IT vendors often choose to market a solution as “full-stack AI,” sometimes adding terms like “enterprise-ready” or “turnkey” to make it sound even more appealing. What does that actually mean, and how does it fit into existing IT infrastructures? And what can AI deliver when deployed to manage said infrastructures themselves? Techzine discussed this during SUSECON 2026 with Francisco Perez van der Oord, founder and president of ITQ, an IT service provider for approximately 500 European organizations.
In practical terms, an AI stack or “AI Factory” is often a mix of homegrown and proprietary solutions wrapped into one validated integration. Depending on the IT vendor, it’s either a solution for adding AI functionality to existing infrastructure or a reference architecture for a turnkey AI environment that current IT systems can integrate with. The theory sounds logical and coherent, but we’re curious to know how a company that implements IT systems for leading clients experiences it in practice.
The Ideal Scenario
ITQ is considered one of the few Pinnacle Partners for VMware in Europe, but also works extensively with companies such as Omnissa (formerly VMware EUC), SUSE, Red Hat, and the three major hyperscalers. In other words: clients can turn to the company to set up or modernize their IT infrastructures of all shapes and sizes. And there is certainly variety, notes Perez van der Oord. Even municipalities with practically identical IT requirements opt for vastly different solutions. For ITQ, this naturally means extra business opportunities, according to its founder, but it isn’t entirely logical for governmental agencies to behave in such an unintegrated fashion. However, these decentralized and autonomous IT choices do maintain a certain form of biodiversity in the IT “animal kingdom”, if you will. After all, if every organization chooses the same solution, that may be efficient in the short term but is detrimental to further innovation.
In practice, AI infrastructures are also more complex than the sophisticated diagrams presented at IT events seem to suggest. At such events, you often see a neatly organized structure of interacting components in a colorful graphic. However, organizations are already fully engaged in integrating AI, wherever possible, into their existing IT infrastructure—which doesn’t track with these graphics very much at all. Companies simply aren’t waiting on any particular vendor to help them run a complete AI stack. In fact, their ambitions are so far-reaching that the entire IT infrastructure could potentially run on AI systems, and they’re trying to get to that stage however they can.
The promise of an AI stack that’s even run by AI is great, says Perez van der Oord. He can well imagine that IT decision-makers dream of a world in which AI agents take over the tasks currently handled by VMware, Nutanix, Red Hat, or SUSE. Rather than just an IT environment for running AI, customers are already thinking of a coherent AI-first strategy. Consider, for example, the prospect of automatically preventing downtime by migrating workloads to a new environment—something that could render traditional High Availability (HA) licenses obsolete.
Containers and VMs
The day-to-day reality is still tied to the familiar IT vendors. The IT world hasn’t progressed beyond dreams of replacing entire applications with agents, whether such moves involve SaaS or infrastructure solutions. At the infrastructure level, Perez van der Oord sees a contrast between traditional virtualization and containers. Organizations constantly use a mix of VMs and Kubernetes, with the desire to manage both in a cohesive manner. VMware is and remains a mainstay within IT infrastructures, despite rising prices and the perception by some that it is a legacy platform. ITQ, on the other hand, has many clients who also run containers within VMware itself, for example via SUSE Rancher Prime.
By 2026, SUSE has also emerged as a latecomer advocating for VMware exits, with its own virtualization solution, Harvester, as a direct alternative. Coriolis was featured on the SUSECON stage—a migration tool capable of transferring VMs from VMware to Harvester (also known as SUSE Virtualization, when referring specifically to the enterprise-ready version of Harvester). Perez van der Oord emphasizes that while this may work well for some customers, it is not a one-size-fits-all solution.
Some VMware users work exclusively with vSphere and are often the ones who find it difficult to cope with the price increases and bundling of VMware services in VMware Cloud Foundation. For them, a quick virtualization-to-virtualization migration is possible. But those who have already fully embraced the private cloud concept of VMware by Broadcom cannot simply switch over. All sorts of crucial settings, such as configuring microsegmentation, HA, and disaster recovery at the VM level, do not easily survive the seemingly simple migration. Additionally, Harvester/SUSE Virtualization is nowhere near as mature as VCF, in stark contrast to the cloud-native solution SUSE Rancher Prime. That’s not surprising news, really: SUSE CEO DP van Leeuwen emphasized in a conversation with us at this year’s SUSECON that the focus over the past year has been primarily on providing support for various enterprise Linux distributions and Rancher, rather than Harvester.
