hcidude's blog

VMware Explore 2026 has arrived in Las Vegas, and this year’s event feels very different from the VMware conferences many of us remember.

The big theme is no longer simply virtualization, Kubernetes or even private cloud.

It is AI.

More specifically, Broadcom is positioning VMware Cloud Foundation as the foundation for the Private AI Cloud – an environment where enterprises can run traditional applications, Kubernetes workloads, AI inference and autonomous AI agents on the same infrastructure.

But AI is only one part of the story.

Looking across the announcements from VMware Explore 2026 and the latest VMware Cloud Foundation releases, several major topics emerge:

• Private AI Cloud
• AI infrastructure and VMware AI Factory
• AI models and model sharing
• Agentic AI
• AI security and identity
• AI-ready data
• Kubernetes and VKS
• Private cloud automation
• GitOps
• Infrastructure efficiency
• Operations and observability
• Open source and software supply chain
• VMware Cloud Foundation 9.1.1

And the last one is particularly interesting because VCF 9.1.1 became generally available during Explore 2026.

The Big Picture

Broadcom’s messaging around Explore 2026 can essentially be summarized in one sentence:

Build a modern private cloud that can run any workload – including production AI – while keeping data, security and operational control inside the enterprise.

VMware Cloud Foundation 9.1 provides the foundation for this strategy.

And now VCF 9.1.1 builds on that foundation.

The result is a platform story that increasingly looks like:

Infrastructure -> Private Cloud -> Kubernetes -> AI -> Agents -> Data -> Security -> Automation

Broadcom’s VMware Private AI Cloud announcement makes this direction explicit: AI inference, agentic applications and traditional enterprise workloads are intended to run together on a single private cloud platform.

1. VMware Private AI Cloud

The headline topic is undoubtedly VMware Private AI Cloud.

Broadcom’s argument is that enterprises increasingly want to run AI where their data already lives.

That means AI doesn’t necessarily belong exclusively in a public cloud.

Private infrastructure can provide:

• Data sovereignty
• Privacy
• Security
• Predictable infrastructure costs
• Control over AI models
• Control over GPU infrastructure
• Integration with existing enterprise applications

Private AI Cloud is therefore not presented as a separate AI product.

Instead, it is an extension of the private cloud model.

The idea is to run:

VMs + Kubernetes + AI inference + AI agents

on the same underlying platform.

This is probably the single most important strategic message coming out of Explore 2026.

2. VMware AI Factory

Getting AI into production is considerably more complicated than running a virtual machine.

Organizations need to prepare hardware, GPUs, networking, storage, Kubernetes, AI runtimes and models before they can even deploy an application.

VMware AI Factory is Broadcom’s answer to that problem.

It is positioned as the software-defined foundation of VMware Private AI Cloud and focuses on automating the journey from infrastructure to production AI.

The focus is not just Day 0 deployment.

Broadcom is also emphasizing Day 2 operations, infrastructure lifecycle management and AI economics – including visibility and control over AI token consumption.

The important concept here is:

from bare metal to model

Broadcom says AI Factory can automate hardware provisioning, software enablement and lifecycle management, with the goal of reducing the journey from bare metal to the first deployed AI model from weeks to hours.

That is a very different proposition from simply installing a GPU driver into an ESXi host.

3. AI Models as a Service

Another major topic is the model layer.

VMware Cloud Foundation is being positioned as a platform on which enterprises can deploy and consume different AI models.

Broadcom announced validated support for a growing set of models and says VCF can run more than 150 open source models using vLLM as the default model runtime.

The important distinction is that VMware isn’t trying to become an AI model provider.

Instead, VCF becomes the infrastructure and operational platform underneath the models.

This gives enterprises the ability to choose models according to:

• Security
• Privacy
• Cost
• Performance
• Licensing
• Domain specialization
• Data sovereignty

4. Multi-Tenant AI Model Sharing

VCF 9.1.1 takes this idea one step further.

One of the new capabilities is Multi-Tenant Model Sharing.

Instead of every business unit or tenant deploying its own independent model stack, models can be shared securely between tenants or lines of business while maintaining data privacy.

This matters because duplicated AI model infrastructure can become extremely expensive.

Imagine ten business units each running their own copy of a model.

10 model runtimes + 10 GPU allocations + 10 operational stacks

Model sharing creates the possibility of:

1 shared model service + controlled tenant access

Broadcom specifically highlights lower TCO, reduced operational complexity and tenant-isolated access controls as benefits.

