Broadcom's Private Cloud Outlook 2026 finds that enterprises are shifting production AI workloads towards private cloud, marking a clear change in infrastructure choices for AI deployment.
The study found that 56% of enterprises globally are running or planning to run production AI inferencing on private cloud. That compares with 41% using public cloud for the same workloads, down from 56% a year earlier.
The figures suggest companies are making a clearer distinction between early AI testing and wider operational use. Public cloud remains in use for pilots and model training, but organisations are increasingly choosing private environments for inference at scale.
The shift is being driven by cost, governance and control. Among the biggest new demands AI places on IT teams, respondents most often cited data protection and privacy at 37%, followed by security and control at 36%.
Cost has also become the top public cloud concern. The share of respondents naming cost as their main issue with public cloud rose to 31%, overtaking security, which held that position a year earlier.
Waste in cloud spending was another recurring theme. The survey found that 97% of IT leaders believe at least some of their public cloud spending is wasted, while 52% estimate that more than a quarter of their total public cloud budget is not delivering value.
Workload returns
Those pressures are feeding a broader move to bring workloads back from public to private infrastructure. The study found that 83% of enterprises are considering repatriating workloads from public cloud, and 50% have already done so.
The leading reasons were security and compliance, cited by 51% of organisations, followed by cost predictability and performance, each on 39%. The report described the rise of cost predictability as one of the sharpest changes in this year's results.
In Asia Pacific and Japan, the trend was even more pronounced. The survey found that 82% of organisations in the region have considered moving workloads from public cloud to private cloud, while 54% have already repatriated some workloads.
Regional respondents pointed to the same core concerns seen globally, with security and compliance, infrastructure cost predictability and performance among the main reasons for the shift. The findings suggest AI deployment is pushing infrastructure choices higher up the corporate agenda as companies seek tighter operational control.
Sovereignty concerns
The survey also found that geopolitical pressures are reshaping IT planning. Four in five IT leaders said geopolitics are affecting their strategy and operations, while 54% named data sovereignty and residency requirements as a leading factor in infrastructure decisions.
That put sovereignty issues ahead of jurisdiction-specific compliance requirements, cited by 51%. The pattern was especially relevant for sectors with strict regulatory obligations, including financial services, the public sector, healthcare and life sciences.
Those industries are dealing with a combination of larger AI-driven data volumes, tougher cross-border governance requirements and rising scrutiny over where information is stored and processed. In that environment, private cloud is being treated as a way to retain direct oversight of sensitive data and workloads.
Investment plans also showed stronger demand for private cloud than public cloud over the next three years. Private cloud spending intent rose by 21 points over that period, compared with 10 points for public cloud, while 58% of IT leaders named building new workloads on private cloud as a top priority.
The research was based on a survey of 1,800 senior IT decision-makers at enterprise organisations with 1,000 or more employees across eight countries in North America, Europe and Asia-Pacific. Radius Tech carried out the work in partnership with Broadcom.
Broadcom also included a regional view from its Asia leadership. "Across Asia Pacific and Japan, organizations are moving beyond AI experimentation and focusing on production deployment. As AI workloads scale, enterprises are prioritizing security, cost predictability, performance and operational control, factors that are accelerating the shift towards private cloud. With 82% organizations in Asia Pacific and Japan considering workload repatriation to private cloud, this marks a turning point in how organizations are thinking about AI infrastructure," said Sylvain Cazard, president, Asia Pacific, Japan & Middle East, Broadcom.