Choices galore, except for legacy
An important consideration for IT infrastructure decisions is sovereignty. Perez van der Oord observes that every client has its own definition of sovereignty, meaning vendors cannot simply respond to it off the bat, and suppliers like ITQ cannot apply a one-size-fits-all approach. For example, while a startup may only need data sovereignty to protect IP, it may be critical for a municipality that the software used is of European origin or remains replaceable in the short term with a sovereign alternative. Although AI is increasingly helping to create designs based on previous successes with clients, ITQ still has to come up with a specific, unique design for virtually every client.
On top of that, modernizing infrastructure is fraught with challenges, whether it’s to promote sovereignty, leverage AI, or to reduce legacy costs. “You have a hundred old applications, and then a hundred new ones are built,” with just as many dependencies on the legacy systems to go with them, explains Perez van der Oord. “If you compare today’s situation to three years ago, there are more VMs in the world.”
Anyone hoping for a revolution in that area may need to adjust their expectations. “There’s no business case for modernizing an application.” That’s because organizations are still running on decades-old, in-house applications. The idea of fully switching to “AI-native” technologies such as cloud-native, containers, microservices, and other modern methods of delivering applications is therefore not really feasible, according to Perez van der Oord. That realization is far from new. “Over the past ten years, I’ve heard nothing but people saying applications need to be modernized. Apparently, it’s very complex.”
The sum of open-source and proprietary
The conclusion, therefore, is that modern IT environments are set to remain complex. AI stacks must interact with them or, ideally, integrate with them. Software vendors seem to have come to terms with this reality by now. In many cases, they attempt to overhaul an entire IT environment and sell it to organizations wholesale, often through partners, but this ideal clashes with the actual choices organizations make within their infrastructure. VMs coexist with containers (and containers often run within VMs), AI tools must function alongside legacy code, and sovereign ideals encounter that same layered IT world.
Where a company like SUSE can distinguish itself is in its mature approach to open source, the ITQ founder believes. For Perez van der Oord, the open-source world is still relatively new; his company has been a SUSE partner since 2024 and, by his own account, it has learned a great deal in that short period. Although he believes the advantages of a proprietary solution like VMware are undeniable, he acknowledges that open source plays an important role. For instance, it offers plenty of innovation and low (or no) licensing costs.
On the other hand, enterprise support is crucial. Not all open-source projects are viable in the long term, and dedicated maintainers are a rare breed. That is why a company like SUSE is so important, the ITQ president points out, because such a company curates and safeguards open-source projects from falling into oblivion. The enterprise-ready version of open source relies on support and validation. For an AI infrastructure, which is teeming with open-source solutions that, together with proprietary applications and hardware, offer a “full stack,” this tension is a given.
Finally, Perez van der Oord makes a point on the SUSECON stage that is as obvious as it is underdiscussed: open-source players like SUSE need money just as much as other IT vendors. Although VMware and Red Hat (like many other companies) have faced criticism for changing pricing models and increasing costs for many customers, it is not the case that enterprise open-source can be free.
Conclusion: the reality is multifaceted and complex
Our conversation with Perez van der Oord once again shows that IT environments rarely adhere to the stylized representations of “full-stack AI” solutions. That’s not to say those diagrams are nonsensical or impossible. For a total greenfield—a brand-new startup with plenty of investment capital—such a diagram might be feasible. But the reality is that every customer, every organization, has different wishes and requirements, even if on paper they have an identical profile to another.
That means IT suppliers like ITQ are constantly busy combining tooling—proprietary or not, open-source and closed-source, cloud and on-premises. It’s important to keep that complex landscape in mind when vendors start talking about complete, off-the-shelf AI solutions. The lesson is not that these solutions are unfeasible or undesirable, but that they are being introduced into an IT landscape filled with diverse needs and legacy systems.