What’s new in AI infrastructure and orchestration in September

We hereby declare September to be scalability month! As the world prepares for a surge of agentic fleets, we are shoring up our AI infrastructure and orchestration offerings to gracefully — and quickly — respond to that demand, all while maintaining workload isolation and security, and keeping costs in check. Read on to learn how these enhancements manifest across Google Cloud’s compute, network, storage, and orchestration offerings, plus new ways customers are using Google Cloud AI infrastructure, and third-party industry validation of our strategy.

Product, technology, and tools updates

Google Kubernetes Engine updates: The GKE team is all about improving the scalability of the platform, and in September, those improvements came in many shapes and sizes:

• New feature: Need an execution runtime with higher density for your agentic workloads? We engineered the new open-source GKE Agent Substrate to run millions of sandboxes with 10x higher density than standard container runtimes. Agent Substrate also delivers sub-500ms resume operations at over 500 suspend/resume activations per second with a native zero-trust kernel and network isolation. 

• Product update: GKE now has scale-to-zero capabilities built-in. No need to configure complex components to scale your workloads down, thanks to the HPA with the Autoscaling Metric and support for KEP-2021, which do the job for you, out of the box. Read the blog to learn more. 

• Product update: Further, the GKE HPA (with the above-mentioned Autoscaling Metric) now lets you scale up and down based on custom PromQL metrics, in addition to standard metrics, allowing you to trigger workloads according to conditions that are meaningful and unique to your business. Read more here. 

• New feature: Yet another scalability feature is GKE Pod snapshots, which lets you save the running state of your workload, including CPU and GPU memory, and restore it on demand. According to internal tests, GKE Pod snapshots can reduce AI inference start-up by as much as 89%. Learn more here.

• New migration tool: Finally, if you’ve always wanted to migrate your container workloads from AWS EKS to GKE but feared a daunting, high-friction engineering endeavor, we’ve just launched GKE agentic migration, a purpose-built agent plugin that replaces brittle, ad-hoc prompting with an AI-assisted migration pipeline protected by deterministic guardrails. Designed as a compilation of agent skills and a local Model Context Protocol (MCP) server, it uses AI to translate complex AWS EKS IaC and Kubernetes manifests directly into GKE landing zones. Get started with the onboarding guide.

• Feature updates: Reinforcement learning (RL) and evaluation workloads are a beast: In a standard agentic RL loop, an LLM policy generates actions like code snippets on GPUs and executes them inside isolated CPU sandboxes to observe a reward signal. However, when scaling up this loop to support tens of thousands of parallel rollouts, infrastructure bottlenecks emerge, for instance idle accelerators, image cardinality, and a saturated control plane. To help, we developed GKE Agent Sandbox optimized for RL, plus an Agent Sandbox RL orchestration SDK and native integrations for popular RL gyms and harnesses. All are now generally available, and you can learn more here.

Storage updates: AI trains and creates lots of data, and that data has to live somewhere — in block storage systems, file systems, object stores and databases. We announced enhancements to our storage portfolio to help this critical layer roll with the agentic punches:  

• Product update: Filestore agent volumes offer high-performance, elastic, persistent file storage for agentic workloads. Thanks to its tight integration with GKE Agent Substrate and GKE Agent Sandbox, Filestore agent volumes automatically allocates and attaches a dedicated, isolated file workspace to GKE agent sandboxes in milliseconds. Request access to the preview here. 

• New product: If you run generative AI and RAG data layers  — think Milvus, Pinecone, Qdrant, Vespa, Redis, and in-memory context caching — you may want to take a look at the M4N family of VMs, now GA, which offers the highest per-core IOPS and throughput of leading hyperscalers. Paired with Google Cloud's custom Titanium offload architecture and paired with Hyperdisk Extreme, M4N instances deliver up to 25,000 MiB/s (25 GiB/s) of aggregate host storage performance and up to 1 million IOPS. 

