The quarterly earnings calls are in, the analyst reports are published, and we now have a clearer picture of where enterprise AI spending actually went in Q3 2025. The patterns are instructive.

The Big Picture

Global enterprise AI spending grew 23% year-over-year in Q3 2025. That sounds impressive until you remember projections earlier this year called for 40%+ growth. The market is growing, but the explosive growth phase has moderated.

Australian enterprises followed global patterns with some local variations. Total AI-related technology spending increased approximately 18% compared to Q3 2024, with concentration in specific categories.

Where Money Flowed

Productivity AI Tools

Microsoft Copilot and similar productivity tools captured the largest share of new spending. This isn’t surprising – it’s the path of least resistance. Buy licenses, roll out to employees, check the “doing AI” box.

Australian enterprises spent approximately $340 million on productivity AI licensing in Q3 2025. That’s real money, though the per-seat costs are declining as competition increases.

The question remaining: how much value are organisations actually extracting from these investments? Early data suggests significant variation – some organisations seeing meaningful productivity gains, others essentially paying for expensive email filters.

Cloud AI Services

AWS, Azure, and Google Cloud AI services grew substantially as enterprises moved from proof-of-concept to production deployments. Consumption-based AI services (inference APIs, managed ML services) showed particularly strong growth.

This reflects a healthy maturation: organisations moving beyond experimentation into sustained production use. The infrastructure spending suggests real applications, not just innovation theatre.

Custom Development

Custom AI development spending grew more slowly than many predicted. Organisations are being selective, funding specific high-value use cases rather than broad AI development programs.

This selectivity is rational. Custom development is expensive, risky, and requires capabilities many organisations lack. The bar for “build rather than buy” has risen.

Where Money Pulled Back

Experimental Initiatives

The “innovation lab” approach to AI faced budget scrutiny. Exploratory projects without clear paths to production value saw funding reduced or eliminated.

This isn’t necessarily bad. Some experimental work was genuinely exploratory and valuable. But much was unfocused activity that generated interesting demonstrations but no business value.

Consulting Spend

AI consulting spend declined from H1 2025 peaks. Organisations that completed strategy work are now in execution mode, requiring less external advisory support.

The exception: implementation-focused consulting. Hands-on help building and deploying AI systems remains in demand. Pure strategy work is harder to sell.

Generic AI Platforms

Generic “AI platform” purchases declined as organisations realised they needed specific solutions, not general-purpose capabilities. The platforms that succeeded were those with strong vertical or use-case focus.

Industry Patterns

Financial Services: Continued strong investment, particularly in fraud detection, risk modelling, and customer service automation. Regulatory compliance AI also growing as institutions prepare for increased scrutiny.

Mining and Resources: Operational AI investment steady. Predictive maintenance and optimisation applications dominate. Less appetite for customer-facing AI given business models.

Retail: Demand forecasting and inventory optimisation remained priority areas. Customer personalisation investment moderated after disappointing results from earlier initiatives.

Healthcare: Cautious growth. Administrative AI (scheduling, billing, documentation) attracted more investment than clinical AI, reflecting risk profiles.

What This Tells Us

Several patterns emerge from the Q3 data:

Consolidation around proven use cases. Organisations are focusing spending on applications with demonstrated ROI rather than exploring broadly. This is mature behaviour.

Buy over build. The custom development surge many predicted hasn’t materialised. Organisations prefer commercial solutions where available, building only when necessary.

Productivity tools dominate. For better or worse, licensing productivity AI is the default enterprise AI investment. Whether this generates proportional value remains to be seen.

Regional variation matters. Australian enterprises are following global trends but with local characteristics – more caution around data sovereignty, different industry mix, smaller absolute numbers.

Budget Implications

For organisations planning Q4 2025 and 2026 budgets:

Expect continued productivity tool investment. Copilot and equivalents will remain the largest category. Focus on maximising value from these investments through training and change management.

Cloud AI services will grow. As applications move to production, consumption-based AI services become recurring costs. Budget accordingly.

Custom development requires strong justification. Unless you have a compelling competitive advantage case, buy before you build.

Don’t forget total cost. AI spending isn’t just vendor fees. Include internal resources, change management, and ongoing operational costs in your planning.

Final Thought

Q3 2025 spending patterns suggest enterprise AI is maturing. The hype-driven investment of 2023-2024 has given way to more considered allocation of resources toward proven applications.

This is healthy. Markets that mature spend more wisely. The question for 2026 is whether the value realised justifies the investment made. That verdict remains pending.