The cloud conversation has changed. Has your AI strategy?

Whether you’re launching global marketing campaigns, monitoring banking transactions for fraud, or analyzing sensor data in real time, there’s no AI without the cloud.

However, when it comes to delivering business value, the spotlight is firmly on data strategy and governance. If you can’t securely operationalize your proprietary data, you have a problem.

This calls for a new way of working: Don’t move your data to your AI; instead, move your AI to where your data is already located.

Private cloud does exactly that. It is where enterprise data, governance and AI converge. Unlike the private cloud of five years ago, which was centered on cost optimization and workload consolidation, private cloud is now the modern infrastructure foundation that makes AI real.

Helping you build that foundation is exactly what we do. In recognition of our strengths in delivering full‑stack private cloud transformation, NTT DATA was named a Leader in Everest Group’s Private Cloud Services PEAK Matrix® Assessment 2025 for our global private cloud services and delivery.

Your AI is only as good as the data it can reach

Enterprise data is found in data centers, at the edge, on manufacturing floors and in healthcare systems, financial networks and government environments — in addition to public clouds. Sending it to a centralized location is not only operationally complex in many regulated industries but may also introduce legal and commercial challenges.

That constraint is affecting where organizations invest in AI infrastructure. According to Gartner®, “worldwide end-user spending on AI-optimized IaaS will total $18.3 billion by the end of 2025 and $37.5 billion in 2026.” It adds: “In 2026, 55% of AI-optimized IaaS spending will support inference workloads and it is projected to reach more than 65% in 2029.”*

Infrastructure as a service (IaaS) refers to cloud infrastructure such as computing, storage and networking that organizations consume on demand. Inference is AI working in production — every time a model responds to a question, flags a transaction or generates a recommendation, that’s inference happening in real time.

The trend of moving to IaaS matters because inference happens where business decisions are made. Running inference at scale demands the ability to handle thousands of simultaneous requests, predictable performance and strict governance over what data an AI model accesses. And for regulated and sovereignty-sensitive workloads, organizations need greater control, governance and predictability than shared environments typically provide.

For many organizations, the answer to that challenge is private cloud — the environment built to run AI where your data lives, under your control.

• ALSO READ → Why cloud architecture now determines AI success

Private cloud is evolving — and so is the conversation

The conversation around private cloud used to center on virtualization, consolidation and cost control. Today, organizations choose private cloud because of what AI demands of their architecture:

• Data sovereignty and compliance: Control over where data resides, who can access it and under which regulatory conditions
• Security and AI governance: Protection of applications, models, training data, intellectual property, prompt histories and inference results
• Predictable economics: GPU use, inference costs, model hosting and data egress (many AI workloads become more economical when running at scale on private infrastructure)
• High-performance infrastructure: Dedicated, low-latency, GPU-optimized environments for production inference

These changing priorities are already influencing cloud investment.

“Worldwide sovereign cloud infrastructure as a service (IaaS) spending is forecast to total $80 billion in 2026, a 35.6% increase from 2025,” according to a Gartner press release.**

In the press release, Gartner also estimates that “due to an increased desire for geopatriation projects, sovereign cloud IaaS spending will shift 20% of current workloads from global to local cloud providers.” It adds: “Hyperscalers face mounting pressure as local cloud providers gain share and governments demand greater platform regionalization to meet regulatory and national security requirements.”

NTT DATA’s own research shows that 99% of organizations expect private cloud adoption to grow, while sovereign cloud adoption is projected to rise 50% in just two years, from 28% of organizations today to 42%.

Sovereign AI is now a business imperative

As control over infrastructure becomes more important, so does control over the AI running on it.

Sovereign AI is becoming a board-level priority, particularly in government, financial services, healthcare, manufacturing and critical infrastructure. It means maintaining control over your models, data and AI operations while meeting jurisdictional requirements.

Our 2026 Global AI Report: A Playbook for Private and Sovereign AI shows that nearly 60% of AI leaders — based on their AI-linked business performance — already cite cross-border data restrictions as a major challenge, while 57% of CEOs rank data privacy and sovereignty across geographies as the number-one security or compliance governance threat their organization faces.

