Broadcom has released VMware Cloud Foundation 9.1, an update to its private cloud infrastructure platform that targets enterprises running AI workloads on their own hardware rather than public cloud. The release builds directly on VCF 9.0 and introduces several changes aimed at reducing costs and cutting the operational overhead that has historically made private cloud maintenance labor-intensive.
The most immediately practical addition is NVMe memory tiering. By pairing conventional DRAM with NVMe memory, organizations can expand their effective memory pools without purchasing additional hardware. Combined with extended vSAN deduplication and compression, the update gives IT teams a path to reduce storage costs using infrastructure they already own. For enterprises managing tight capital budgets while AI workload demands keep climbing, that combination addresses a real constraint.
On the maintenance side, VCF 9.1 extends live patching to cover up to 80% of use cases without requiring host evacuation. In practical terms, that means fewer maintenance windows, less disruption to running workloads, and continuous compliance enforcement for environments handling sensitive data. Cluster upgrades now run four times faster than in previous versions, and fleet capacity has doubled, which matters for organizations scaling AI infrastructure across larger deployments.
The release also adds real-time observability, giving teams actionable insight from telemetry data as it arrives rather than after the fact. For organizations trying to manage growing AI infrastructure without proportionally growing their operations staff, that visibility layer reduces the manual monitoring burden considerably.
A Broadcom representative noted that memory tiering alone has generated enough efficiency in some customer environments to offset the cost of VCF licensing entirely. That kind of concrete financial case tends to move budget conversations faster than broader platform arguments.
The update also reflects a wider pattern Broadcom is observing across its enterprise customer base. Organizations that spent the past several years moving workloads to public cloud are now rebuilding private infrastructure strategies, particularly for AI applications where data governance, latency, and cost predictability push back against pure cloud models. VCF 9.1 sits squarely in that context, supporting AMD, Intel, and NVIDIA processors within the same environment and bringing Kubernetes-native capabilities alongside traditional virtualization.
For channel partners, the migration journey from basic virtualization to full private cloud represents a substantial services opportunity, covering assessment, design, deployment, and ongoing managed operations.
