65% of Australian Firms See AI Returns — But Readiness Is Falling

Sixty-five percent of Australian business leaders report positive returns on AI and machine learning investments, up eight percentage points year over year — while their self-assessed readiness to manage external risks fell ten points to 23%. That contradiction sits at the centre of Kyndryl's latest Readiness Report, published November 11, 2026, and based on responses from 200 senior business leaders in Australia. The numbers describe companies earning money from AI while dismantling their own confidence in handling what AI brings.

The stakes are straightforward. Australia's AI adoption has moved past the experimentation stage — 89% of surveyed organisations now run AI in workflows beyond pilot projects. Returns and risk management are diverging at exactly the moment AI becomes embedded in everyday operations. The report's most useful finding may be what it reveals about how uneven that embedding actually is.

What does the readiness gap look like in numbers?

The decline in preparedness is broad and, in two categories, steep. Only 18% of Australian leaders said they felt completely ready to manage regulatory change — a drop of 18 points from the previous year. Readiness for competitive disruption fell 16 points to 22%. Readiness to manage external risks overall averaged 23%, down from 33%.

The adoption side of the ledger is noisier than headline returns suggest:

• 28% of respondents reported a failed AI investment.
• 25% said they had cancelled a major AI initiative.
• 66% said they had more AI and automation pilots than they could scale.
• 56% said they had substantially changed their AI strategies over the past year.

Two-thirds of organisations sitting on pilots they cannot scale is not a picture of mature deployment, whatever the 89% adoption figure implies. The failed-investment and cancellation rates, nearly a third of the sample combined, go unmentioned in the framing of "stronger returns" — a gap worth noting.

Martin Schroeter, Chairman and Chief Executive Officer of Kyndryl, described the pattern as a structural problem rather than a temporary one. "AI ambitions are moving faster than the systems needed to support them, and the bar for readiness keeps moving as AI becomes more integrated in enterprises," he said. "As AI creates significant opportunity, organisations are now focusing on how to make early returns sustainable." His prescription — "focusing on people, technology and governance" — is directionally sensible but thin on specifics. What those systems should consist of, the report does not say.

Why is cybersecurity the sharpest pressure point?

Security shows the widest divergence between Australian and global respondents. Some 61% of Australian leaders said cybersecurity risks were growing faster than their organisation's ability to address them, against a global average of 52%. Confidence collapsed in parallel: the share saying they were completely ready to manage cyber risk dropped 13 points, from 41% to 28%.

Yet the same respondents credit AI with defensive value. More than four in five — 83% — said AI helps their organisations identify threats faster. That figure coexists with declining confidence in their ability to respond to cyberattacks, which suggests firms are buying detection speed without building response capacity. Faster alarms, same fire brigade.

Concern extends to AI itself. Two-thirds of Australian respondents, 66%, said they worried about AI system malfunction or unintended error. Only 20% said they were completely ready to manage that risk. An organisation can fear its own tools and still lack the governance to control them; that is the report's implicit admission.

Is the infrastructure actually ready for AI at scale?

The short answer, by the respondents' own account, is no. While 65% of Australian leaders described their core IT estate as mostly or fully modernised, 64% said their infrastructure still required significant re-architecting before it could support AI at scale. Modernisation, in other words, is not the same qualification as AI readiness — a distinction many vendors blur.

Technical debt (the accumulated cost of ageing systems and shortcuts in code and architecture) is the named culprit. In Australia, 49% of leaders said technical debt was holding them back, above the global figure of 39%. One easing constraint did appear: the share of Australian leaders saying innovation efforts were delayed by weaknesses in their technology stack fell 12 points to 51%. That improvement is real but modest — half the sample still reports delays.

How far behind is the workforce?

Further behind than the infrastructure. Only 26% of Australian respondents said their workforce was completely ready to use AI. Just 29% said their organisations had formally updated roles and responsibilities around AI. Change management (the structured process of preparing staff for operational change) looked weaker still: a bare 22% said their organisation had fully implemented it for AI adoption.

Graeme Beardsell, Managing Director of Kyndryl Australia and New Zealand, framed the workforce and governance gaps as the binding constraint on the next phase. "Australian organisations are seeing encouraging returns from AI in some areas, but the findings also show how much work is still required to support AI transformation at scale," he said. "Capturing it sustainably will depend on building the resilience, readiness and security to support AI as it becomes part of everyday operations." Kyndryl sells the services that close such gaps, and the report functions partly as a market argument — a caveat readers should weigh against its data.

What are firms willing to pay for control?

The survey also tested how much sovereignty Australian organisations want over their digital operations. Some 64% said they would pay a premium of 10% or more for solutions offering full operational control if needed. Another 65% said greater control was worth slower deployment and slower adoption of advanced technology. For cloud and infrastructure vendors, that is a signal: a majority of Australian buyers will trade speed for control — assuming, as with every figure here, that stated willingness matches actual procurement behaviour.

The Australian results come from a broader Kyndryl study of 3,700 senior business leaders across 21 countries, which gives the local sample context but also means the Australian figures rest on 200 respondents — self-reported, unweighted for company size, and drawn by a vendor with a commercial interest in the readiness problem it documents.

The pattern to watch is whether next year's edition shows readiness recovering as governance and change-management investments mature, or whether returns keep outrunning the systems meant to contain them — the second year of that divergence would be harder to dismiss as growing pains.