• New product: Another new Compute Engine product, Z4D, is GA, and a strong storage solution for AI/ML training and inference workloads. When configured as a bare metal instance, Z4D provides both the high local SSD (LSSD) capacity and low latency required by agentic microVMs, so you can run thousands of isolated sandboxes per host with native performance and efficiency. Learn about Z4D machines here.

• Product update: Today’s AI training and inference pipelines create data faster than most storage management systems can keep up, creating challenges for teams trying to understand their storage estates. A new version of Storage Intelligence advisor makes it easier to answer the question: "What’s in my buckets?" and quickly identify unexpected changes. Then, enhanced batch operations let you automate bulk changes across your buckets — say, move storage classes, mass-delete stale or temporary data, or apply metadata, tagging, retention, or encryption changes. Learn about the latest in Storage Intelligence advisor here.

• New product: Last but not least, a new version of AlloyDB for PostgreSQL brings together pioneering Google infrastructure — Colossus distributed file system, and Jupiter network — to power a new, no-compromises database architecture for the agentic era. 

Practitioner guides and how-tos

• How-to: Wish you could make GPUs and TPUs scattered around the globe behave as a single pool behind a single entry point? In this blog, we show you how to do just that. At the edge, the multi-cluster GKE Inference Gateway focuses on global, multi-region traffic distribution and high availability. Beneath that, the LLM-d router handles complex, memory-aware scheduling algorithms to keep utilization high. This architecture is deliberately runtime-, model-, and accelerator-agnostic, and in tests, routing traffic through the multi-cluster GKE Inference Gateway added less than 1% overhead.

• Guide: Using or planning to use GKE on TPUs for AI model training or inference? Training massive Large Language Models (LLMs) or running high-throughput inference serving represents a significant investment in specialized AI hardware, such as Cloud TPUs and GPUs. To get the most out of every dollar spent, you need to understand workload lifecycle metrics in GKE. This detailed guide explains how to turn opaque cluster behaviors into actionable telemetry.

Customer and partner updates

• GKE customer: Learn why gaming startup SeaVerse relies on GKE Agent Sandbox for its multi-tenant workloads, and how the platform helped it decrease its infrastructure costs by 60%. 

Research, reports and deep-dives

In case you missed it, we’re also thrilled to share that Google has been named a leader, including achieving the highest score on either product or strategy, in three key analyst reports from Gartner and Forrester. 

• Google is a leader in The Forrester Wave™: Public Cloud Platforms, Q3 2026: Highest overall score of any cloud provider!
• Google named a Leader in 2026 Gartner® Magic Quadrant™ for Strategic Cloud Platform Services: Positioned furthest for “Completeness of Vision” of all vendors evaluated.
• Google is a Leader in the 2026 Gartner Magic Quadrant for Container Management: Positioned highest in “Ability to Execute” of all vendors evaluated.
• Google Cloud achieved a Gold rating in the latest SemiAnalysis ClusterMax 3.0 report, which evaluates the reliability, performance, support, pricing, and security of GPU providers globally. See the full report for more.

August 2026

Product, technology, and tools updates

• Product update: Filestore, Google Cloud’s first-party, secure, scalable NFS file service, has emerged as a popular storage platform for AI and agentic workflows, and now, it’s even better suited to the task, with a new backend storage layer built directly on Colossus, Google’s foundational distributed storage system. This new backend lets you provision IOPS independently from storage capacity, and is deeply integrated with GKE. In AI environments, this can help you service so-called agentic swarms — large groups of agents that need to read and write to a common dataset — without a drop off in performance. For more, check out the blog post. 

• New feature: gVisor sandboxes are now available in distributed Ray clusters on GKE. In partnership with Anyscale, we introduced an experimental library for Ray that brings gVisor, Google’s open-source application kernel, directly into distributed Ray clusters. gVisor provides lightweight environments with stronger isolation than ordinary containers, plus fast startup times and low memory overhead. To try out these sandboxing capabilities on GKE, head over to the Ray sandboxing User Guide.

• Product update: Looking for high-performance, easy-to-use infrastructure on which to run a personal AI agent, but don’t want to spend a lot of money? New Cloud Run instances are dedicated, singleton compute runtimes on Cloud Run that won’t shut down when the agent is idle. Better yet, the cost to run a Cloud Run instance with 1 vCPU and 1 GiB of memory continuously for 30 days is just $5.70.  

