Broadcom Releases 'Private Cloud Outlook 2026' Report: AI Reaches Inflection Point, Production AI Inference Decisively Shifts to Private Cloud

*This content is a press release issued by Broadcom on June 9, 2026 (US time).

Broadcom Inc. (NASDAQ: AVGO), a global technology leader designing, developing, and delivering semiconductor and infrastructure software solutions, today announced its latest report, 'Private Cloud Outlook 2026.' Key findings reveal that the experimental phase of AI is over, and enterprises are increasingly deploying AI workloads on private cloud to ensure security and scalability.

Last year's report highlighted a 'cloud reset' aimed at balancing public and private cloud usage. In 2026, AI is reaching a definitive inflection point. This shift is driven by three factors—cost, complexity, and control—and public cloud environments are increasingly unable to support large-scale production AI operations. Key findings from the report include:

- 56% of enterprises are currently running or planning to run production AI inference on private cloud, while public cloud usage for similar workloads has dropped 15 percentage points year-over-year from 56% to 41%. - The top AI-related challenges facing enterprise IT departments are data protection and privacy (37%) and security and governance (36%). - For the first time, 'cost' has surpassed 'security' as the top concern for public cloud, rising from 26% in 2025 to 31% in 2026. - 97% of IT leaders believe some portion of their public cloud spending is wasted, and 52% estimate that over 25% of their total public cloud budget is wasted. - 83% of enterprises are considering repatriating workloads from public to private cloud, with 50% already having done so. The second-largest driver for repatriation is 'cost predictability,' cited by 39% of enterprises, marking a sharp rise. - Four out of five IT leaders say geopolitical factors are directly impacting their IT strategy and operations. 'Data sovereignty and data location requirements' (54%) have now surpassed 'compliance requirements by jurisdiction' (51%) as the primary geopolitical factor influencing infrastructure decisions.

Prashanth Shenoy, Vice President of Marketing, VMware Cloud Foundation, Broadcom Inc., commented:

'As enterprises move AI from pilot to production, infrastructure and operational costs are skyrocketing, security challenges are emerging, and complexity is increasing. As our findings show, enterprises are increasingly choosing private cloud for production AI workloads.'

The Shift in AI Inference: Production AI Moves to Private Cloud

The most significant finding in this year’s report is the scale and speed of change in where enterprises run AI workloads. While public cloud remains effective for AI pilot projects and model training experiments, its cost efficiency for large-scale inference processing is different.

56% of enterprises are currently running or planning production inference on private cloud, while only 41% are doing so on public cloud—a complete reversal from last year when both were nearly equal. The 15-point drop in public cloud usage for production AI workloads in just one year is one of the most dramatic changes in this year’s report.

The reasons are clear. As IT leaders in the survey pointed out, public cloud remains expensive and lacks the governance and control needed for large-scale AI operations. These disadvantages may be acceptable during pilot or training phases where agility is prioritized. However, as enterprises seek to scale, cost and governance requirements become critical, leading them to bring workloads back in-house. 62% of IT leaders say they are 'very concerned' or 'extremely concerned' about infrastructure costs for generative AI and agent-based AI, and 36% say AI is creating new requirements for data protection, privacy, security, and risk management.

Data Sovereignty Demands: Infrastructure Strategy Reshaped by Geopolitics

In 2026, geopolitics has become a central theme in infrastructure discussions. Currently, four out of five IT leaders say geopolitical factors are directly affecting their IT strategy and operations. Data sovereignty has moved from a mere compliance requirement to a board-level priority. 'Data sovereignty and data location requirements' (54%) have now surpassed 'compliance requirements by jurisdiction' (51%) as the primary geopolitical factor influencing infrastructure decisions. Industries with high security and compliance demands—financial services, public sector, healthcare, and life sciences—are at the forefront of this shift. With AI-driven data growth, cross-border data governance complexity, rising public cloud costs, and governance burdens, private cloud infrastructure—keeping sensitive data under internal control—is becoming an increasingly compelling option for enterprises.

The Cost Reality: Public Cloud Cost-Effectiveness Reaches Its Limits

Currently, the top concern for public cloud usage is 'cost,' surpassing 'security' with concern rising from 26% (2025) to 31% (2026). Additionally, findings on cost waste are noteworthy: 97% of IT leaders believe some portion of their public cloud spending is wasted, and 52% estimate that over 25% of their total public cloud budget is wasted.

These cost factors are directly driving the trend of workload repatriation to private cloud. Currently, 83% of enterprises are considering moving workloads from public to private cloud, and 50% have already migrated at least some workloads. The top three drivers for repatriation are security and compliance (51%), cost predictability (39%), and performance (39%). The sharp rise of 'cost predictability' as a primary repatriation driver is one of the most notable changes from last year’s report, highlighting the sharply declining cost-effectiveness of public cloud in the AI era.

In response, investment intent in private cloud is accelerating. Intent to invest in private cloud is expanding at twice the pace of public cloud (projected over the next three years, public cloud