient encounters, but its performance depends on the availability of high-performance computing and data centers.

Datagrid, a Singapore-based firm, has been identified in the context of these infrastructure needs as the region grapples with a global boom in artificial intelligence [1, 2]. The race to secure AI infrastructure has become a priority for nations looking to maintain technological competitiveness.

Industry experts said that as AI continues to advance, infrastructure must evolve to enable access and delivery of real-time information at scale [3]. This evolution involves the construction of specialized data centers capable of handling the intense processing requirements of large language models, and real-time audio transcription.

Currently, the gap between software capabilities and physical hardware in New Zealand prevents the full rollout of the scribe's features. This limitation highlights a broader challenge where AI software development is outpacing the physical build-out of the data centers required to host them [1, 2].

"Some capabilities required to run Heidi at full capacity are not yet available,"

The situation in New Zealand illustrates a growing global tension between the rapid deployment of AI software and the slower pace of physical infrastructure development. For the healthcare sector, this means that even when advanced tools are available, their utility is capped by the local availability of high-performance computing and data residency capabilities.