Through my 20-year career working in the datacentre industry, I’ve seen infrastructure priorities change many times. Looking back, I can’t remember another period where the conversations shifted as quickly as they have over the last few years.
My discussions with customers, technology vendors and integration partners used to start with technical questions. Which server platform should we deploy? Which storage platform should we standardise on? How should we approach our next infrastructure refresh?
Today, the questions are more likely to focus on reducing operating costs, improving cyber resilience, meeting sovereignty requirements and preparing for AI. Technology remains part of the conversation, but it’s rarely the starting point.
Organisations are looking at infrastructure through a broader business lens, assessing how technology decisions affect cost, risk, resilience and their ability to support future initiatives. As a result, decisions that once sat largely within IT teams are now drawing interest from finance, security, risk and executive leadership.
I see several pressures driving that shift, and they’re rarely separate conversations anymore. Whether it’s managing cloud consumption, meeting sovereignty requirements or preparing for AI, organisations are being asked to make technology decisions with greater discipline and accountability than ever before. In this blog, I’ll explore the key pressures shaping infrastructure decisions today, and why organisations need to consider them together rather than in isolation.
Refresh cycles, licensing changes and growing data volumes are forcing organisations to weigh platform choices against long-term operating costs, rather than treating them as separate conversations. That’s showing up as a genuine capex versus opex question at the board level, not just a procurement detail. AI adds a new layer on top of that, with GPU infrastructure, platform readiness and token consumption starting to appear as their own line items.
Cloud has given organisations flexibility they didn’t have before, and most have taken advantage of it. The trade-off is that consumption-based spending is hard to forecast once it’s running across multiple platforms without a consistent view of what’s being used. I see this most often in organisations that moved to the cloud for agility and only started asking about cost governance after the first few invoices arrived. Getting ahead of that requires visibility into consumption before it becomes a budget problem, rather than after.
For government, federal and health customers, sovereignty isn’t an abstract principle. Trusted organisations increasingly want their data closer to the source rather than spread across global infrastructure, and that expectation is shaping procurement decisions well before anyone talks about specific technology. Cyber resilience and sovereignty now sit alongside cost as things a CIO has to answer for directly, and in regulated industries, they’re often the first questions asked.
A lot of organisations jump straight to models, copilots and agents. In my experience, those who realise real value start earlier are the ones backed by secure infrastructure, governed data and operational and financial discipline. Get that foundation wrong, and AI initiatives struggle to scale no matter how good the use case looks on paper.
I think of it as a shift we’ve already lived through once. A decade ago, the risk was Hypervisor sprawl, with environments growing faster than anyone could track. We’re heading towards the same pattern with AI tokens if organisations don’t apply the same discipline to consumption early. It’s a familiar problem wearing a new name.
There’s one idea I keep coming back to. When you build your AI capability within your own infrastructure, you own the intellectual property that comes out of it, rather than renting it via someone else’s platform and subscription. That’s a different economic position to be in, and it’s one more organisations are starting to ask about.
One of the more common mistakes I see is organisations committing significant budget to a platform based on an assumption rather than a tested one. Before that kind of investment goes ahead, it’s worth validating workload suitability, platform strategy, security requirements and the operational impact of the change against the business outcome it’s meant to deliver, not just the technology spec sheet. A proof-of-value step earlier in the process costs far less than discovering a mismatch after the platform is live.
These are all reasons to be more deliberate in your decision-making. The organisations getting this right are asking how to modernise, reduce costs, improve cyber resilience and create the capacity to invest in AI all at once, without solving one problem at the expense of another.
I’ll unpack this further at the Dell Technologies Forum in Sydney on August 11 2026. If you want to discuss where your organisation sits on infrastructure cost, sovereignty or AI readiness, come find me at the stand or register to attend the full session. If you can’t attend, reach out to us via the form below or visit our Dell Technologies page to learn more.
Speak to our team of Specialists today
Information provided within this form will be handled in accordance with our privacy statement.