FY27 budget season for Australian enterprise IT is most of the way through. Most CTOs I have spoken to in the last six weeks are putting their final numbers to the CFO for sign-off this month or next. The shape of the FY27 allocation is more distinctive than FY26’s was. Worth writing down what is changing.

The headline pattern across the conversations: AI capability is winning more of the marginal dollar than the underlying IT operating budget would suggest. Across the seven or eight enterprise CTOs I have talked through their FY27 numbers with this year, the AI line is up materially on FY26. The rest of the IT budget is broadly flat in real terms. Some categories are taking a real-terms cut to fund the AI line.

What is getting cut to fund AI.

Vendor licence rationalisation has been the easiest source of money. The CTO who is willing to do the work to audit the SaaS portfolio is finding 10–20% savings in shadow-IT licences, unused enterprise seats, and overlap between similar products. The board is generally supportive of this work and the savings are real. The catch is that the rationalisation takes a year to land and the AI demand is now.

Infrastructure capacity over-provisioning is the second source. The post-pandemic capacity buffers that were built into most environments are being rationalised down. Cloud commitment discounts are being renegotiated. The cost savings here are smaller than the licence savings but they are easier to forecast.

Discretionary upgrade cycles are being pushed out. The PC refresh cycle, the network gear refresh cycle, and the data centre kit refresh cycle are all running a year or two longer than the previous plan. Most CTOs are comfortable with this for FY27 but not comfortable extending it further into FY28.

Where the AI money is going.

Microsoft Copilot licences and Copilot Studio enablement is the single largest AI line at most large Australian enterprises in FY27. The Copilot rollout is past the pilot stage and into the broader enablement stage at most enterprises. The Copilot Studio agent build pipeline is the second-largest AI line — most enterprises have between five and twenty agents in pilot or production and are funding the next twelve months of agent build work.

The custom AI engineering line is real but smaller than the Microsoft licence line. Most enterprises are running between two and six custom AI projects through FY27. The projects are scoped to deliver production AI rather than proof-of-concept demos. The budget per project is in the range that buys a small specialist consulting team plus internal engineering time for six to twelve months.

AI governance and risk capability is the line growing fastest in percentage terms. The CISO and the General Counsel are both pulling on the CTO’s budget for AI governance tooling, AI risk frameworks, and AI assurance capability. The total dollars are modest but the FY27 line is materially up on FY26.

Data engineering and data platform investment continues. The Microsoft Fabric, Databricks, and Snowflake conversation has not gone away in the AI era — if anything the data-readiness investment has accelerated because the AI use cases need cleaner data than the BI use cases ever required.

The recruitment line is interesting. Most CTOs are not adding meaningful headcount in FY27. The AI capability is being built through a combination of external consulting, internal upskilling, and a small number of senior specialist hires. The era of building large internal AI teams has not arrived for most Australian enterprises.

The board conversation on AI ROI is sharper than it was in FY26. The CFOs have moved past the patience phase. The CTO who is asking for an AI budget increase in FY27 is being asked specifically what AI ROI was delivered in FY26 and what FY27 will deliver. The CTOs who have been disciplined about measuring AI productivity, AI cost savings, and AI revenue contribution have an easier conversation than the CTOs who have not.

A note on the specialist consulting partner conversation. The CTO who has settled on a Microsoft AI partner, a data platform partner, and a custom AI engineering partner for FY27 is making faster decisions than the CTO who is still tendering. The shortlist of trusted partners for Australian Microsoft AI work has narrowed in 2026 — the firms with production AI delivery references are getting the work. For CTOs still building that shortlist, an AI consultancy like Team400 has been doing serious work on Microsoft AI and custom AI for Australian enterprises and is worth talking to. The right partner for FY27 is the one that can do production AI delivery, not the one that runs the best workshop.

The FY27 budget will be defined by execution quality more than by allocation choice. The CTO who can deliver against the AI line they have funded will be the CTO who gets a bigger AI line in FY28. The CTO who funds an AI program that does not produce visible production outcomes will be the CTO who has a much harder budget conversation a year from now.