Company Interview / Are Aussie businesses paying a hidden AI tax?

Key Points:

Hidden “AI tax” seen in over‑provisioned infrastructure, token overspend and centralised architectures Flexibility, cloud portability and clear exit strategies viewed as essential for AI investment Australia’s infrastructure seen as too city‑centric for nationwide and global AI ambitions

Jay Jenkins from Akamai Technologies states that many Australian organisations are paying a “hidden AI tax”, as they invest heavily in models while neglecting the underlying infrastructure. Jenkins argues that businesses often over‑provision cloud capacity, overspend on tokens and measure success by AI spend rather than outcomes. Centralised architectures built over decades are viewed as ill‑suited to an AI era that requires real‑time responsiveness where data and users actually reside, not just in data centres.

Jenkins frames flexibility as critical in a fast‑moving AI landscape. Workloads should be portable across clouds, talent should not be locked into a single hardware or software stack, and leaders should plan clear exit strategies from vertically integrated platforms. Australia’s AI adoption is described as rapid and relatively mature, yet infrastructure is seen as concentrated in Sydney and Melbourne while users and industries are dispersed nationwide and globally, raising concerns over latency, responsiveness and egress costs.

Jenkins highlights Akamai’s long-standing distributed edge network and its deployment of Nvidia Blackwell GPUs, alongside open technologies to support multiple vendors and providers. Jenkins expects a shift from large foundation models to smaller, domain‑specific language models, orchestrated as multi‑agent systems to deliver better value at lower cost.