If you logged into Google Cloud Console this week looking for Vertex AI, you didn't find it. As of May 21, 2026, Vertex AI no longer exists as a product. Google completed its migration to the Gemini Enterprise Agent Platform, and the old name is gone from the user interface. This isn't a rebrand. It's a hierarchy inversion that signals where enterprise AI development is heading.
The shift moves model training, deployment, and registry under an agent-first architecture. What used to be the primary container—the model—is now a sub-feature under agents. Google is betting its entire AI roadmap on agents being the primary unit of work, not the model. For CTOs and engineering leaders, this changes how you think about building on Google's stack.
What Actually Happened
Google announced the Gemini Enterprise Agent Platform at Cloud Next 2026 on April 23, but the console-level cutover only happened in the last 48 hours. The replacement is structural, not cosmetic.
What you knew as Vertex AI—Model Garden, Custom Training, AutoML, Model Registry, Endpoints, and Pipelines—now lives under a "Models" sub-menu inside the Agent Platform. A new primary "Agents" menu sits above that, containing Agent Garden (prebuilt templates), an MCP Server registry, Memory Bank for long-term context, Sessions for stateful interactions, and an Agent Registry that catalogs every agent, tool, and connector you build.
The most surprising part? The API stays the same. The endpoint remains aiplatform.googleapis.com, which was the name of the service before it was even called Vertex AI. If you have code that hits Vertex AI today, it keeps working without changes. To use the new agentic features, you enable a handful of new APIs like agentregistry.googleapis.com and modelarmor.googleapis.com, but the core connection stays where it was.
Here's the mapping:
| What you knew (Vertex AI) | Where it lives now (Agent Platform) |
|---|---|
| Model Garden | Models → Garden sub-menu |
| Custom Training, AutoML | Models → Training sub-menu |
| Model Registry, Endpoints | Models → Registry sub-menu |
| Pipelines | Models → Pipelines sub-menu |
| Vertex AI Agent Builder | Agents → primary menu |
| New (no Vertex equivalent) | Agent Garden, Memory Bank, Sessions, Agent Identity, Gateways, Topology |
Why This Is Bigger Than It Sounds
Making "Models" a child of "Agents" inside the console is Google putting a stake in the ground: agents are now the thing you build, and the model is one ingredient inside the agent. The choice of menu hierarchy is the clearest signal of where the platform's investment will go over the next 18 months.
Memory Bank, MCP Servers, Agent Registry, Topology, and Agent Evaluation are all features that have no equivalent in the Vertex AI era because they only make sense if agents are the primary object. Traditional machine learning tools are now nested inside an agent-first workflow.
Three Under-Reported Angles
1. Identity Propagation
Agent Identity gives each agent a cryptographic ID, and the platform can restrict an agent so it only reads what the specific user who triggered it is allowed to read. Model-only vendors like OpenAI and Anthropic don't have a clean equivalent to this.
If you're building an agent that needs to query a corporate data lake on behalf of different employees with different permissions, identity propagation is the kind of capability that closes the deal versus building on top of a raw API key. This matters for compliance, governance, and security in regulated industries.
2. The $750 Million Fund
Google announced a $750 million fund for partners building agents on the platform, alongside an Agent Marketplace seeded with named partners like Adobe and Atlassian. The platform's "Agent Gallery" needs content at launch, and Google is paying for it.
That's fine for now, but it tells you the ecosystem isn't organic yet. Solo builders who want to publish their own agent for free are competing for shelf space against partners with funded engineering teams. The marketplace is currently partner-first, not community-first.
3. Scale and Adoption
Google disclosed that its models are now processing more than 16 billion tokens per minute via direct API calls, up from 10 billion the previous quarter. Nearly 75% of Google Cloud customers are using its AI products. Those numbers tell me the platform has demand and Google is comfortable doubling down on infrastructure rather than only competing on raw model quality.
What This Means for Technical Leaders
The hierarchy inversion is the news. The rebrand is the symptom.
