Companies Migrate AI to Private Clouds for Cost and Control

The Tipping Point for Enterprise AI

For years, the dominant narrative dictated that AI would run on the public cloud. Hyperscalers offered APIs, GPU capacity, and the inertia of a decade of cloud investment. However, Broadcom's Private Cloud Outlook 2026: The AI Tipping Point report, based on a global survey of 1,800 senior IT executives, reveals a radical shift: production AI is finding its home in the private cloud.

What Happened?

The study shows that public cloud usage as the primary environment for production AI inference dropped from 56% in 2025 to 41% in 2026. In contrast, 56% of companies now run or plan to run inference on private clouds. More telling: 43% of companies actively repatriating workloads are doing so with AI workloads — training, large language models, and inference — a category that didn't even exist in last year's report.

Overall repatriation has also accelerated: 83% of companies consider repatriating workloads, up from 69% in 2025, and half have already moved at least some workloads, a 15-percentage-point jump in a single year.

Why It Matters

The forces pushing AI toward private clouds are the same ones that previously attracted storage, security-sensitive applications, and regulated data: security, control, cost, and governance. But with AI, the consequences of getting it wrong are much harder to absorb at production scale.

For the first time in this study, cost has surpassed security as the top concern about public cloud. Nearly all IT leaders (97%) believe some of their public cloud spending is wasted, and more than half (52%) say that waste exceeds 25% of total spending. Generative and agentic workloads are compounding the pressure: 62% of IT leaders are very or extremely concerned about AI infrastructure costs.

As a result, the net intention to increase private cloud investment over three years jumped from 51% to 72%, and private cloud investment is growing at twice the rate of public cloud. Cost predictability has become the second biggest driver of this shift, cited by 39% of organizations.

Market Implications

This move has profound implications. Hyperscalers like AWS, Azure, and Google Cloud could see their AI growth stall as companies opt for more controllable environments. Private cloud infrastructure providers, such as VMware (now part of Broadcom), Nutanix, and OpenStack, benefit directly. Additionally, data sovereignty has become a boardroom priority: 86% of IT leaders say geopolitical and regulatory factors directly affect their IT strategy, and 54% cite data residency requirements as the top concern.

What Readers Should Know

For companies planning to scale AI, it's crucial to reassess the premise that public cloud is the default choice. The report suggests private cloud offers predictable costs, direct control, and regulatory compliance — critical aspects for production AI workloads. Startups and tech companies should consider hybrid models, but the data indicates a clear trend toward repatriation. Investors should monitor private cloud infrastructure providers, which are well-positioned to capture this growing spend.

Cost has surpassed security as the top concern about public cloud. Nearly all IT leaders (97%) believe some of their public cloud spending is wasted.

Conclusion

We are at a tipping point: enterprise AI at scale is opting for private cloud. Security, control, cost, and data sovereignty are the drivers. Companies that ignore this trend could face unpredictable costs and compliance risks. AI infrastructure is changing hands, and private cloud is the new home.