Broadcom launches VMware Private AI Cloud to secure on-premises inference

The launch of VMware Private AI Cloud signals a strategic shift toward localized AI infrastructure as enterprises prioritize data sovereignty over public cloud convenience. By integrating AgentMinder and zero-trust controls, Broadcom is addressing the specific security risks of autonomous agents that traditional firewalls cannot manage. This move challenges the dominance of external API-based AI models by offering a validated on-premises alternative for regulated industries. As streaming and media organizations scale AI-driven personalization, the ability to run 150+ models without data leaving the enterprise becomes a critical cost and compliance lever. Watch for adoption rates of VMware Cloud Foundation 9 as a proxy for enterprise readiness to move agentic AI into live production environments.

Broadcom has positioned VMware Private AI Cloud as a direct counter to hyperscaler dependency for enterprise AI workloads. The platform builds on VMware Cloud Foundation 9, which Broadcom made generally available in May 2025 with integrated AI infrastructure capabilities, including support for NVIDIA-accelerated servers and unified management across on-premises environments. The company's broader AI Factory initiative, announced alongside VMware Cloud Foundation 9 at VMworld 2025, targets enterprises that need to run inference at scale without sending proprietary data to public cloud endpoints. Ram Velaga, Broadcom's senior vice president and general manager of the Cloud Infrastructure Business Group, has framed the offering as a way to reduce per-token inference costs by 40 to 60 percent compared to public cloud API pricing for sustained workloads.

The competitive landscape for private AI infrastructure has intensified considerably. NVIDIA announced its own AI Enterprise platform expansion in March 2025, adding support for both VMware and Red Hat virtualization layers to give enterprises more deployment flexibility. Meanwhile, Dell Technologies launched its Dell AI Factory with VMware in late 2024, bundling validated designs that pair PowerEdge servers with VMware's AI stack to reduce deployment friction. On the security side, Broadcom's zero-trust approach aligns with NIST's updated guidance on AI system security published in early 2025, which recommends isolating model inference from untrusted network paths, a requirement that on-premises architectures inherently satisfy more cleanly than multi-tenant cloud environments.

For streaming and media operators specifically, the technical economics of private inference matter as AI-driven personalization and content recommendation workloads scale. Amdocs reported in its fiscal 2025 results that media and entertainment customers were among the fastest-growing adopters of on-premises AI inference, citing data residency requirements in European and Asia-Pacific markets. The broader market reflects this trend: IDC projected in June 2025 that enterprise spending on private AI infrastructure would reach $42 billion by 2027, up from $18 billion in 2024, driven primarily by regulated industries including media, financial services, and healthcare. Broadcom's AgentMinder governance layer addresses a specific gap in this market by providing policy controls for autonomous AI agents, a capability that .