However, the systems most organizations built for borderless data flows aren’t designed for AI that must run in controlled, jurisdictionally bounded environments.

That calls for an architecture built around sovereignty from the outset. It starts with a sovereign AI platform that establishes governance, compliance and operational control. On top of that sits private AI, providing a secure environment for managing models and data. Underpinning both is purpose-built AI infrastructure that delivers the performance, scalability and resilience required to run enterprise AI workloads. Each layer reinforces the others, turning sovereignty into a practical operating model.

One global industrial manufacturer faced exactly this challenge. Operating in a highly regulated environment, they needed assurance of strict data confidentiality, regulatory compliance and seamless collaboration across their subsidiaries before deploying a single AI model. We delivered a full-stack sovereign AI platform that combined infrastructure, software and managed services in a secure, unified environment, giving more than 100 data scientists the foundation to shorten research and development cycles and move AI from the lab into production.

The right architecture runs every workload in the right place

A strong AI strategy uses every environment — whether on public of private cloud — for what it does best. Enterprise AI will remain hybrid because different workloads have different requirements, and each workload needs to be matched to the environment it was built for.

Public cloud speeds up experimentation and model training. Private cloud governs production inference and the data that can’t leave your environment. Edge brings AI closer to where data is generated, wherever latency and data locality — keeping it physically close to where it’s created, stored or used — matter most.

That’s especially true in factories, hospitals, retail environments and transportation networks, where the combination of edge, private cloud and sovereign AI becomes increasingly important.

Training and inference: Not the same workload

The clearest example is the difference between training and inference workloads.

Training and inference place fundamentally different demands on infrastructure, yet many organizations assume the same environment can efficiently support both. In reality, their performance, scalability and cost requirements are very different.

• Training and fine-tuning require highly scalable GPU clusters, high-speed networking and nonblocking architectures to process large models and datasets as quickly as possible.
• Inference prioritizes latency, concurrency and cost efficiency. Large models or high-volume applications may also require distributed architectures.

These differences directly influence cost per token — one of the most important measures of GenAI economics. Matching each workload to the right infrastructure improves performance while controlling costs.

How NTT DATA and Dell Technologies work together

Building private cloud and AI is both an infrastructure challenge and a business transformation challenge.

If, like many other organizations, you want to own your private cloud environment but not run it, our managed services model is designed exactly for that. We bring together consulting, AI strategy, cloud, data and security to design, operate and optimize architectures that place every workload where it belongs, with governance in place from the outset.

Dell Technologies is our strategic global AI infrastructure partner, supporting the full AI lifecycle from core infrastructure and storage to AI factories, edge computing and sovereign AI environments.

Together, we continuously invest in AI-ready infrastructure. By the time we walk into a client conversation, we’ve already built, tested and validated the latest AI architectures, absorbing the risk so you don’t have to. That matters because of the rapid pace of evolution of AI infrastructure. Without a partner doing this work continuously, it’s easy to base today’s decisions on yesterday’s architecture.

Dell’s NVIDIA-powered infrastructure underpins our private and sovereign AI architectures, supporting everything from Dell AI Factory and edge environments to Microsoft Azure Local and Google Distributed Cloud deployments.

The objective isn’t private cloud for its own sake. It’s helping you operationalize your data, govern AI responsibly and deploy AI securely at enterprise scale.

• ALSO READ → Cloud modernization: AI is exposing the limits of legacy systems

* Gartner Press Release. Gartner Says AI-Optimized IaaS Is Poised to Become the Next Growth Engine for AI Infrastructure. 15 October 2025.

GARTNER is a registered trademark and service mark of Gartner, Inc. and/or its affiliates in the U.S. and internationally and is used herein with permission. All rights reserved.

** Gartner Press Release. Gartner Says Worldwide Sovereign Cloud IaaS Spending Will Total $80 Billion in 2026. 9 February 2026.