Digital Realty, Lenovo, and ePlus are partnering to deliver Private AI Infrastructure-as-a-Service, a pre-validated, production-ready AI foundation that takes enterprises from initial assessment through global-scale deployment, on PlatformDIGITAL®.

AI is expected to reshape enterprise infrastructure requirements at a rapid pace the industry.

Enterprises are accelerating AI adoption across industries, from financial services and healthcare to retail and manufacturing. Yet, while experimentation is widespread, many organizations still struggle to move AI from pilot projects to reliable production systems. Research shows that only about half of AI projects successfully move from pilot to production. The challenge is not ambition. It is infrastructure.

Earlier this year, we launched the AI Infrastructure Solutions Program, a five-stage framework for packaging validated AI infrastructure with leading OEM and technology partners so enterprises can move from experimentation to production-scale execution with confidence. The five stages are:

Discover
Validate
Deploy
Connect
Scale

Today, we are pleased to introduce the first solution in this global program:

Private AI Infrastructure-as-a-Service (IaaS), delivered in partnership with Lenovo and ePlus.

Why this matters now

Enterprise demand for AI-ready data center capacity is accelerating at a pace the industry has not seen before. In 2025, Digital Realty signed nearly $340 million in colocation and interconnection bookings across our enterprise product set, a full-year record and more than a third above 2024 levels. We added nearly 600 new customers for the second consecutive year, many of them deploying AI-driven workloads for the first time.

What’s changed is not just the volume of demand, but the nature of it. Enterprises are increasingly looking for higher-density configurations that can support localized inference, agentic AI, MCP (Model Context Protocol), analytics, AI-enhanced digital twins, and fine-tuning workloads. As a result of this shift, daily, high-volume AI tasks are becoming embedded in business operations. At the same time, the infrastructure requirements for these workloads — GPU compute, high-speed interconnection, thermal management, and multi-cloud connectivity — are more complex than anything most enterprise IT teams have provisioned before.

This complexity is precisely why packaged, pre-validated solutions matter. Enterprises shouldn’t have to become data center engineers to deploy AI. They need partners who can meet them where they are and hand them a proven starting point.

A clear path from AI ambition to AI execution

Many enterprises face a difficult choice when deploying AI infrastructure.

Public cloud environments provide flexibility but may introduce unpredictable, usage-based GPU costs and shared tenancy concerns.

Private AI Infrastructure-as-a-Service combines dedicated infrastructure, predictable OpEx economics, and global scalability within Digital Realty facilities — without the variability of metered cloud billing or the capital burden of building your own.
This solution brings together:

Lenovo AI-optimized systems powered by NVIDIA GPUs
ePlus design, deployment, support, and lifecycle expertise
Digital Realty’s global high-density colocation and interconnection platform

Together, these capabilities deliver a production-ready AI foundation engineered for enterprise workloads.

How it works: A repeatable path from discover to scale

Private AI IaaS follows the five-stage AI Infrastructure Solutions Program framework, compressing the path from initial interest to global, repeatable production:

Discover. A Solution Architect workshop where Digital Realty, Lenovo, and ePlus align on the enterprise’s AI use case, workload characteristics, performance requirements, and infrastructure readiness. Using PDx® Visualizer and Builder, the team produces an AI Infrastructure Readiness Assessment and a recommended OEM solution shortlist — giving the enterprise a clear picture of what success looks like and which configurations to validate.
Validate. A hands-on proof-of-concept in Digital Realty Innovation Lab (DRIL), where the enterprise runs actual workloads on Lenovo GPU platforms under real-world power, cooling, and network conditions. The DRIL — now available in Northern Virginia, with locations expected to open in Singapore, Tokyo, and London in 2026 — supports up to 150kW per cabinet. Digital Realty captures performance telemetry and environmental data. ServiceFabric® connectivity is mapped during this phase, testing multi-cloud and partner network paths alongside compute. The output is a formal DRIL PoC (Proof-of-Concept) Report with benchmark results, validated configuration specs, and token economics — a documented business case before any capital commitment is made. PDx Modeler and the PDx Token Calculator are used throughout.
Deploy. Based on validated results from DRIL, the enterprise moves to a live production environment, typically a one, three, or six-rack footprint on Digital Realty Colocation, with Lenovo GPU platforms commissioned and operational on day one. The validated DRIL configuration and OEM bill of materials flow directly into production, eliminating rework. PDx Modeler supports final rack and power design.
Connect. With production infrastructure live, ServiceFabric ports and virtual cross-connects (VXCs) are activated, connecting the enterprise to 300+ cloud on-ramps and 700+ connected data centers across PlatformDIGITAL. The Private AI Exchange (AIPx) can provide direct, private connectivity to leading AI platforms and model providers, removing the public internet from the inference path and creating the performance layer for enterprise AI.
Scale. With the architecture proven and connected, the validated design can become a repeatable blueprint, deployable across additional racks, higher GPU density configurations, new metros, and liquid-cooled environments as workloads grow. PlatformDIGITAL’s 30+ country, 55+ metro footprint means enterprises can replicate validated Lenovo configuration globally with consistent infrastructure standards, sovereign deployment options, and regional partner activation.