For me, this is one of the more practical additions in VCF 9.1.1.

5. Agentic AI

The next evolution beyond generative AI is agentic AI.

Instead of simply answering questions, an AI agent can perform actions.

• It can access data.
• It can call APIs.
• It can invoke tools.
• It can modify systems.
• It can potentially execute entire business processes.

That changes the infrastructure and security requirements dramatically.

Agentic AI is therefore one of the major themes at Explore 2026.

The question is no longer simply:

„Can we run an LLM?“

It becomes:

„How do we safely allow an autonomous system to perform actions inside our environment?“

6. AgentMinder

AgentMinder is Broadcom’s approach to AI agent governance and runtime control.

The concept is relatively straightforward:

An AI agent needs an identity, just like a user or application.

But identity alone isn’t enough.

AgentMinder evaluates an agent’s:

• Identity
• Mission
• Intent
• Context
• Authorized tools
• Authorized resources
• Current risk

before allowing actions to take place.

This creates a fundamentally different security model for autonomous applications.

Instead of simply asking:

„Who are you?“

the platform can also ask:

„What are you trying to do, and are you allowed to do it?“

Broadcom combines AgentMinder with VMware vDefend and Avi Load Balancer to provide multiple layers of security around agentic AI workloads.

7. AI-ready Data

AI is only as useful as the data it can access.

That makes the data layer another major Explore 2026 topic.

Broadcom announced new AI-ready data foundations for VMware Tanzu Platform.

The objective is to combine governed AI agents with governed enterprise data inside the private cloud.

This is intended to address several problems at once:

• Data leakage
• Data sovereignty
• Data quality
• Data governance
• Unexpected cloud egress costs
• Incorrect AI responses caused by poor or uncurated information

The idea is to make enterprise data available to AI agents without requiring organizations to move all of that data into an external AI environment.

8. Tanzu and the Application Platform

Tanzu is becoming an increasingly important part of the overall architecture.

The role of Tanzu is expanding beyond traditional cloud-native application development.

It is now being positioned as the application and agent platform for Private AI Cloud.

That brings together:

• Applications
• Kubernetes
• AI agents
• Data
• APIs
• Platform engineering
• Developer self-service

This is important because infrastructure alone doesn’t create an AI platform.

Developers need somewhere to build and deploy their applications and agents.

Tanzu provides that layer.

9. Kubernetes and VKS

Despite the enormous focus on AI, Kubernetes hasn’t disappeared from the VMware story.

Quite the opposite.

Kubernetes is becoming one of the fundamental execution environments for modern applications and AI workloads.

VCF is therefore increasingly being positioned around a common platform for:

Virtual Machines + Kubernetes + AI

VCF 9.1.1 continues that direction with the vSphere Kubernetes Service 3.7 Add-on Management Framework.

The framework provides clearer lifecycle management and support ownership for Kubernetes ecosystem add-ons such as networking, observability, security and storage.

This matters enormously for existing VMware customers.

Most enterprises are not going to throw away their existing VM estates because they want to adopt AI.

They need a platform that allows them to add modern workloads without creating another completely separate infrastructure silo.

10. VMware Cloud Foundation 9.1.1

One of the most important developments during Explore 2026 was the general availability of VMware Cloud Foundation 9.1.1 on September 3, 2026.

It is tempting to look at 9.1.1 as simply a maintenance release.

That would miss quite a lot.

VCF 9.1.1 adds functionality across several areas:

AI

• Multi-Tenant Model Sharing
• AI Assistant for VCF Operations
• Additional AI operational capabilities

Kubernetes

• VKS 3.7 Add-on Management Framework
• Improved Kubernetes observability

Automation

• Native GitOps service based on Argo CD as a Tech Preview
• Improved application delivery workflows

Operations

• AI-assisted troubleshooting
• Full-stack Kubernetes observability
• Enhanced security operations

Infrastructure

• EVPN architectural enhancements
• Compact VCF deployment option
• Improved lifecycle and management capabilities

Storage

• vSAN Object Storage as a Tech Preview

So 9.1.1 is much more than a collection of bug fixes.

It continues the transformation of VCF into a broader private cloud platform.

11. AI Assistant for VCF Operations

One of the more visible additions in VCF 9.1.1 is an AI Assistant for VCF Operations.

The idea is to give administrators a conversational interface for daily operational workflows.

