When I was brainstorming ideas for this piece as Data#3’s new Dell Technology-aligned EUC specialist, one question kept surfacing: If everyone is using AI on their devices, what real benefits are we seeing in return? That gap between activity and return is what I want to talk about, as well as how we can unlock a genuine business advantage for organisations that know how to use it and use it well.
The reality for most Australian businesses, though, is that turning AI adoption into measurable business value still feels a little out of reach. The challenge is turning growing adoption into something secure, manageable and capable of delivering a measurable return. In my experience, that’s where the endpoint starts to become much more important than many organisations realise.
Most of the IT and end-user compute leaders I meet are under pressure to do something with AI. Some of that pressure comes from the top, and some from staff who already use these tools at home and now expect them at work. Yet, in Deloitte’s 2026 State of AI in Enterprise report, only 65% of Australian respondents plan to increase AI investment in the next financial year, compared to a global average of 84%. Alarmingly, just 12% of respondents report that generative AI is already transforming their business and industry.
The truth is that adoption and productivity are not the same thing, and while many organisations have AI in the building, few can point to a clear, repeatable business return.
The organisations that pull ahead will be the ones that can operationalise AI, not simply access it. Turning scattered, individual use into dependable work is a harder job than switching a tool on, and it is the job worth focusing on.
This is where end-user hardware becomes part of the conversation. Much of early AI adoption, especially among developers, relies on born-in-the-cloud tools. That works, yet it carries a running cost that is hard to predict and even harder to cap. Cloud AI spend can climb quickly once a team moves from the odd prompt to always-on agents. For a developer cohort, this raises a fair question – how much of this work truly needs to sit in the cloud, and how much could run at the edge closer to the person doing it?
Modern Dell hardware is built with that question in mind. The neural processing units (NPUs) in their current AI PCs are designed to run AI workloads on the device itself. Your AI strategy is only as strong as the infrastructure beneath it, so where that processing happens matters more than people assume.
Moving more of it to the endpoint can make costs more predictable, keep sensitive data local and maintain performance even when the network is not. A developer cohort is a sensible place to prove that before deciding what belongs on the device across the wider business.
As soon as AI starts running locally, questions about governance are never far behind. If a model runs on a device, does it slip outside the governance and controls that IT has spent years building? Cloud AI tools send prompts and data to services you do not own, which is the harder flow to see and govern. Keeping that processing on the endpoint keeps the data inside your environment and where your existing controls already reach.
A smart security posture already spans identity, applications and the endpoint, and governing AI draws on those same levers. Identity governs who and what can call a model, application controls determine which AI tools are approved and the endpoint sets the boundaries for local processing. None of that is new ground for a well-run IT team.
On Dell hardware, it’s taken a step further. The Dell Trusted Device application feeds PC security telemetry directly into tools such as CrowdStrike’s Falcon console, so security teams can monitor device integrity from the same place they monitor everything else. Results from Dell’s off-host BIOS verification, which checks the BIOS against a known-good image stored securely off the device, appear directly in that console. Paired with Intel silicon-level security, the protection runs from the chip up to the cloud. An endpoint that performs local AI work sits within that view, rather than becoming a blind spot.
That’s how an organisation gets meaningful AI impact while still meeting the governance, security and budget requirements already in place.
A common concern I hear is that giving people this kind of flexibility weakens security, but I see it the other way around. A smart security posture across identity, applications and the endpoint lets an organisation do both at once. People get to work the way they want to work, whether that be from the office, from home or from a cafe, while the business holds a high security bar. The choice between flexibility and control is a false one, but if you get the foundations right, nobody has to pick a side.
Applications and how people actually do their work sit at the centre of all of this for me. Hardware, security and management only earn their keep when they let people get on with the job.
What most of us are learning is that the hardest step in AI is moving from a promising pilot to something that runs reliably at scale every day. That journey from experimentation to production is where AI starts to create real value, and it’s also where most efforts stall. Innovation accelerates when technology stops getting in the way of the outcome and removing that friction is exactly where we can help.
Data#3 does more than ship a device and wish you well. Our broader end-user compute practice is built around a range of deployment, management, adoption and lifecycle planning services including:
If your team is grappling with the gap between using AI and realising value from it, don’t wait to have the conversation. Reach out to the Data #3 team to see how we can help.
Speak to our team of End User Compute Specialists today
Information provided within this form will be handled in accordance with our privacy statement.