“Enterprise AI is no longer a vision — it’s an execution challenge. Organizations know the value AI can unlock but scaling it securely and efficiently requires the right foundation. Private AI Infrastructure-as-a-Service delivers that foundation. By combining Lenovo’s AI-optimized platforms with Digital Realty’s global high-density data centers and ePlus’s deep integration and lifecycle expertise, we’re providing a production-ready Enterprise AI backbone — enabling organizations to operationalize AI with speed, scale, and confidence.”— Flynn Maloy, Chief Marketing Officer, Lenovo Infrastructure Solutions Group

A dedicated AI foundation, built for production

Private AI IaaS delivers single-tenant GPU clusters deployed within Digital Realty data centers. It is designed for organizations running sustained AI workloads, not short-term pilots.
The solution is designed to support a range of enterprise AI use cases, including:

GPU-as-a-Service
Retrieval-augmented generation (RAG)
Model Context Protocol (MCP)
High-volume inference
AI-embedded enterprise applications
Fine-tuning domain-specific models
Advanced analytics and digital twin simulations

Unlike shared cloud GPU services, this model provides dedicated capacity with defined power and cooling envelopes. Infrastructure is standardized and repeatable, enabling predictable deployment timelines and expansion paths.

Global scale with integrated connectivity

AI workloads rarely operate in isolation. They depend on secure access to cloud platforms, data sources, partners, and users, and the economics of that connectivity matter as much as the compute itself.

Digital Realty’s PlatformDIGITAL provides global reach across 30+ countries and 55+ metros worldwide, while ServiceFabric enables secure, low-latency connectivity to 300+ cloud on-ramps and 700+ connected data centers across PlatformDIGITAL. The Private AI Exchange (AIPx) extends this further, offering direct, private connectivity to leading AI platforms and model providers, keeping inference traffic off the public internet and turning interconnection into a performance layer. By integrating connectivity into the architecture from day one, enterprises can support hybrid and distributed AI strategies without compromising performance or governance requirements.

Private AI IaaS is delivered through a fixed OpEx model with defined capacity. Enterprises can gain dedicated GPU allocation and transparent pricing, without exposure to metered billing variability or shared-resource contention, enabling long-term infrastructure planning with cost confidence.

Built for industry-specific AI use cases

Private AI Infrastructure-as-a-Service supports deployments across industries where AI is becoming increasingly critical:

Financial services: real-time fraud detection and algorithmic trading
Healthcare and life sciences: AI-accelerated imaging analysis and clinical documentation
Retail and eCommerce: AI-driven personalization and operational optimization
Manufacturing: predictive maintenance and digital twin simulations

Each deployment follows validated architectural standards designed for reliability, scalability, and operational efficiency.

Our enterprise customers are not struggling with AI ambition — they are struggling with operationalizing it at scale. The complexity of infrastructure design, integration, and lifecycle management can slow progress significantly. ePlus Private AI Infrastructure-as-a-Service provides a structured, repeatable deployment model that moves organizations from evaluation to sustained production with greater speed and predictability. By integrating validated hardware platforms within Digital Realty's global infrastructure and connectivity ecosystem, we're delivering a clear path to enterprise-grade AI execution and scalability. — Ken Farber, President, ePlus Technology

The first in a series of AI infrastructure solutions

Private AI Infrastructure-as-a-Service with Lenovo and ePlus marks the first solution to launch under the AI Infrastructure Solutions Program and it sets the template for what follows. Additional solutions targeting different workload profiles, OEM platforms, and route-to-market partners are in development, and the DRIL network is now expanding globally to Singapore, Tokyo, and London, expected to support validation at scale in 2026.

The objective is clear: translate platform strength into structured, ready-to-deploy AI infrastructure solutions that accelerate enterprise adoption, moving customers through a repeatable five-stage path from Discover to Scale.

AI is expected to reshape industries. The infrastructure supporting it must be equally resilient, scalable, and future-ready. With Private AI Infrastructure-as-a-Service, enterprises gain a clear foundation for deploying AI at scale, built on Digital Realty’s global platform, validated in DRIL, and delivered through trusted technology partners.

Partner with Digital Realty, Lenovo, and ePlus to accelerate your AI deployment

Delivering AI at scale requires more than infrastructure. It requires the right team around the table from day one.

Digital Realty, Lenovo, and ePlus bring together the platform, the hardware, and the services expertise to take enterprises from initial discovery through sustained production — without the guesswork. Digital Realty provides the global AI-ready colocation platform and interconnection fabric. Lenovo delivers AI-optimized infrastructure, purpose-built for GPU-dense workloads. ePlus provides the design, integration, and lifecycle management that makes the deployment model repeatable and scalable across your organization.

Learn more about the AI Infrastructure Solutions Program or contact us to schedule a Discover workshop and begin your five-stage path to production AI, validated in DRIL, deployed on PlatformDIGITAL, and connected through ServiceFabric.

Introducing Private AI Infrastructure-as-a-Service with Lenovo and ePlus