Google Cloud Modernize puts its migration agents in one hub

Google has folded the products it sells for moving businesses onto its cloud into one portfolio, and put an AI agent in front of them. Google Cloud Modernize was announced on the company's cloud blog by Souvik Choudhury, senior director of product management, and Tom Nikl of the cloud modernisation and migration team.

The portfolio gathers Migration Center, Google Cloud VMware Engine, Google Cloud Mainframe Modernization and a new EKS-to-GKE Migration Agent under a single banner, and Google is promising agentic capabilities across infrastructure assessment, platform modernisation and application modernisation. The pitch is blunt: enterprises have multi-year roadmaps they would rather not spend multi-year budgets on.

A hub that reads your source code

Sitting at the middle of the offering is Modernization Hub, a new in-console experience where developers and architects can analyse source code, map out dependencies and accelerate modernisation work on Java, .NET and mainframe applications. It is the part of the announcement aimed squarely at the people who have to do the work rather than the people who sign it off.

Bigger machines for agent workloads

Google is also refreshing the hardware underneath, on the argument that AI agents querying backend systems will hit I/O bottlenecks that older instance families cannot absorb. The X5 Series is now generally available with single-node 43 TiB memory configurations, lifting a previous 29 TiB ceiling that forced enterprises to split ERP estates across nodes.

The M4N Series is also generally available, pairing 26.57 GiB of RAM per vCPU with Hyperdisk Extreme. Google says that stops customers overprovisioning cores just to reach a memory target, and claims licensing savings of more than 20% for Oracle and other core-licensed databases. For storage-heavy work, Z4D is generally available and Z4M is in preview, offering up to 84,000 GiB and 168,000 GiB of local NVMe respectively, 400 Gbps networking on both, and RDMA support on Z4M.

Cost modelling moves into a chat window

Assessment is where the AI branding does the most work. New Gemini capabilities in Migration Center power an Agentic Quick Estimator, now generally available, which turns VMware inventory exports such as RVTools reports into total-cost-of-ownership projections for Compute Engine. Teams can then interrogate those numbers in a chat interface, testing multi-region footprints or comparing bring-your-own-licence terms against pay-as-you-go. Google is also offering a free modernisation assessment through its Rapid Migration and Modernization Program.

VMware customers who are not ready to re-architect get a self-managed option on Bare Metal Z3 shapes running VMware Cloud Foundation 9.1, wired into Compute Engine, Google Kubernetes Engine, BigQuery and Gemini Enterprise through native global VPC links, so agents can be grounded in operational data without code changes.

Our opinion

Consolidation announcements are usually a packaging exercise, and there is a reading of Modernize in which Google has simply drawn a box around things it already sold. The interesting detail is that the box has an agent in it. The EKS-to-GKE Migration Agent is the clearest admission yet that Google is willing to automate the boring half of a competitor's migration, because whoever runs the container transition also owns the Kubernetes estate afterwards.

The memory and storage numbers are the practical story. A 29 TiB to 43 TiB jump on a single node, and claims of more than 20% off core-based licensing, are the sort of arithmetic that gets a modernisation project funded, because they attack the two costs that stall these programmes: re-architecting around memory limits, and paying per-core prices while doing it. Google is betting that enterprises will accept AI agents in the assessment phase before they accept them anywhere near production, and on the evidence of this launch, that is exactly where it has put them.