Instead of navigating through multiple dashboards and manually correlating information, administrators can use an AI assistant to help investigate issues.

VCF 9.1.1 can use locally configured AI models, including models running through Private AI Services or a private Google Gemini instance.

This is an interesting development because VMware is now using AI not only as a workload that runs on VCF, but also as a capability that helps operate VCF itself.

That creates an interesting feedback loop:

AI workloads run on the private cloud.
AI also helps operate the private cloud.

The AI Assistant can also assist with management pack creation and correlate health information, configuration and logs to help diagnose problems.

12. GitOps Comes to VCF Automation

Another significant VCF 9.1.1 announcement is the introduction of a native GitOps service.

The service integrates Argo CD into VCF Automation and is available as a Tech Preview.

The goal is to make GitOps part of the private cloud consumption model rather than requiring platform teams to build and maintain separate integrations.

Users can self-service Argo CD instances, authenticate through VCF Automation and attach vSphere Namespaces and VKS clusters as deployment targets.

That creates a workflow along the lines of:

Git -> Argo CD -> Kubernetes / VKS -> Application

with VCF Automation providing the surrounding organization, identity and service management.

This is an important step toward a true platform engineering model.

13. Operations and Observability

VCF 9.1.1 also strengthens the operations story.

As environments become more complicated, administrators need visibility across:

• VMs
• Kubernetes
• Containers
• AI workloads
• Networks
• Storage
• Applications

Traditional VM monitoring is no longer enough.

VCF Operations now adds deeper Kubernetes visibility and AI-assisted troubleshooting.

The platform can correlate health alerts, configuration and logs and provide real-time Kubernetes operational visibility.

This is becoming increasingly important because AI infrastructure introduces another layer of complexity.

• A GPU can be underutilized.
• A model can be overloaded.
• A Kubernetes workload can be waiting for resources.
• A network can become the bottleneck.
• A storage system can limit inference performance.

The operations platform needs to understand these relationships.

14. Security for the AI Era

AI also changes the security model.

There are now additional components to protect:

• LLMs
• AI agents
• MCP servers
• APIs
• AI tools
• Kubernetes workloads
• Data stores
• Model endpoints

Broadcom is therefore putting significant emphasis on security around Private AI Cloud.

The broader VMware security story includes AgentMinder, VMware vDefend and Avi Load Balancer.

The objective is to provide security across multiple layers rather than relying on a single AI-specific security mechanism.

The overall direction is clear:

AI security needs to become part of the infrastructure platform.

15. Infrastructure Efficiency

There is another topic that shouldn’t get lost behind the AI announcements:

cost.

AI infrastructure is expensive.

GPUs are expensive.

Power is expensive.

Memory and storage are expensive.

And public cloud AI workloads can quickly generate significant consumption costs.

VCF therefore focuses heavily on infrastructure efficiency.

The broader Private AI Cloud strategy explicitly addresses hardware CapEx, operational complexity and token economics.

Production AI needs to be economically sustainable.

16. Networking and EVPN

Networking is another area that becomes more important as VCF evolves into a larger private cloud platform.

VCF 9.1.1 introduces EVPN architectural enhancements.

Among the changes are direct VXLAN tunnels between transit gateways and leaf switches for East-West traffic and additional networking services such as NAT, NSX Load Balancing, Avi Load Balancing and DHCP relay in EVPN VXLAN distributed connectivity mode.

This is particularly relevant for organizations building larger-scale data center fabrics and integrating VMware environments with physical networking infrastructure.

17. vSAN Object Storage

VCF 9.1.1 also continues the expansion of vSAN beyond traditional VM datastore use cases.

vSAN Object Storage is available as a Tech Preview.

Object storage is particularly relevant for modern applications and AI because many AI pipelines consume large volumes of unstructured data.

That includes:

• Datasets
• Documents
• Images
• Video
• Model artifacts
• Training data
• Logs

Adding object storage closer to the VCF infrastructure therefore fits naturally into the Private AI Cloud strategy.

18. Open Source and TrueSource

One topic that sits slightly outside the traditional VMware story is TrueSource by Broadcom.

TrueSource focuses on commercially supported and verifiably built open source software.

The portfolio includes coverage for:

• Spring
• Java
• Python
• Node.js
• Container images
• PostgreSQL
• RabbitMQ
• MySQL
• Valkey

This is particularly relevant in an AI world where developers are generating more code, consuming more open source packages and deploying applications faster than ever.

