Google unveiled AI-driven data delivery upgrades and improved metadata generation for Google Cloud Storage during its Next ’26 conference. The update includes higher-performance object and Lustre storage, automated metadata tagging, MCP-enabled AI agent connectivity, zero-configuration management dashboards, consolidated activity views, and optimized batch operations. Google also expanded its capabilities for NetApp Volumes, GKE Filestore, and backup & data protection services.
Sameet Agarwal, VP/GM of Google Cloud Storage, and Senior Product Management Director Asad Khan stated: “We are upgrading every layer of our storage stack — performance, intelligence, and management — to ensure data matches the speed and usability of your AI models, applications and agents.”
Google’s key announcements cover Cloud Storage Rapid, Managed Lustre with Dynamic Tier, Smart Storage, and upgraded Storage Intelligence.
Cloud Storage Rapid
Built for extreme throughput, frequent I/O and sub-millisecond latency, Cloud Storage Rapid delivers durable, scalable object storage with cost-effective auto-tiering. Natively compatible with PyTorch and JAX, it contains two components:
Rapid Bucket runs on Google’s Colossus distributed storage architecture, delivering 15+ TB/s bandwidth, 20 million requests per second and ultra-low latency within a single zone. It supports high-performance gRPC and S3-compatible APIs, offering 5x faster checkpoint restoration and 3.2x quicker checkpoint writing than traditional object storage. Against regional storage, it cuts GPU idle time by 50% and accelerates data loading by 2.5x for multimodal training.
Google noted regional storage delivers cost-efficient performance for in-region workloads, but compromises availability and resiliency.
Rapid Cache, rebranded from Anywhere Cache, boosts bursty workload bandwidth with 2.5 TB/s aggregate read throughput for existing buckets without code changes. Its new ingest-on-write feature accelerates checkpoint restores by up to 2.2x, caching data during upload to enable instant cache hits on first reads.
Google Cloud Managed Lustre
Powered by DDN Lustre and EXAScaler, this parallel file system ranks among the cloud’s highest-performance managed Lustre services. It delivers up to 10 TB/s throughput — a 10x year-over-year increase and 4–20x faster than rival hyperscaler Lustre deployments. Built on C4NX VMs and Hyperdisk Expools, it achieves 2.6x faster checkpoint writing and restoration than other Google Cloud storage services.
The newly launched Dynamic Tier is priced at $0.06 per GB monthly. It delivers low-latency performance via persistent disks to eliminate performance degradation, offering simplified single-SKU billing without complex tiering rules.
Managed Lustre serves as a shared KV cache for AI inference, cutting time-to-first-token by over 40% compared to host-memory KV cache alone. DDN confirmed the solution improves overall inference throughput by 75%, positioning Managed Lustre as the ideal external KV cache storage.
Smart Storage
Smart Storage simplifies ML dataset selection with semantic filtering and eliminates custom retrieval pipelines. Key features include automated inline image and content annotations, where one-time write-time tagging benefits all downstream systems throughout a file’s lifecycle. Additionally, a Cloud Storage MCP server enables standardized protocol-based data reading, writing and analysis for AI agents.
Storage Intelligence
Google enhanced Storage Intelligence with zero-configuration dashboards that detect cost anomalies and integrate DSPM governance to identify storage security risks without manual setup. New object event and bucket activity tables support deeper cost auditing and streamlined operations. Upgraded batch tools allow bulk ACL and storage-class adjustments across billions of objects and multiple buckets.
Google rolled out three additional ecosystem upgrades:
NetApp Volumes: Flex Unified provides a unified cloud-data center storage pool supporting block (iSCSI, NVMe/TCP) and file (NFS/SMB) access. The new ONTAP mode enables native use of Terraform, Ansible and ONTAP APIs.
Filestore for GKE: AI developers can start with minimal 100 GiB shares and independently scale capacity and IOPS. Tighter Colossus integration enhances scalability and enterprise-grade capabilities.
Backup and DR: New agentic AI autonomously audits backup environments and fixes coverage gaps, with newly integrated GA support for AlloyDB and Filestore.
Separately, NetApp is adopting Gemini Enterprise internally to empower AI-driven sales and product development within its Google Cloud partnership.
Users can explore these new functions via the Google Cloud Storage console, with dedicated documentation and Next ’26 technical sessions available for further reference.
Beijing Qianxing Jietong Technology Co., Ltd.
Sandy Yang/Global Strategy Director
WhatsApp / WeChat: +86 13426366826
Email: yangyd@qianxingdata.com
Website: www.qianxingdata.com/www.storagesserver.com
Business Focus:
ICT Product Distribution/System Integration & Services/Infrastructure Solutions
With 20+ years of IT distribution experience, we partner with leading global brands to deliver reliable products and professional services.
“Using Technology to Build an Intelligent World”Your Trusted ICT Product Service Provider!
Your message must be between 20-3,000 characters!
English
