There’s been a lot of handwringing in tech media this month about enterprise AI spending hitting a plateau. Forrester’s latest quarterly data shows year-over-year growth in enterprise AI budgets dropped to single digits for the first time since 2023. Headlines are calling it a “correction.” Some are calling it worse.

I think they’re reading it wrong.

After sitting across the table from CIOs and CFOs in Melbourne and Sydney for the better part of two years, I can tell you that the spending slowdown isn’t a sign of disillusionment. It’s a sign of maturity. And honestly, it’s overdue.

The Binge Phase Is Over

Let’s be real about what happened between 2024 and mid-2025. Enterprises threw money at AI with the same discipline a teenager shows at an all-you-can-eat buffet. Proof-of-concept budgets ballooned. Vendor contracts were signed with minimal due diligence. “AI strategy” often meant “buy whatever the CTO’s favourite podcast recommended.”

The numbers bear this out. According to McKinsey’s latest State of AI report, only about 26% of organisations have moved even one AI use case into full production. That means three-quarters of the money spent so far is still sitting in sandbox environments, pilot programs, and PowerPoint decks.

So when budgets flatten, what we’re really seeing is organisations pausing to figure out what actually worked.

What the Smart Money Is Doing

The enterprises I work with that are spending more wisely aren’t cutting AI budgets. They’re reallocating them. Here’s what that looks like in practice:

From broad experimentation to targeted scaling. Instead of funding fifteen pilots, they’re picking the two or three that showed genuine ROI and investing in production-grade infrastructure. One financial services client in Melbourne recently consolidated seven experimental AI projects into two production deployments, and their per-project spend actually went up. But total AI spend stayed flat.

From vendor subscriptions to internal capability. There’s a growing realisation that paying $50 per seat per month for AI tools only makes sense if people actually know how to use them. I’ve seen multiple organisations redirect 20-30% of their tool budgets toward training and change management. Team400 has been doing interesting work in this area, helping enterprise teams build internal AI capability rather than just stacking more vendor contracts.

From general-purpose models to domain-specific solutions. The era of “just plug in GPT-4 and see what happens” is winding down. Enterprises are investing in fine-tuned models, RAG architectures built on their own data, and custom workflows that actually fit their processes.

The CFO Perspective Matters More Now

Here’s something that doesn’t get enough airtime: CFOs are finally asking the right questions about AI investments. Not “how much should we spend?” but “what did we get for what we spent?”

That shift changes everything. When the CFO is asking for ROI evidence rather than just rubber-stamping innovation budgets, the quality of AI projects improves dramatically. Bad projects die faster. Good projects get the support they need.

I spoke with a CFO at a mid-tier Australian insurer last month who put it bluntly: “We spent $4.2 million on AI in 2025. I can point to maybe $1.8 million in identifiable value. That doesn’t mean AI is bad. It means we need to be better at picking projects.”

That’s exactly the right takeaway.

What This Means for 2026

If you’re an enterprise leader watching your AI budget flatten, don’t panic. Instead, use this moment to do three things:

1. Audit your existing investments ruthlessly. Kill the zombies. Every organisation has AI projects that are technically “active” but haven’t produced meaningful results in months. Shut them down and redirect the budget.

2. Invest in measurement infrastructure. You can’t prove AI ROI if you didn’t set up proper baselines before deployment. This is the single most common gap I see, and it’s worth fixing now.

3. Focus on adoption, not just deployment. A model in production that nobody uses is worse than no model at all, because you’re paying for infrastructure without getting value.

The Bottom Line

The AI spending plateau isn’t a crisis. It’s a course correction. The organisations that treat it as an opportunity to sharpen their approach will come out ahead. The ones that panic and either slash budgets entirely or double down on unfocused spending will struggle.

Maturity doesn’t make for exciting headlines, but it makes for better outcomes. I’d take a plateau over a bubble any day of the week.

Sarah Chen is a Melbourne-based enterprise consultant specialising in AI strategy and digital transformation.