Google Cloud makes an Anthropic-run option for Claude generally available
Michael Gerstenhaber, Friday: the new deployment puts Anthropic as the data processor, on Google Cloud infrastructure, with day-zero access to Claude Managed Agents, skills and code execution. The existing Google-managed option keeps Google as the processor, with VPC Service Controls, customer-managed keys, and residency in the US, EU and Asia-Pacific. Both, the post says, get the same models on day zero, at the same price, on the same Google Cloud bill. No rate card is printed.
Feature2 minIn-depth
Priya Shah2 min read
- Google Cloud
- Anthropic
- Claude
- Agents

Photo: Google CloudSource page
Google Cloud on Friday made a second way to run Claude generally available. Michael Gerstenhaber, vice president of product management for Gemini at Google Cloud, wrote that an Anthropic-managed deployment now sits beside the Google-managed option the company already sells. Both, he wrote, deliver the same frontier models on day zero, at the same pricing, billed through Google Cloud. The difference he names is who acts as the data processor, and whether new Claude platform features are there on the day Anthropic ships them. The post does not print a price, a token rate, or a list of regions for the Anthropic-managed stack.
Anthropic holds the keys to the new option
On the new option, Anthropic operates the stack as the data processor, on Google Cloud infrastructure, under Anthropic’s commercial terms. Customers still authenticate with Google Cloud IAM, keep Cloud Audit Logging, and draw the spend down against existing Google Cloud billing and eligible commitments. Gerstenhaber says this path delivers Claude models and Claude platform features on day zero. The features he names are Claude Managed Agents, which he marks as beta, plus skills and code execution. A team that wanted those features without leaving a Google Cloud account is the buyer this paragraph is written for. The post does not say Anthropic can see customer content beyond what a data processor’s terms already cover, and it does not attach a compliance certification to this option.
The Google-managed option is the one Gerstenhaber describes for stricter governance. Google Cloud remains the data processor, under the customer’s existing Google Cloud agreement. He lists VPC Service Controls, customer-managed encryption keys, and data residency in the United States, the EU and Asia-Pacific. New Claude platform features, on this path, arrive once they have been integrated into Google’s infrastructure, not on Anthropic’s ship date. The same paragraph says teams on this option get native ties to Google Cloud services, and names BigQuery and Model Armor. Billing and eligible commitment drawdown apply here too. The post’s comparison table gives both columns “available on day 0” for the models themselves, and gives first-party tools and betas to the Anthropic-managed column on day 0 and to the Google-managed column only after integration.
The models named at the bottom of the post are Claude Fable 5.1, Sonnet 5.5, Haiku 5.5 and Opus 5.5. Gerstenhaber says new platform features are available immediately on the Anthropic-managed option and roll out on the Google-managed option once integrated. He does not date that integration. He also says developers get Google Cloud IAM and Marketplace billing either way. What Friday’s post adds is a choice of processor for a set of models Google Cloud was already offering. It does not say the weights moved, and it does not say the price changed. A buyer who needs the residency list, the customer-managed keys and the VPC boundary stays on the Google-run option, and waits for Managed Agents. A buyer who wants those features now takes Anthropic as the processor, on Google’s machines, and does not get the residency line Gerstenhaber attached only to the Google-managed column.


