Research: AI Workloads in Companies Are Trending Towards Private Cloud
Broadcom is pivoting to a so-called “tipping point” for enterprise AI in its latest private cloud research, saying production AI workloads are changing the way companies evaluate cloud architecture, costs, security, and governance.
The company’s Private Cloud Outlook 2026 report, titled “The AI Tipping Point,” is based on a global survey of 1,800 senior IT decision-makers in the Americas, Europe, and Asia Pacific. In a press release, Broadcom claimed that the AI experimentation phase is over and that the private cloud is where enterprise AI workloads will be deployed for security and scalability reasons. The company said change will be driven by “cost, complexity and control” as companies move AI workloads into production.
Production AI is Moving Towards the Private Cloud
According to Broadcom, 56% of companies surveyed are running or planning to run production AI inference in a private cloud. The company also said public cloud usage for the same workloads fell 15 percentage points year-over-year, from 56% to 41%.
[Click on image for larger view.] Where AI-based applications or workloads are currently running (Source: Broadcom).
According to the report, public cloud remains part of the enterprise IT strategy for experimentation, elastic capacity, and specialized services. However, according to Broadcom, AI production workloads introduce persistent computing needs, sensitive data flows, governance requirements, and performance expectations that can expose limitations with a purely public cloud approach.
The shift is also reflected in the return data. According to Broadcom, 83% of organizations are considering or have already repatriated workloads from the public cloud to the private cloud, and 50% have already repatriated some workloads. In 2025, 69% are considering repatriation or have already done so, and 35% have already done so.
[Click on image for larger view.] Return trends for AI-based applications or workloads (Source: Broadcom).
AI first appeared as a return category in the 2026 study. According to Broadcom, 43% of organizations repatriating workloads are moving AI training, large language models, and inference from the public cloud to the private cloud.
In a Broadcom blog post about the report, Prashanth Shenoy, Vice President of Marketing for Broadcom’s VMware Cloud Foundation Division, wrote that “the direction has changed as enterprises look to scale.” The same post states that enterprise AI “has found its infrastructure. And it is a private cloud.”
Cost Replaces Security as the Public Cloud’s Biggest Problem
Broadcom said that in the 2026 study, cost has overtaken security as the public cloud’s biggest concern. According to the report, 31% of respondents cited cost management as the top public cloud challenge, up from 26% in 2025.
According to the report, 97% of IT leaders surveyed believe some of their public cloud spending is wasted, while 52% said waste is more than 25%. Broadcom linked these findings to AI infrastructure pressures, including computing power, storage, bandwidth, GPU prices, data transfer fees, and unpredictable usage patterns.
For further details, the full report can be accessed Here.
“`
