The Great Repatriation: Enterprise IT Infrastructure Returns to On-Premises

After more than a decade of "cloud-first" strategies, a major counter-trend is reshaping enterprise IT: the systematic repatriation of workloads, data, and infrastructure back to on-premises and private data centers. Driven by spiraling cloud costs, data sovereignty regulations, AI performance requirements, and workload maturity, the shift is no longer a fringe phenomenon — it is a mainstream strategic decision

The scale and drivers of the repatriation trend, real-world examples and savings, the Total Addressable Market (TAM) for infrastructure automation; and specifically, where Puppet comes into play.

Scale of the Trend

The numbers are striking and accelerating:

"83% of enterprises plan to move workloads back to private clouds or on-premises — a finding that Michael Dell called 'not surprising.'" — Barclays CIO Survey 2024 / Dell Technologies

Market Size & TAM Growth

The repatriation wave is expanding several adjacent technology markets simultaneously. Below are the key segments most directly impacted.

IT Infrastructure Management Tools

This is the broadest measure of the market that benefits from enterprises managing, automating, and operating their own infrastructure — whether on-prem, hybrid, or private cloud.

Note: The on-premises segment alone held 55.7% of this market in 2024 — a dominant share that is expected to grow further as repatriation accelerates. Large enterprises (companies with 1,000+ employees) represented 75.5% of market spend.

Configuration Management & Infrastructure Automation

This is the segment most directly tied to Puppet: tools that automate, configure, enforce compliance, and manage infrastructure at scale.

This segment is growing at nearly double the rate of the broader IT management tools market — driven directly by the complexity of managing hybrid environments where organizations now run workloads across public cloud, private cloud, and on-prem simultaneously.

Automated Infrastructure Management Solutions

A narrower, automation-focused segment tracking tools for provisioning, monitoring, and orchestrating infrastructure assets — the core use case for Puppet Enterprise.

Bottom line on TAM: The configuration management and automation market, is expected to nearly double from ~$5.8B in 2025 to $12.2B by 2030, growing at 16% CAGR. This is one of the fastest-growing segments in enterprise software infrastructure.

What Is Driving the Trend

1. Cost — The Primary Catalyst

Public cloud pricing models — with variable compute, storage tiers, API call charges, and egress fees — compound in ways that are nearly impossible to forecast. The economics only worsen at scale.

• 27% of cloud infrastructure spending is wasted on underused resources (Flexera 2025).
• 59% of organizations overspent their cloud budgets in 2024 (IDC).
• Predictable, long-running workloads are almost always cheaper to own than rent over a 3–5 year horizon.

2. Workload Maturity

As cloud migrations aged, many enterprises found that workloads which once needed cloud elasticity had stabilized into predictable, steady-state patterns. At that point, the scalability premium charged by cloud providers no longer delivers ROI. Moving these workloads back on-prem — or to colocation — is a straightforward financial optimization.

3. Data Sovereignty & Regulatory Pressure

Legal and regulatory frameworks are tightening globally, making public cloud untenable for certain workloads:

• The EU AI Act, GDPR, California CPRA, and data localization laws in the UK, India (DPDP), and China (PIPL) are all creating requirements that on-prem deployments satisfy inherently.
• The US CLOUD Act is a flashpoint: Microsoft's France GM testified before the French Senate that he could not guarantee French citizens' data was protected from US government access — a disclosure replicated by Google, Amazon, and Salesforce.
• Gartner forecasts that 60% of financial firms outside the US will adopt sovereign or on-premises deployments by 2028.
• Legal and professional services firms face additional confidentiality requirements that make on-premises deployments mandatory.

4. AI Infrastructure Performance

The explosion of enterprise AI workloads is creating a new forcing function for on-prem investment. AI inference and training require massive compute, fast local data access, and cost structures that are often prohibitive in public cloud at scale.

• Fortune 1000 enterprises are weighing on-premises GPU infrastructure (often in colocation facilities) against cloud GPU instances — and increasingly choosing the former for production inference.
• Data gravity is a key factor: training on proprietary data works best when the data and compute are co-located on-prem.
• Regulated industries like finance, healthcare, and legal services often cannot use cloud-hosted AI models for sensitive inference tasks.

5. Performance & Reliability

The "noisy neighbor" effect in shared public cloud environments creates inconsistent performance for latency-sensitive workloads — financial trading systems, industrial automation, real-time data processing, and healthcare applications. Organizations with SLA requirements often find dedicated on-prem hardware more reliable than shared cloud infrastructure.

Real-World Examples & Savings

The following high-profile repatriations have been publicly documented, providing validated proof points for the business case:

How Puppet Plays a Role 

Puppet is an infrastructure automation and configuration management platform purpose-built for the exact challenge that repatriation creates: managing large-scale, complex infrastructure consistently and securely across on-premises, hybrid, and multi-cloud environments.

When an enterprise moves workloads back on-prem, they face an immediate set of operational challenges that Puppet is uniquely positioned to solve. 

Configuration Drift & Compliance Enforcement

On-premises infrastructure requires active configuration management to prevent drift — where servers gradually diverge from their intended state through manual changes, patches, or failures. Puppet's declarative Infrastructure as Code model continuously enforces desired state across thousands of nodes, automatically correcting drift before it causes security or compliance failures.

Hybrid Environment Consistency

The repatriation trend does not mean full cloud exit — most organizations are moving to hybrid models. Puppet provides a single control plane for managing configurations consistently across on-prem servers, private cloud, and remaining public cloud instances. This is the core challenge of hybrid infrastructure: ensuring the same security baselines, logging configurations, and software versions exist everywhere, regardless of where the workload runs.

Scale & Automation at Enterprise Depth

Puppet manages Linux and Windows servers, network peripherals, and edge devices from one platform. As enterprises spin up on-prem hardware at scale — buying server racks to replace cloud compute — automating that infrastructure provisioning and ongoing management is non-optional. Manual administration at thousands-of-node scale is not viable.

Compliance & Security at Scale

Regulated industries (BFSI, healthcare, government) are disproportionately driving repatriation due to data sovereignty requirements. These are exactly the industries where Puppet's compliance enforcement capabilities — role-based access control, automated patch scanning, security baseline enforcement, and audit reporting — deliver the highest value.

Consolidation of Automation Tooling

As enterprises build out on-prem infrastructure, many find themselves with fragmented automation tooling; multiple scripts, agents, and platforms acquired during cloud migration. Puppet positions itself as a consolidation platform: replacing multiple point tools with a single, strategic automation layer across the entire infrastructure estate.

The Destination: Hybrid Infrastructure as Steady State

It is important to note that cloud repatriation is not a wholesale rejection of public cloud. The market is maturing toward a model where infrastructure decisions are workload-specific:

• Stable, predictable, data-intensive workloads → On-premises or colocation
• Regulated, sensitive data workloads → Private cloud or on-premises (sovereignty)
• AI training and production inference (sensitive data) → On-premises GPU infrastructure
• Bursty, elastic, globally distributed workloads → Public cloud
• Development and testing environments → Public cloud (cost-efficient, disposable)

In 2026, hybrid infrastructure is no longer a transitional phase — it is the permanent, steady-state architecture for large enterprises. This means the market for tools that manage complexity across all environments simultaneously is structurally larger than either a pure-cloud or pure-on-prem world would imply.

Gartner forecasts that worldwide public cloud spend will reach $723.4 billion in 2025 — up 21% from $595.7B in 2024. Cloud is not shrinking. But on-premises infrastructure is growing alongside it, expanding the total addressable market for infrastructure automation tools that span both worlds.

Key Takeaways

Explore Puppet Enterprise Read More From Puppet