Most organisations adopting AI will never build the infrastructure that powers it. The compute behind the tools their people use sits with hyperscalers and cloud providers, consumed as a service. Gartner expects worldwide AI spending to reach US$2.59 trillion in 2026, with more than 45% going to infrastructure, almost all of it built by vendors and cloud providers rather than by the businesses using AI.
That one fact should change how business leaders in Australia think about AI. If the heavy lifting happens elsewhere, AI adoption should not automatically trigger a major hardware refresh.
And yet, the rush to embrace AI has created no shortage of technology decisions. New tools arrive weekly, investment is climbing and boards are under pressure to show they have a strategy. In that climate the conversation slides quickly from adopting AI to buying things: software, infrastructure, devices.
AI is not a procurement exercise. It is a business transformation exercise, and the organisations getting the most from it are not the ones buying the most technology. They are the ones making better technology decisions. Before buying anything, leaders should ask three questions, in order:
What business outcome are we trying to improve?
Where should that workload run: cloud, data centre, edge or endpoint?
What technology genuinely needs to change?
The first keeps the focus on the business. The second determines what technology, if any, is required. The third stops AI adoption becoming an automatic reason to replace devices. Taken together, these questions shift the conversation away from buying technology and towards making smarter decisions about where technology can create the greatest value.
Start with the business outcome, not the hardware
AI is a business transformation decision before it is a technology decision. The question is not “where can we use AI?” but “which decision, process or customer experience do we want to improve, and how will we know?” Only once that is clear does the second question earn an answer. For most organisations, most of the time, that answer is “in the cloud”, which means much of the existing device fleet is already fit for purpose.
There are exceptions. Employees handling highly sensitive data, working in low-connectivity environments or running graphics-intensive engineering and creative workloads may benefit from on-device AI capability. These are targeted use cases and they call for targeted refreshes, not a fleet-wide program justified by a need that applies to a fraction of the workforce.
Even when organisations identify a compelling AI use case, technology is rarely the primary barrier to adoption.
The real constraint is capability, not hardware
More often than not, the obstacle to AI value is not the device on the desk, but the confidence of the person using it. AWS research found businesses report that only 28% of employees feel confident using advanced AI tools. That is a capability gap, and a new laptop does not close it. What closes it is training, change management and clear governance: which tools are sanctioned, what data can be shared and how outputs are validated.
This matters for budgets. Money spent on a premature fleet-wide refresh cannot also be spent on the training and governance that actually drive adoption. Knowing the condition and performance of existing assets lets leaders tell the difference between where technology is holding people back and where the greater need is capability.
That distinction becomes even more important as AI begins reshaping the economics of technology investment itself.
AI is changing the economics of the technology estate
One of the stranger consequences of the AI boom is that it’s making ordinary technology more expensive. AI driven demand for memory is diverting manufacturing capacity away from the DRAM and storage that go into everyday PCs and servers. Gartner forecasts memory prices will rise around 130% by the end of 2026, pushing average PC prices up 17% on 2025, and expects business PC lifetimes to stretch by 15% as organisations hold on to devices for longer. AI PCs are not expected to reach half the market until 2028.
The implication is the opposite of the popular narrative. For most organisations, AI is extending refresh cycles, not accelerating them. Devices cost more, stay in service longer, and refresh decisions become more selective, with a handful of roles getting new AI capable hardware while everyone else’s equipment keeps working.
As devices stay in service longer and refresh decisions become more selective, understanding the condition, performance and value of existing assets becomes increasingly important.
That is where asset lifecycle management stops being IT housekeeping and becomes strategically relevant. When assets are expensive and long lived, knowing what you own, where it is, how it is performing and what useful life remains is the foundation of every good technology decision. For every asset, the outcome should be a conscious choice: maintain and optimise it, redeploy it where it is needed, refresh it selectively, retire it securely, or recover the residual value in it.
Risk, value and sustainability
These decisions are no longer purely operational. Increasingly, they have direct implications for risk, security and governance.
Longer device lives and more internal movement raise the stakes on those choices. AI is increasing the volume of sensitive information passing through endpoints. Prompts, proprietary documents, customer data and generated outputs all move through devices that may remain in service longer than planned and change hands more often. Visibility, chain of custody and certified data sanitisation when data bearing equipment is retired are no longer back office hygiene. They are conditions of responsible AI adoption.
The same discipline recovers value. Redeploying a suitable device instead of buying a new one, extending useful life before replacing, and remarketing displaced equipment can offset part of the cost of transformation at a time when hardware quotes are arriving above budget. Refurbishment and responsible recycling at end of life then turn the lifecycle into a practical sustainability outcome. Organisations have little control over the energy consumed by hyperscale AI infrastructure. They have complete control over their own estates, and extending device life is one of the most effective ways to cut emissions and e waste, with no separate sustainability initiative required.
Decide first, buy last
The organisations that get the most from AI will not be those that bought the most hardware, nor those that froze investment altogether. They will be the ones that can answer three questions for every AI initiative: which outcome will it improve, where will it run, and what genuinely needs to change.
Asset lifecycle management is what allows leaders to answer that third question with confidence. Sometimes the right answer will be extensive change. Often, it will be very little. Neither outcome is a missed opportunity. Both reflect technology decisions grounded in business need rather than assumptions about refresh cycles.
Ultimately, good AI strategy is not measured by how much technology is purchased, but by how effectively technology decisions align with business outcomes. AI should drive better lifecycle decisions, not automatic refresh decisions. That is why asset lifecycle management is becoming more important than ever.