Practitioner guides, documentation and how-tos

• How-to guide: Big news in Model Context Protocol (MCP) land: As of the 2026-07-28 specification, the protocol core is “completely stateless. The handshake is gone. The initialize / initialized handshake (SEP-2575) and the logical Mcp-Session-Id header (SEP-2567) have been removed entirely. Instead, every request is now self-describing and independent.” Whoa. Learn more about the changes that the latest MCP specification brings, and more importantly, how to implement them, in this Google Developers blog.  
• Guide: Real-time AI systems make a mess of traditional network load balancing techniques. “Instead of handling isolated requests, the backend has to manage a continuous, live bidirectional stream. You’re dealing with a constant stream of audio chunks, transcripts, model outputs, and synthesized speech flowing back and forth simultaneously.” Things only get worse when the user gets involved. “The server has to immediately halt its current speech generation, pivot to update the context, maybe trigger a new tool, and start drafting a different response; this must be done without dropping the connection.” For a new approach to managing load in the AI era, read Scaling real-time AI agents with session-aware load balancing.
• How-to: Learn how to build an elastic, scalable LLM inference platform on GKE, even with a mix of different GPU accelerators. The proposed architecture combines Capacity Advisor and Compute Advisor, plus high-performance storage like RunAI:model streamer or GCPFuse with parallel downloads. Get all the details here.
• Documentation: The thing about hosts with GPUs or TPUs is that you can’t use live migration to update them, setting up a maintenance challenge. In this new docs page, learn how to update accelerator-equipped hosts according to your tolerance for downtime for your training and inference workloads.    
• Documentation: Advanced Compute Images, or ACIs, are standardized image stacks for AI/ML and HPC infrastructure, so you don’t need to manually build your own custom images. In this new docs page, learn how to create an ACI image using the Google Cloud CLI, console, or SchedMD's Slurm workload manager. 
• Guide: AI workloads are notoriously difficult to architect, resource-intensive, and bursty, which can also lead to scaling bottlenecks and large pools of underutilized — or misutilized — compute resources. A new blog outlines the three main ways to achieve dynamic capacity management in Google Cloud: 1) scheduling capacity for planned downtime; 2) maintaining automated fallback capacity for unplanned downtime; and 3) relying on GKE’s core orchestration capabilities to automate resource allocation. 

Customer and partner updates

• Business orchestration software provider UiPath was dealing with spiky workloads, and wanted more predictable costs. To get there, it re-architected its infrastructure, moving from isolated clusters to a shared Google Cloud GPU fleet that included both A3 VM instances (NVIDIA H100 GPUs) for training with G4 VM instances (NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs) for inference. You can read more about their architecture here. 

• Mirendil, an frontier AI lab focused on accelerating AI development, announced that it is using AI Hypercomputer with both TPUs and NVIDIA GPUs to support its model pre-training and post-training applications. 

• Replenit, a retail CRM provider, built its AI decision engine in Google Cloud, using BigQuery, Gemini Enterprise Agent Platform, and open-source Gemma models that it runs on Cloud TPUs. This latter combination provided Replenit with 90% lower pipeline costs than their previous cloud provider, the company reports. Read the full case study for more. 

• Malachyte architected its AI-powered e-commerce recommendation platform on top of Bigtable, Managed Service for Apache Kafka, Pub/Sub, Compute Engine, and last but not least, GKE. See how it all comes together in this blog.

July 2026

Product, technology, and tools updates

• Product update: Google Cloud Managed Lustre is now GA, and available in four distinct performance tiers that deliver throughput ranging from 125 MB/s, 250 MB/s, 500 MB/s, to 1000 MB/s per TiB of capacity — with the ability to scale up to 8 PB of storage capacity. The Managed Lustre solution is powered by DDN’s EXAScaler, combining DDN's decades of leadership in high-performance storage with Google Cloud's expertise in cloud infrastructure.

