ReleaseBytes
gcp Google Cloud Blog ·

Claude on Google Cloud for Enterprise Production AI

blogaigcpengineerhealthcarefinancemediagovernmentgcp-cloud-rungcp-gkegcp-vertex-ai
feature announcement

Google Cloud now offers Claude, a frontier AI model, for enterprise production, providing managed infrastructure, global reach, and compliance features. This integration aims to simplify AI deployment by allowing teams to use Claude like any other Google Cloud service, leveraging existing IAM and observability tools. The offering includes managed infrastructure, global and regional endpoints for low latency and data sovereignty, and optimizations for cost and performance.

  • →Fully managed infrastructure frees engineering time for feature development
  • →Global, regional, and multi-region endpoints ensure low latency and data residency
  • →Enterprise-grade security and data sovereignty inherited from Google Cloud
  • →Optimized cost and performance with native model and serving layer features
  • →Claude on Google Cloud integrates frontier AI for enterprise use
Features (4) ›
  • Fully managed infrastructure frees engineering time for feature development

    Claude on Google Cloud runs on managed infrastructure, handling compute provisioning, auto-scaling, load balancing, and failover. Teams can leverage this by using Claude as a Model-as-a-Service via REST/JSON endpoints, with an AnthropicVertex client for integration. This approach mirrors calling other Google Cloud services, using the same IAM policies, VPC controls, and Cloud Logging/Monitoring for observability.

  • Global, regional, and multi-region endpoints ensure low latency and data residency

    The platform offers three endpoint types: Global endpoints route requests to available AI compute capacity for maximum availability and cost-efficiency. Regional endpoints keep data within specific geographical boundaries for low latency and data-sovereignty needs. Multi-region endpoints provide U.S. or EU data residency with built-in resilience against regional outages.

  • Enterprise-grade security and data sovereignty inherited from Google Cloud

    Claude on Agent Platform inherits Google Cloud's security posture, supporting FedRAMP High and HIPAA compliance for regulated industries. VPC Service Controls allow organizations to define perimeters around Agent Platform resources, preventing data exfiltration, while IAM-native access control manages Claude endpoints similarly to other Google Cloud resources.

  • Optimized cost and performance with native model and serving layer features

    Claude's native features, such as prompt caching (reducing latency and cost), streaming responses, extended context windows (up to 1M tokens), and adaptive thinking, are fully supported. Google Cloud's serving infrastructure adds capabilities like batch prediction for asynchronous workloads and provisioned throughput for guaranteed inference capacity.

Notes (1) ›
  • Claude on Google Cloud integrates frontier AI for enterprise use

    Claude is now available on Google Cloud, designed for enterprise production demands such as managing accelerators, ensuring low latency globally, handling regulated data in-region, and serving long-context requests reliably. The integration treats Claude calls as standard Google Cloud service operations, utilizing existing Identity and Access Management (IAM), VPC Service Controls, and observability tools.

Read the original announcement →

https://cloud.google.com/blog/products/ai-machine-learning/claude-at-scale-on-google-cloud-frontier-ai-built-for-enterprise-production/

Related releases