Report: Enterprise AI workloads are trending toward private cloud
In a pivotal move for enterprise AI, Broadcom has highlighted a significant shift toward private cloud deployments in its latest research. The report, dubbed “The AI Tipping Point,” indicates a transformative phase where AI workloads are reshaping corporate cloud strategies, influencing considerations around architecture, costs, security, and governance.
The AI Tipping Point: A Closer Look
Broadcom’s Private Cloud Outlook 2026 report is grounded in a comprehensive survey of 1,800 senior IT decision-makers spanning the Americas, Europe, and Asia Pacific. According to this research, the experimentation phase of AI is concluding, and enterprises are increasingly favoring private clouds for deploying AI workloads due to enhanced security and scalability. This shift is primarily driven by the need for better cost management, reduced complexity, and greater control over AI operations.
Production AI is moving towards the private cloud
Data from Broadcom reveals that 56% of surveyed companies are either running or planning to run production AI inference in private clouds. In contrast, reliance on public cloud services for similar workloads has decreased by 15 percentage points from the previous year, now standing at 41%.
[Click on image for larger view.]
Despite this trend, public cloud platforms remain integral to enterprise IT strategies for purposes like experimentation, elastic capacity, and access to specialized services. However, Broadcom’s findings suggest that AI production workloads, which demand consistent computing power, secure data handling, stringent governance, and reliable performance, are highlighting the limitations of an exclusively public cloud-based approach.
The Repatriation Trend
Reflecting this shift, Broadcom’s report indicates that 83% of organizations are contemplating or have already begun repatriating workloads from public to private clouds, with 50% having completed such transitions. By 2025, these numbers are expected to grow, with 69% considering or already having undertaken repatriation, and 35% having fully executed the move.
[Click on image for larger view.]
In the 2026 study, AI emerged prominently as a category for repatriation. Broadcom reports that 43% of organizations moving workloads back to private clouds are focusing on AI training, large language models, and inference tasks.
Cost: The New Concern in Cloud Computing
In a notable development, cost concerns have surpassed security as the primary issue associated with public cloud usage. The report reveals that 31% of respondents now view cost management as their main challenge with public cloud services, up from 26% in 2025.
Moreover, a staggering 97% of IT leaders acknowledge that a portion of their public cloud expenditures is wasted, with 52% estimating the waste to be more than 25%. Broadcom attributes these findings to the pressures of AI infrastructure requirements, including the need for computing power, storage, bandwidth, GPU pricing, data transfer fees, and unpredictable usage patterns.
For more detailed insights, visit the full article Here.
“`
