Google Launches Gemini Enterprise Agent Platform for Enterprises

Google Cloud launched the new Gemini Enterprise Agent Platform, positioning it as the successor to Vertex AI and the central stack for enterprise agent development. The company announced a $750 million partner fund to accelerate agentic AI adoption, introduced security-focused agents, and disclosed a multi-year collaboration with Merck valued at up to $1 billion. Google also committed infrastructure capacity to startups such as Thinking Machines Lab using A4X Max VMs with GB300 GPUs.

The new platform consolidates model selection, model building, and agent orchestration into a single product. Key capabilities highlighted include:

• •Agent Studio low-code and visual tooling plus code-first workflows for production agent development
• •First-class access to over 200 models via Model Garden, including first-party Gemini variants and support for third-party models like Claude Opus, Sonnet, and Haiku
• •Built-in DevOps, orchestration, and observability for agent fleets, with governance controls, policy management, and role-based access
• •New security agents: Threat Hunting, Detection Engineering, and Third-Party Context to detect novel adversarial behavior and enrich security workflows

the company says all future Vertex AI roadmap items will be delivered through the Agent Platform, making migration and integration a planning priority for existing Vertex users. The platform emphasizes hybrid delivery, deep integration with Google Cloud services such as Kubernetes Engine and Spanner, and partner-driven proofs-of-concept and practice-building via the fund.

This is a strategic pivot from isolated model and pipeline tooling toward operationalized, multi-model agent fleets that act across enterprise systems. By packaging orchestration, governance, model choice, and security agents, Google is targeting enterprises that need robust controls before delegating outcomes to software agents. The $750 million partner fund and the headline $1 billion Merck engagement signal that Google expects agentic workflows to become a major enterprise transformation story, not a niche research topic. Compatibility with third-party models undercuts any single-provider lock-in narrative and frames Google Cloud as a platform play that stitches models, infra, and systems together.

Watch enterprise adoption signals from early anchor customers like Merck, partner uptake from systems integrators, and how effectively Google operationalizes governance and security at scale. Also track infrastructure shifts, including Google's use of A4X Max VMs and GB300 GPUs, since these affect cost, latency, and vendor competition for GPU supply and pricing.