LAS VEGAS — Broadcom has expanded storage, security and compliance capabilities and has added Model Context Protocol (MCP) support beyond just a perfunctory effort for its cloud infrastructure platform for the public and private cloud.By seeking to improve compliance, security and other challenges when making the leap to creating and implementing agentic AI into developer pipelines and operations, Broadcom is at least addressing rampant concerns in this new world of IT that organizations are articulating.Broadcom executives described VMware Cloud Foundation (VCF) as providing IT-level Infrastructure as a Service (IaaS) and a core set of platform engineering and developer services during a conference call with analysts and journalists ahead of its annual VMware Explore user conference taking place here this week.Meanwhile, the responses to the AI-related concerns include — in addition to the MCP support — VCF Intelligent Assist (currently in tech preview) for improved debugging, diagnosis and resolution. The Multiaccelerator Model Runtime supports AI model deployments across AMD and NVIDIA GPUs without the need for refactoring AI, the company says. The MCP support serves as a standardized system to integrate AI assistants with internal content repositories and external tools such as Oracle, Microsoft SQL Server, ServiceNow, GitHub, Slack, PostgreSQL and others without building and maintaining custom connectors. Multitenant Models as a Service is designed to improve TCO and resource optimization for sharing of AI models between tenants or different user teams in their namespaces, while maintaining isolation for each tenant, the company says.Free of ChargeTo address the elephant in the room from the outset, users are not charged extra for the new AI, MCP and other new capabilities. This may appease some amid the controversy following the differences in licensing schemes after the Broadcom acquisition. It shows the large scope of a platform that addresses containers running side by side with virtual machines (VMs) through integrated management capabilities and can accommodate different Kubernetes versions. (Broadcom is a Kubernetes co-creator and remains a leading contributor.) It also includes storage components like vSAN, as well as security features through NSX and other Broadcom security tools, including integration with S3, as described below.Developers are ostensibly one of the key targets for the release. Broadcom executives said they intend to help developers “build modern applications at scale while maintaining IT controls.” The operational burden on developers, who spend 30-40% of their time on IT integration, is intended to be reduced. The VCF release improves on the integration of infrastructure services into the developer’s CI/CD pipeline, they said.“The winners in enterprises are the ones who are able to converge and look at their app data and AI as an integrated stack with complete common safety and governance,” Purnima Padmanabhan, vice president and general manager of Broadcom’s Tanzu Division, told analysts and journalists during the conference call.Additional New FeaturesOther features Broadcom described during the conference call include:Private AI as a Service: AI and generative AI (GenAI) are moving from pilot projects to production environments. Over 70% of enterprises plan to run AI models on a private cloud due to cost, privacy and compliance requirements. A challenge is the low utilization of GPUs, with over 80% of organizations seeing GPU capacity utilization below 35% in a shared model. Broadcom Private AI Services are now a native part of the VCF subscription. At the infrastructure level, VCF provides GPU reservations and bare-metal performance with the virtualization layer. Benchmarks show that VCF retains 99% of bare-metal performance with virtualization capabilities, such as vMotion and smart profile placement.With increased ransomware and silos, the threat surface and recovery time from cyber threats have increased. A key focus is on providing a cyber-resilient and compliant service from both an infrastructure and data perspective.Native S3 object storage: VCF will now have native S3 object storage as part of vSAN. This provides support for file, object and block data types, offering unified storage operations and simplified workflows. It is intended to support AI workloads and data recovery use cases with low latency and integration with Broadcom live recovery.Canonical partnership: A partnership with Canonical provides three key benefits. It allows customers to run Kubernetes images faster and lighter with chiseled Ubuntu containers, which have minimal package dependencies and a reduced attack surface. It also includes pre-compiled virtualized GPU drivers to streamline AI deployments in air-gapped environments by reducing dependency on external repositories. Enterprise-grade support for these containers is provided through Broadcom.AI services for developers: With the new VCF release, new capabilities are being added to help developers deploy and manage AI workloads. This includes support for MCP for integrating AI assistants with external tools without building custom connectors. An OpenAI-compatible API enables developers to connect to a standardized interface. A model runtime service helps data scientists create and manage model endpoints.Broadcom’s partnership with NVIDIA in support of VCF is being expanded to support its latest GPUs, including the NVIDIA Blackwell B200 Tensor Core GPU and the NVIDIA RTX 6000 Ada Generation.GitOps, Argo CD and Istio for secure application delivery: VCF streamlines secure application delivery by embedding GitOps, using Git as the source of truth for Kubernetes. Infrastructure and apps are stored as code in Git, with Argo CD automating consistent, auditable deployments. VCF also integrates Istio Service Mesh, offering zero trust networking, traffic control and rich observability for container communication. Enhanced Kubernetes multicluster management provides centralized control, visibility and policy enforcement via a single interface across environments.Broadcom is expanding the Broadcom Private AI ecosystem and partnering with AMD to advance enterprise AI infrastructure. Customers can leverage VCF with AMD ROCm Enterprise AI software and AMD Instinct MI350 Series GPUs to fine-tune large language models (LLMs), deploy Retrieval-Augmented Generation (RAG) workflows and run inference workloads within their data centers, addressing concerns related to privacy, choice, cost, performance and compliance.S3 storage expansion: S3 storage lends expanded object storage to vSAN. This capability will provide support for the full spectrum of data types, including file, block and object, and will offer unified storage operations with a simplified workflow. It is intended to be used for AI workloads, data recovery and other secondary storage use cases. Developers can directly deploy and consume this storage while removing the complexity of the underlying infrastructure. The VCF automation layer includes self-service capabilities for S3 storage, Broadcom says.Our take is that VCF’s new MCP features, at least on paper, are impressive. However, some of the features are still in development. It looks like a very solid effort, and Broadcom has provided ample support, patents and engineering as well. At the very least, it should be interesting to see how the cloud native community makes use of it, and we’ll see what their feedback is.The post VCF 9 Gets MCP Support, GPU Optimization at No Extra Cost appeared first on The New Stack.