Guest: Dominic Wilde
Company: Mirantis
Show Name: An Eye on AI
Topic: AI Infrastructure

AI is reshaping enterprise infrastructure in ways few anticipated. What began as a rush toward GPUs and cloud scalability has now collided with the realities of sovereignty, repatriation, cost control, and hybrid complexity. In this clip, Dominic Wilde, SVP & GM of the Core Business at Mirantis, explains why private cloud is returning as a strategic pillar — and how Mirantis is guiding enterprises through an end-to-end “metal to models” transformation.

The infrastructure world tends to move in cycles, but every cycle is shaped by a new dominant force. Today, that force is AI. As organizations prepare for large-scale model development and inference, they’re discovering that much of the conventional wisdom about cloud modernization no longer holds. Wilde notes that Mirantis is seeing a “renewed interest in private cloud,” driven by several overlapping pressures: repatriation, data sovereignty requirements, cost predictability, GPU scarcity, and the new operational demands of AI.

Private cloud isn’t replacing public cloud — but it’s becoming a critical part of the hybrid cloud puzzle. Enterprises need environments they can control, optimize, and scale with predictable economics, especially as GPU clusters become core business assets. With its long history in OpenStack, Kubernetes, and large-scale infrastructure services, Mirantis finds itself at an advantage in this new landscape. Wilde explains that Mirantis has deployed private cloud technologies “at incredible scale,” paired with the services capability to guide customers through both architecture and operations.

Hybrid cloud, he notes, “is now a brutal reality.” Organizations are no longer talking about hybrid as an aspiration; they are living it every day. This hybridity brings complexity that multiplies as GPU workloads enter the picture. Enterprises don’t just need cloud flexibility — they need a coherent strategy to move from bare metal all the way to LLMs. That’s where Mirantis’ “metal to models” vision comes into focus.

Wilde breaks the journey into layers: bare metal, Infrastructure-as-a-Service, GPU platform services, AI platform services (or NeoPaaS), and finally the delivery and lifecycle of AI models. Mirantis’ strength is in providing a consistent operational approach across these layers, enabling customers to choose the right technology building blocks without being boxed in by a particular vendor’s agenda. Wilde emphasizes this point repeatedly: Mirantis gives customers “the opportunity to have a fast on-ramp, but an exit if they need it,” a philosophy grounded in openness and choice.

One of the most striking insights from this clip is Wilde’s observation that many enterprises “don’t even know the questions to ask” when preparing for AI. This isn’t a reflection of capability — it’s a sign of how quickly the ecosystem is evolving. To address this gap, Mirantis introduced an AI maturity model, a practical framework available on their website. The model guides organizations through essential topics: readiness of their data layer, current GPU strategy, hybrid integration, cost control, performance considerations, and which platform layers they actually need to manage.

This maturity model has gained strong interest because it meets organizations at a critical inflection point. Many want to adopt AI but are unsure whether their existing cloud, Kubernetes, or virtualization environments can support the demands of large models. Others are already running GPUs but are now confronting the complexity of scaling up while maintaining sovereignty and operational control.

Another major factor driving private cloud interest is cost. Training and inference pipelines at scale require sustained GPU access — something that is dramatically more predictable and often more economical in private environments. Public cloud remains valuable for burst capacity, experimentation, and global reach, but long-running GPU pods can quickly overwhelm cloud budgets. Combined with rapidly growing sovereignty requirements — especially in regulated industries and certain regions — private cloud has re-emerged as a solution to both operational and governance challenges.

Mirantis’ breadth of expertise allows the company to guide customers not only in technology choices but in mapping out a forward-looking AI strategy. Wilde stresses that Mirantis views modernization as a journey, not a mandate. Customers can adopt pieces of the stack gradually, integrate with existing tools, and make decisions based on their own business priorities rather than vendor pressure.

The shift toward AI presents both an opportunity and a risk for enterprises. Those who architect thoughtfully — from metal to models — are poised to build infrastructure that accelerates innovation. Those who patch together tactical solutions risk creating siloed GPU islands, fragmented Kubernetes estates, and unsustainable cost structures. Wilde’s message is clear: AI demands an end-to-end plan, not scattered upgrades.

Private cloud’s resurgence is part of this broader strategic picture. It provides the control plane for sovereignty, the cost stability for GPUs, and the operational consistency needed across hybrid environments. Combined with Mirantis’ commitment to open, composable architecture, it enables enterprises to enter the AI era without abandoning the flexibility they’ve built over the past decade.