The software supply chain therefore becomes another part of the security story.

19. Hybrid Cloud

Private AI does not necessarily mean isolated infrastructure.

Broadcom continues to support hybrid deployment models.

The important idea is that organizations should be able to choose where a workload runs based on:

• Cost
• Performance
• Data sovereignty
• Availability
• Capacity
• Regulatory requirements
• Hardware availability

Private infrastructure can therefore be the primary AI platform while public cloud environments remain an extension of that architecture where appropriate.

The important part is consistency.

The private cloud and the public cloud should not require completely different operational models.

20. What Does This Mean for VMware Administrators?

This may be the most important question for the VMware community.

The VMware administrator of the future will probably need to understand considerably more than ESXi and vCenter.

The skill set is expanding toward:

• VMware Cloud Foundation
• Kubernetes
• GPUs
• AI infrastructure
• Automation
• APIs
• GitOps
• Platform engineering
• Security
• Observability
• Data services

That doesn’t mean virtualization is going away.

It means virtualization is becoming one component of a much larger platform.

The traditional VMware administrator is evolving into something closer to a private cloud platform engineer.

Putting It All Together

The announcements at VMware Explore 2026 make much more sense when viewed as one architecture rather than a collection of individual product announcements.

At the top are the applications that ultimately consume the platform.

Traditional enterprise applications continue to run alongside a new generation of AI agents.

Those workloads are supported by the Tanzu Platform, which brings together application development, Kubernetes, data and AI services.

Below that sits the Private AI Cloud layer, providing the environment in which AI inference and other cloud workloads can operate under enterprise control.

The underlying execution environments remain familiar:

VMs + Kubernetes + AI Inference

All of these workloads are ultimately delivered through VMware Cloud Foundation.

Below VCF, Broadcom is adding the capabilities required to build and operate the infrastructure:

AI Factory + Automation / GitOps

And underneath everything is the physical infrastructure:

CPU + GPU + Memory + Storage + Network

This is the important architectural shift.

VCF is no longer simply the layer that provides virtualization.

It becomes the platform connecting physical infrastructure with applications, Kubernetes, AI and automation.

And around the entire architecture sits the security and operations layer:

AgentMinder + vDefend + Avi + VCF Operations

This is the bigger story behind the individual Explore announcements.

What Changed with VCF 9.1.1?

If I had to summarize the difference between the original VCF 9.1 release and 9.1.1 in a few words, I would put it this way:

VCF 9.1

Build the modern private cloud.

VCF 9.1.1

Make that private cloud smarter, more automated and more AI-aware.

VCF 9.1 established many of the architectural foundations.

VCF 9.1.1 starts adding the operational experiences around them.

• AI-assisted operations
• Multi-tenant model sharing
• GitOps
• Better Kubernetes management
• More observability
• More automation
• Additional networking capabilities
• Object storage
• Continued security improvements

That makes 9.1.1 particularly interesting because it starts connecting the individual pieces into a more complete operating model.

My Take

For me, the most interesting thing about VMware Explore 2026 isn’t any individual product announcement.

It is the direction of the platform.

VMware is moving further away from being perceived simply as a virtualization stack.

• VCF is becoming the foundation.
• Kubernetes becomes another workload layer.
• AI becomes another workload class.
• Tanzu becomes the application and agent platform.
• Security becomes integrated throughout the stack.
• Automation becomes the mechanism that turns all of this into a private cloud service.

And VCF 9.1.1 adds another important dimension:

the platform itself is becoming AI-assisted.

That is a significant evolution.

The real test will now be execution.

• Can Broadcom make this architecture simple enough to operate?
• Can customers justify the economics?
• Can enterprises run AI securely at scale?
• Can platform teams operate VMs, Kubernetes, GPUs, AI agents and data services through one coherent platform?
• Can VMware maintain the operational simplicity that made virtualization so successful while adding all this new functionality?

Those are the questions I will be watching after VMware Explore 2026.

But one thing is already clear:

VMware Explore 2026 isn’t really about virtualization anymore.

It is about what comes after virtualization – and how the private cloud becomes the platform for the AI era.

---

VMware Explore 2026 took place August 31 – September 3, 2026, in Las Vegas. VMware Cloud Foundation 9.1.1 became generally available on September 3, 2026, during the conference. The release adds new capabilities across AI, Kubernetes, operations, GitOps, networking and storage while continuing the broader VCF private cloud strategy.