- AI
- Danoffice IT
- Cloud & Data Platform
- HPE
- Article
The possibilities with AI are not up for debate. The old saying, the future is already here, seems more present than ever.
However, questions and paradoxes are lining up side-by-side – especially on the discussion on the very foundation of IT: Our infrastructure. Are there any value in having your own infrastructure, on-premises, in the work with AI? And is it even possible to achieve the full impact with it? Absolutely, says AI-specialist. To both scenarios.
“We are often helping manufacturing companies, and they have IT everywhere in their production. There are, however, many things that are not online, and which should never be. So-called air-gapped infrastructure. But maybe these companies wish to operate AI in their production with either robots, Computer Vision or with a critical language model-system. However, the company wishes to be able to trust its system 100 % and 24/7, and in these situations, solutions with no dependency on the cloud have extreme value. Here, HPE has presented an entire portfolio of on-premises solutions, which are geared to work with AI in these private-private-environments”.
Leif Elgaard Høj, Head of AI & Development i Danoffice IT
Extreme demands
If you thought time had healed the mythical conflict between the IT-department and “corporate operations”, then you are wrong. According to Leif, AI has lit the fire even more. “Operations typically expect that AI is something you take down from the shelve and install it, plug & play. There is very little understanding of the fact that, if they want an AI-solutions that crunches its tasks in milliseconds, then it takes massive infrastructure”, he says and continues: “If you choose to install an AI-solution on your own infrastructure, then the demand goes up even more”.
In other words, there is a need for solid infrastructure. Operations are making demands – so is the AI itself. “HPE’s 2024-launch, PCAI – Private Cloud AI, is a turn-key solution for companies wanting a disconnected cloud, on-premises, in their production, which can run very complex language models – without plugging in the cable to the big internet”, he says.
Sizing 2.0
With the new demands, a new big challenge arises, which all manufacturers of IT-equipment seek to conquer. “What we are all working on right now, is to translate AI-demands to infrastructure-demands. Meaning adapting the actual demand for actual infrastructure so you don’t break the IT-spine of the company”, Leif says and states that the pace of the development brings yet another obstacle with it. “There is a newfound anxiety of getting behind in the race. Everything renews itself every other month, so a lot of companies buys too much tech. Over capacity, and over-powering, which might harm the IT-environment’s performance overall”.
The flexibility, which AI demands, is answered by HPE with their infrastructure as-a-service. Leif points out that the entire portfolio is available to you and it brings you the answers. Traditional infrastructure-specialists might think that heavy machinery is needed when upscaling for AI. But infrastructure of the future is in fact more about micro-segmentation of your compute power. Few companies can realize the full potential of a Nvidia H200 GPU anyway, so there you would micro-segment it and create smaller, virtual GPU’s instead. Scalability today is everything – or else you end up with unused hardware”.
A utopia
Answers are the core of the AI-development. ”If you, today, think that your infrastructure specialist can speak fluidly about demands for language models in the development of your R&D division and their AI-needs, then you are wrong. It is a utopia. In that perspective, HPE’s PCAI is a possible answer to that”, he explains. “It gives you the answer on data governance and data segregation, or if you want high-data-streaming or a complex Nvidia models implemented. These things are thought in from the start as a complete solution which you don’t need to change in order to adapt to new demands or regulations”
An AI-ready infrastructure is also an answer to the unavoidable international engagement in our IT. “The state of the world forces us to view things differently, and with geopolitics in the picture, then on-premises and private cloud-infrastructure is very much in the picture, too. The balance is all about mapping your actual needs right now and setting out the best possible estimates on your demands in a near future. Every other month, my image of my work with AI changes. The things we are talking about today was not even present a year ago. So, make timely investments, and keep your AI-advisor on speed-dial”, concludes Leif Elgaard Høj, Head of AI & Development in Danoffice IT.
From Infrastructure to Impact: AI That Delivers
At Danoffice IT, we specialize in delivering AI Solutions that go beyond theory and into real-world execution. Our team integrates artificial intelligence directly into the operational environments of our clients—whether through secure, air-gapped systems or scalable, on-premises deployments. These AI implementations are not future concepts—they’re active, high-performance systems transforming industries today.
We help companies unlock measurable value by aligning AI capabilities with robust infrastructure strategies. From manufacturing to logistics, our clients are already seeing improved efficiency, reliability, and scalability. Explore our success stories and discover how Danoffice IT can help you build an AI-ready infrastructure tailored to your business needs.
-
CASE: Product counting with AI & Machine LearningDiscover how food manufacturing can optimize production efficiency by using AI & Machine learning. -
CASE: Advanced harbor trackingDiscover how a harbor can become more efficient and secure with AI. Explore our advanced harbor tracking system that shows all the traffic in real time.