• Product update: C4N network and storage optimized VMs are now GA. C4N is our first network- and block-storage-optimized VM series built to eliminate data-transfer bottlenecks. Powered by 5th Gen Intel Xeon Scalable processors and built on Google's Titanium offloading hardware, it achieves 400 Gbps network bandwidth, 95 million packets per second (MPPS), and up to 25 GiB/s of block storage throughput when paired with Hyperdisk Extreme.

• New feature: GKE Dataplane V2 up to 15K Nodes with Network Policies (GA). This capability enables standard GKE clusters to scale up to 15,000 nodes while maintaining full active Network Policy enforcement, supporting the massive infrastructure needs of large enterprise and AI/ML customers.

• New feature: Co-operative time-slicing in llm-d. If you’re running reinforcement learning (RL) workloads, you can now interleave independent RL jobs onto shared physical hardware, increasing aggregate accelerator duty cycles from a ~40% baseline up to 70% without impacting model convergence or accuracy. 

• New AI security tool: Looking to secure your AI supply chain on GKE, deploy AI workloads safely, and cut down on shadow AI? We open-sourced k8s-aibom, a lightweight, unprivileged Kubernetes controller that continuously monitors container clusters to automatically detect running AI runtimes (like vLLM and Triton) and generate standard CycloneDX Machine Learning Bill of Materials (ML-BOMs). Check out the k8s-aibom project and get involved.

Practitioner guides and how-tos

• How-to guide: On July 27, Google announced Day 0 support for Moonshot AI’s Kimi K3 2.8-trillion-parameter open-weight model, the day weights were released. Whichever your preferred deployment path — via Model Garden, custom orchestration, or GKE with llm-d recipes — this guide offers detailed step-by-step instructions to help you evaluate and pilot Kimi K3 in Google Cloud. 

• How-to guide: Google Kubernetes Engine (GKE) managed DRANET supports both GPUs and TPUs. There are several configurations to use this implementation, including standard cluster (where you have full control) and autopilot cluster (where Google does the heavy configs for you). Take a deeper dive in the hands-on lab, GKE Autopilot clusters with TPUs, GKE managed DRANET and Gemma 4.

• How-to guide: Learn to run Ray on TPUs, not GPUs. In Part 1 of this two-part series, we discuss TPU slices (hint: Ray thinks of them as just another accelerator on which to schedule), then walk through Ray’s various AI libraries (Part 2).

• How-to guide: Evaluate TPUs for sample workloads using a new microbenchmark suite that helps you accurately assess whether a device is achieving its theoretical performance specifications, and to identify specific performance gaps or architecture-specific bottlenecks. Dive in here. 

• How-to guide: Scale your agents without killing your budget. Learn how GKE orchestration can help you safely pack more agents onto a fixed compute footprint with GKE Agent Sandbox and Pod snapshots. Whether your goal is performance or cost optimization, we teach you how to turn the right dials for optimal agent efficiency. 

• Technical blueprint: Inside the optimization of Mistral 3 large inference on Ironwood. This blog outlines how one Google team optimized Mistral 3 large MoE model inference on Google’s Ironwood (TPU v7x), achieving a 1.5x performance gain. They did so with hybrid sharding, replacing linear VPU summations with tree reductions, optimizing GMM/MLA kernels, and adopting asynchronous scheduling. As a result, they boosted throughput by up to 48% while maintaining benchmark accuracy neutrality. Read the full blog here.

Research, reports and deep-dives

• Report: Google was named a Leader in the inaugural GartnerⓇ Magic Quadrant™ for AI Infrastructure, positioned highest for ‘Ability to Execute’ and furthest for ‘Completeness of Vision’. Gartner called out Google’s proprietary scalable compute, integrated AI Hypercomputer architecture, and the scale of our AI compute capacity as key strengths. Download a copy here.

• Report: We recently surveyed more than 1,400 senior IT leaders for our State of AI Infrastructure report, and a resounding pattern emerged: The gap between AI ambition and infrastructure reality is widening. In fact, 83% of organizations say they require infrastructure upgrades to support production-grade agentic AI. Read the accompanying blog to understand how adapting your infrastructure to meet the demands that agentic applications place on your systems will help you move from pilot to production.

