VMware Cloud Foundation (VCF) 9.1 is built as a private cloud platform to run production AI, modern apps, and traditional workloads on infrastructure you own and govern, all under one operating model.

It is engineered around three pressures most IT teams are facing in 2026:

• AI demand: AI is being embedded into everyday apps and operations. IDC forecasts worldwide AI spending will grow at a 31.9% CAGR through 2029, reaching $1.3 trillion and exceeding 26% of total worldwide IT spending. VCF 9.1 is designed to host these AI-enabled workloads efficiently.
• Geopolitics and sovereignty: Data sovereignty rules, export controls on accelerators, and fragmented compliance regimes make workload location a board-level issue. VCF 9.1 lets you keep sensitive data and AI models on infrastructure you control, while still operating at cloud scale.
• Budget compression: Many organizations are expected to fund AI from flat or shrinking budgets. VCF 9.1 focuses on lowering cost per workload and simplifying operations so you can do more with the same footprint.

Key capabilities that support this include:

• Cost and density:
  • Enhanced NVMe Memory Tiering extends effective memory pools by tiering DRAM and NVMe, which is especially useful for memory-bound AI and database workloads.
  • Extended vSAN Dedup and Compression reduces cost per usable terabyte.
  • Topology Aware Scheduling places workloads with NUMA and accelerator locality in mind, improving performance for GPU and AI pipelines.
  • Ubuntu OS Enterprise Support broadens the supported workload ecosystem without adding operational overhead.
• Operational efficiency:
  • Real-Time Operational Observability turns telemetry into actionable insights.
  • Increased VMware vSphere Kubernetes Service (VKS) scale raises the Kubernetes ceiling for containerized workloads.
  • vSphere Elastic Provisioning enables zero-touch fleet expansion.
  • Expanded fleet size and upgrade scale allow a single VCF instance to govern larger estates with fewer maintenance windows.

The net effect is more workloads per rack, fewer operators per thousand workloads, and a measurable reduction in cost-to-serve, while keeping AI, container, and traditional workloads on a single, governed platform.

VCF 9.1 is designed to reimagine the private cloud as a single platform that can deliver VMs, containers, AI services, and data services with the same level of consistency.

For application and platform teams, this shows up in a few concrete ways:

• Faster provisioning for both VMs and Kubernetes:
  • VKS and VM Fast-Deploy shorten time-to-running for Kubernetes clusters and virtual machines, so teams can move from request to usable capacity more quickly.
  • Simplified Container-as-a-Service turns VKS into a self-service surface for developers, while platform teams keep central guardrails.
• Integrated data and middleware services:
  • Native Object Storage (S3-compatible, in tech preview in 9.1.x) brings object storage directly into the platform, which is particularly relevant for AI and data-intensive workloads.
  • Tanzu Marketplace integration provides a curated path to certified middleware and data services.
  • SQL Server DBaaS elevates SQL Server to a first-class, managed database service within the private cloud control plane.
• AI-specific visibility and control:
  • Private AI Model and GPU Metrics expose utilization, memory pressure, and model-level visibility in the same console used for the rest of the estate.
  • Live Application Stack Blueprints let teams version and redeploy entire application topologies as code, from legacy three-tier apps to RAG pipelines running on the latest AMD GPUs.

For you, this means one platform and one operating model that can handle:

• Traditional three-tier enterprise applications
• Modern containerized microservices
• Data services and databases, including SQL Server as DBaaS
• AI pipelines and inference workloads using GPUs and large in-memory datasets

This unified approach helps you avoid building separate silos for AI, containers, and legacy apps, and instead run them under a single governance and operations framework.

VCF 9.1 treats resilience, security, and ecosystem integration as core architectural properties, not add-ons. This is especially relevant if you are running sensitive data and high-value AI models.

Security and resilience

• Ransomware and disaster recovery:
  • vSAN for Recovery and On-prem Ransomware Recovery provide a sovereign, in-platform path to recover from destructive attacks without relying on external escrow.
  • CrowdStrike EDR integration for Ransomware Recovery adds more choice for endpoint detection in the recovery workflow.
• Encryption and patching:
  • Encrypted vMotion with Intel QAT offloads cryptography to dedicated silicon, removing much of the historical performance tax of end-to-end encryption in motion.
  • Live Patching for TPM-enabled hosts reduces one of the largest sources of unplanned downtime in large estates by allowing security updates without taking hosts offline.
• Compliance and lateral security:
  • Continuous Compliance Enforcement turns compliance from a periodic audit into a runtime guarantee.
  • Self-Service Lateral Security and Automated Load Balancing put micro-segmentation and traffic management in application teams’ hands, while still being governed by central policies.

Ecosystem and multi-vendor integration

• Hardware and accelerators:
  • Enhanced DirectPath I/O for the latest AMD GPUs delivers near-bare-metal accelerator performance on the newest AMD silicon, which is important for AI training and inference.
• Networking fabrics:
  • Unified EVPN with Arista, Cisco, and SONiC provides a consistent overlay fabric across three dominant data center networking stacks, including the open-source SONiC option.
  • This gives network teams one operating model regardless of which switch vendor is in which rack, and helps shorten the time to onboard new sites or integrate acquired environments.
• Cloud-native ecosystem:
  • New VKS Reference Architectures with cloud-native ISVs give platform teams validated starting points for Kubernetes and AI-centric designs.

For existing customers on VCF 9.0, there is a supported in-place upgrade path to 9.1, with a dedicated upgrade guide and compatibility matrix available on the Broadcom support portal. Organizations earlier in their journey can use the VCF 9.1 Reference Architecture library to design validated AI, container, and traditional workload environments that align with their sovereignty and security requirements.