For CIOs and CTOs: If you're already building on Vertex AI, nothing breaks. Your code keeps working. But if you're planning net-new AI development on Google's stack, you need to think agent-first, not model-first. The platform's roadmap, feature set, and pricing structure are now optimized for agents, not traditional ML workflows.
For VPs of Engineering: The Agent Development Kit (ADK) is open-source and supports non-Google models, including OpenAI and Anthropic. You're not locked into Gemini if you want to build on the platform. This is a competitive advantage versus proprietary agent frameworks.
For Chief Architects: Identity propagation and auditable agent trails solve compliance problems that raw API keys don't. If you're in a regulated industry (finance, healthcare, government), this is worth a closer look. The platform can enforce user-level permissions across multi-agent systems, which is non-trivial to build yourself.
What This Means for Business Leaders
For CFOs: The $750 million fund signals Google is subsidizing early adopters. If you're evaluating enterprise AI vendors, this is leverage. Ask for co-development funding, reference architecture support, or extended pilots. Google needs partners to populate the Agent Marketplace, and that creates negotiating room.
For COOs: Agent-first architecture means you're buying automation systems, not models. The platform's value prop shifts from "better predictions" to "better workflows." If you're budgeting for AI in 2026-2027, align spend with process automation ROI, not model accuracy benchmarks.
For CMOs and CROs: The Agent Marketplace is partner-first today. If you're building customer-facing AI (chatbots, personalization, recommendation engines), ask whether you're competing with Adobe and Atlassian for platform visibility or whether you can partner with them for go-to-market leverage.
What This Means for Indie Builders
Don't switch to the Agent Platform just because Google made a noisy announcement.
The Vertex AI to Agent Platform shift adds value when your agent needs to talk to enterprise data with user-level permissions, or when you need an auditable trail across many agents running in production. For a single-purpose agent or a wrapper SaaS, the API overhead alone (a dozen-plus services to enable, IAM roles to configure, identity bindings to set up) is overkill.
If you're currently building with a single OpenAI or Anthropic API key, keep doing that. The Agent Platform is a hyperscaler stack: deeper governance and tools, higher friction to get started, and probably overkill unless you need identity-aware data access or you're pre-paying for compliance posture.
The Competitive Landscape
Google is the first hyperscaler to invert the hierarchy and make agents the primary container. AWS and Azure still treat models as the top-level object. OpenAI and Anthropic don't have platform-level identity propagation or agent registries—they're model vendors, not platform vendors.
This is Google's differentiation play: if you're building multi-agent systems that need to interact with enterprise data under user-level permissions, Google now has infrastructure that OpenAI and Anthropic don't. If you're building a single agent that calls a single model, Google's stack is heavier than you need.
The Agent Marketplace, $750M fund, and identity propagation are all aimed at enterprise buyers who need governance, compliance, and auditability. Solo developers and startups pay the complexity tax without getting the enterprise benefits.
Bottom Line
Vertex AI is gone. The Gemini Enterprise Agent Platform replaces it with an agent-first hierarchy, identity propagation, and a $750 million fund to seed the Agent Marketplace. The API stays the same, so existing code doesn't break. But the platform's roadmap is now optimized for agents, not traditional ML workflows.
If you're a CTO or VP of Engineering: Think agent-first for net-new development on Google's stack. The platform's feature set, pricing, and investment priorities are all aligned with multi-agent systems, not standalone models.
If you're a CFO or business leader: The $750M fund creates negotiating leverage for co-development deals. If you're evaluating enterprise AI vendors, this is the time to ask for funded pilots and reference architecture support.
If you're an indie builder: Don't migrate unless you need identity propagation or agent-level governance. The platform is overkill for single-agent use cases. OpenAI and Anthropic are still simpler for most startups.
Google made a bet that agents are the primary unit of work, and the console hierarchy proves it. Whether that bet pays off depends on whether enterprises buy into agent-first development or whether they keep treating models as the top-level primitive. Based on the 16 billion tokens per minute and 75% customer adoption, Google has demand. The question is whether the demand is for agents or whether it's for models that happen to live under an agent menu.
