Enterprises are rethinking their reliance on hyperscalers such as AWS, Azure, and Google Cloud as operational costs rise and infrastructure needs become more specialized. The shift isn’t a retreat from cloud computing but a reassessment of where workloads belong and how control, performance, and cost efficiency can be balanced. Across sectors, IT leaders are turning to smaller vendors, colocation providers, and modern private-cloud environments that offer predictable economics and data sovereignty without sacrificing scalability.
A Broadcom-commissioned Private Cloud Outlook 2025 report found that 69 percent of enterprises are considering repatriating workloads from public to private cloud, with more than one-third having already done so. Far from signaling disillusionment, this repatriation reflects a more nuanced strategy: placing each workload in the environment that best meets its cost, security, and latency requirements. “Private cloud now outperforms public cloud,” said Hock Tan, CEO of Broadcom, at VMware Explore 2025, citing VMware Cloud Foundation 9.0 as an end-to-end platform that restores performance and control. “It has better security, better cost management, and of course, greater control,” he said.
Rising cloud bills are one catalyst. Hyperscalers’ variable-use pricing and layered managed services often produce unpredictable expenses once workloads scale. Many enterprises initially drawn by pay-as-you-go flexibility now describe double-digit annual cost growth that outpaces revenue. Smaller providers, by contrast, promise transparent pricing and infrastructure tuned to specific workloads. The economics are especially compelling for steady, compute-intensive applications, AI inference, analytics, or ERP, where predictable demand justifies dedicated capacity.
Security and compliance pressures add weight to the recalibration. With generative-AI adoption accelerating, companies are increasingly wary of sensitive data traversing public-cloud environments. “The excitement and related fears surrounding AI only reinforce the need for private clouds,” said Dave McCarthy, research vice president for cloud and edge services at IDC, in a recent CIO.com interview. IDC found that 92 percent of IT leaders trust private cloud more than public options for compliance-critical workloads. In industries such as healthcare, finance, and defense, control over data location and auditability often trumps the agility benefits of public platforms.
Smaller Vendors Find Their Moment
A growing ecosystem of smaller vendors and infrastructure alliances is capturing enterprise workloads that once defaulted to hyperscalers. In October, Hitachi Vantara announced a partnership with Supermicro combining GPU-accelerated compute systems with Hitachi’s VSP One unified storage platform. The goal is to integrate compute and data more tightly for mission-critical and AI workloads while maintaining enterprise-class governance. “The convergence of Supermicro’s leadership in AI compute with the scale and resiliency of Hitachi Vantara’s platform marks an important step in building the foundation that will guide the future of enterprise AI,” said Sheila Rohra, CEO of Hitachi Vantara, in the company’s statement. Vik Malyala, president at Supermicro, added that the joint offering “supports compute-intensive workloads for dynamic vertical applications, yielding better performance and efficiency for customers.”
At the infrastructure edge, Digital Realty, Dell, and DXC Technology have formed a similar coalition. Their initiative brings Dell’s AI Factory and DXC’s managed-services framework directly into Digital Realty’s colocation data centers. The approach allows enterprises to deploy private AI frameworks closer to data sources, reducing latency and data-transfer costs while maintaining operational control. “AI success requires more than infrastructure—it demands the right people, processes, and technology,” said Holland Barry, global field CTO at DXC, in the companies’ joint announcement.
These collaborations show how second-tier providers are leveraging specialization and proximity to win enterprise trust. Rather than replicating hyperscalers’ global scale, they compete on integration depth and workload optimization. Many enterprises now view these relationships as part of a deliberate multi-cloud strategy, retaining hyperscalers for elastic or global workloads while moving stable or sensitive functions to smaller or private clouds.
The operational implications are significant. By distributing workloads across multiple environments, enterprises can minimize vendor lock-in, optimize for performance, and negotiate pricing more effectively. In some cases, the shift even enhances sustainability profiles, as smaller or regional providers design data centers around local renewable-energy sources and more efficient cooling systems.
Energy demand is another factor reshaping strategies. As Michael Dell recently noted, “At some point there’ll be too many of these [AI] data centers built, but we don’t see any signs of that yet.” His company’s AI-server sales rose nearly 70 percent last quarter, and Dell has partnered with smaller providers such as CoreWeave to supply high-density compute systems outside traditional hyperscale networks. Yet Dell also cautioned that power availability is emerging as a bottleneck: “Many customers will tell us, ‘Don’t deliver it until this day because we won’t have power in the building to support it.’”
The broader result is a more distributed and pragmatic cloud landscape. Enterprises no longer treat migration to a hyperscaler as the end goal but as one of several options in a modular infrastructure plan. Workload placement, rather than cloud adoption itself, is becoming the key performance metric. Analysts describe this evolution as a “cloud-appropriate” approach, where efficiency, governance, and cost predictability guide decisions more than brand or market share.
Enterprises that once saw hyperscale cloud as the only viable model are finding credible alternatives in smaller, more adaptable providers. The transition reflects a maturing view of infrastructure, valuing strategic control over one-size-fits-all scale. In that sense, the cloud era is diversifying, as enterprises reclaim agency over where and how their computing power resides.