June 2026

Product, technology and tool updates

• Product update: Protecting sensitive data used with AI is a critical part of advanced and secure cloud infrastructure. Confidential Computing cryptographically protects data in use in hardware-based Trusted Execution Environments (TEEs) with verifiable data integrity, and is now available on the accelerator-optimized G4 machine series, featuring NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs. Get started with Confidential G4 VMs and Confidential G4 GKE Nodes.

• Developer resource: The new TPU Developer Hub is the place to go for model builders, optimizers, and developers to learn to unlock the full performance of Google Cloud TPUs. Read more in this blog. 

• New product: Scale your AI workloads with the new OpenTelemetry-Based TPU AI Telemetry Collector Agent. For the first time, you can route high-fidelity TPU hardware telemetry to Google Cloud Monitoring, Google Managed Prometheus, or your own self-hosted Grafana stack.

Practitioner guides and how-tos

• How-to guide: Learn how to build high availability into an AI inference workload running on GKE Inference Gateway with TPUs, Cloud Storage FUSE and Dynamic Resource Allocation (DRA). This blog provides an overview, or you can get all the technical details in the hands-on codelab.

• How-to guide: Did you know you can connect your AI agents to unstructured data in Cloud Storage via Model Context Protocol (MCP)? In this blog, learn about why would want to do that from three customer examples, then how to do it, choosing either a fully managed service, or a self-managed local server for more customization and control. 

Research, reports and deep-dives

• Report: According to an independent benchmark report, GKE Inference Gateway outperforms the next leading managed Kubernetes service with 15.7% higher throughput, 92.8% shorter wait times, and 62.6% lower inter-token latency. This performance can be attributed to its use of prefix caching, which optimizes LLM performance by storing the KV cache (activation states) of long, repetitive prompt prefixes. Learn more in the blog. 

• Architecture deep dive: A closer look at the cold start problem, this time for TPUs and GKE, and how the Run:ai Model Streamer can help change the dynamic. 

Customer and partner updates

• Customer win: Leveraging GKE, BigQuery, Cloud SQL, and Gemini Enterprise Agent Platform, Pager Health is eliminating operational fragmentation to deliver a simplified, personalized U.S. healthcare experience that transforms lives.

• Customer win: Trustpilot, the customer review platform, built a high-volume streaming pipeline using fine-tuned Gemma models with Dataflow and Gemini Enterprise Agent Platform running on cost-optimized A2 VMs using A100 GPUs, as well as optimized version of vLLM maintained by Gemini Enterprise Agent Platform.

May 2026

Product, technology and tool updates

• Product update: GKE Agent Sandbox is now generally available.

• New open-source project: Agent Substrate is a new open-source project aimed at continuing to push the limits of agentic infrastructure density

• New feature: Google AI Edge Portal, a solution for testing and benchmarking on-device machine learning (ML) at scale, now supports benchmarking and debugging on-device LLMs. Read more here. 

• Product deep dive: We went into depth about Cloud Storage Rapid, a new family of high-performance storage offerings for AI workloads. At launch, offerings include Rapid Bucket (formerly Rapid Storage), a high-performance zonal object storage offering, and Rapid Cache (formerly Anywhere Cache), which accelerates reads on-demand and colocates compute and data for workloads in existing buckets. 

Research, reports and deep dives

• Architecture deep dive: Google Global Infrastructure VP Bikash Koley and Engineering Fellow Arjun Singh provide a high-level overview of the challenges that AI workloads pose to network infrastructure, and discuss the deep enhancements we’ve made to our data center fabrics, WAN, and global networks to better support them. 

• Architecture deep dive: We unveiled a new cluster-level reliability model for developing frontier AI models on TPUs, ditching instance-level reliability 

Customer and partner updates

• Customer win: Visual media provider Imgix serves more than 8 billion images and videos from AI Hypercomputer equipped with G4 VMs powered by NVIDIA RTX PRO 6000 Blackwell GPUs.