Google Cloud adds Anthropic-managed Claude deployment

Google Cloud has launched an Anthropic-managed deployment option for Claude models on its cloud platform, giving customers a second way to run Anthropic's models on Google Cloud.

The new option sits alongside Google Cloud's existing Google-managed deployment model. Both routes offer access to the same Claude models at launch and use the same Google Cloud billing arrangement.

Under the Anthropic-managed option, Anthropic operates the stack on Google Cloud infrastructure and acts as the data processor under its own commercial terms. Customers still use Google Cloud IAM for authentication, retain Cloud Audit Logging, and can consolidate spending within existing Google Cloud billing and eligible commitments.

By contrast, the Google-managed option keeps Google Cloud as the data processor under existing customer agreements. It is aimed at customers with governance, compliance and sovereignty requirements, and includes VPC Service Controls, Customer-Managed Encryption Keys, and data residency options in the US, EU and APAC.

Two models

The main difference between the two options is the operating model and when customers get access to Anthropic platform features. The Anthropic-managed route gives customers access to new Claude platform features as they launch, including Claude Managed Agents in beta, skills, and code execution.

Customers using the Google-managed route will still get access to the latest Claude models on the day they are released. However, additional Anthropic platform features will arrive later, once they have been integrated into Google-managed infrastructure.

The arrangement reflects growing demand from companies that want to standardise artificial intelligence workloads on a single cloud provider while choosing different operating and compliance models for specific applications. It also shows how cloud providers and model developers are expanding commercial structures around generative AI services as customers seek clearer control over data processing and regulatory obligations.

Both deployment models carry standard pricing and draw down eligible Google Cloud commitments. That means customers can choose between Anthropic-managed and Google-managed processing without changing the core billing framework.

Model access

The Claude models available on Google Cloud include Claude Fable 5.1, Sonnet 5.5, Haiku 5.5, and Opus 5.5. These models are available on day zero through both deployment options.

The Google-managed option also includes native integrations with services such as BigQuery and Model Armour. These integrations are part of Google Cloud's existing approach to embedding third-party and proprietary models within its wider cloud software portfolio.

The Anthropic-managed route gives organisations another way to access the latest Claude releases while keeping procurement and identity management within Google Cloud systems. For customers already using Google Cloud for infrastructure and software services, that may reduce the need to manage separate billing channels for AI model access.

At the same time, the split deployment structure highlights a broader issue in enterprise AI adoption: whether customers want direct access to a model provider's newest tools or prefer a tighter compliance framework operated by a cloud provider. The two options are designed to address those different priorities without changing the underlying cloud environment.

In practice, the difference comes down to who acts as the customer's data processor and whether immediate access to newly launched Claude platform features is required.