At Google Cloud Next '26, Google introduced cross-cloud infrastructure for the agentic enterprise, delivering new innovations in fluid compute, secure cross-cloud connectivity, unified data layer, and digital sovereignty to help organizations run AI agents securely and efficiently across environments. These updates address the performance, security, and data demands of agentic AI workloads that generate rapid machine-to-machine traffic and require real-time adaptability.
Fluid compute enables adaptive AI agent workloads
Google Cloud introduced the C4N Series VMs, which process up to 95 million packets per second—up to 40% faster than other leading hyperscalers—to eliminate I/O bottlenecks for security appliances, streaming media, and open-source databases under agentic AI demands. These VMs help enterprise workloads maintain performance even during spikes in agent-generated traffic.
The M4N Series with Hyperdisk Extreme delivers industry-leading per-core IOPS and throughput, providing 26.57 GB of RAM per vCPU to scale mission-critical workloads cost-effectively on fewer cores. For Oracle workloads, this configuration reduces total cost of ownership by over 20% compared to leading hyperscale clouds by minimizing overprovisioning and data pipeline bottlenecks.
GKE Agent Sandbox secures AI agents with gVisor isolation and launches up to 300 sandboxes per second per cluster, offering up to 30% better price-performance than competitors when running AI agents on GKE with Google Axion N4A. It supports demand spikes and secure execution for agentic orchestration, reinforcement learning, and RAG workflows.
Secure cross-cloud connectivity governs agentic traffic
Agent Gateway acts as an air traffic controller for enterprise agentic traffic within the Gemini Enterprise Agent Platform, natively understanding protocols like MCP and A2A to inspect and govern every agent interaction. It integrates with Google and third-party identity and AI safety services to verify access, block attacks, and protect sensitive data while maintaining compliance.
Cloud Network Insights provides end-to-end visibility across hybrid and multi-cloud environments, using synthetic traffic analytics and AI-powered insights from Gemini Cloud Assist to deliver hop-by-hop network path visibility and faster troubleshooting for agentic workloads across Google Cloud, AWS, Azure, and data centers.
Enhanced Cloud Next Generation Firewall and Cloud Armor deliver machine-speed, AI-powered protection against AI-generated polymorphic malware and zero-day exploits. Cloud NGFW’s advanced malware sandbox prevents threats in real time, while Cloud Armor’s managed rules defend against known and unknown CVEs, working with Model Armor to analyze the intent and content of AI agent communications.
Unified data layer and digital sovereignty turn data into actionable context
Smart Storage transforms dark data into a knowledge asset by embedding semantic intelligence directly into data objects through automated annotation, entity extraction, and semantic search. This enables AI agents to instantly find and use specific data in spreadsheets, PDFs, and other unstructured formats across the organization, accelerating AI development from months to weeks.
Knowledge Catalog maps business meaning across the data estate, providing a grounded source of truth so agents can deliver accurate results without requiring data migration. Agents interact with it directly wherever data lives, with full context and governance, supporting AI training and inferencing as part of the Agentic Data Cloud.
Digital sovereignty capabilities include Confidential External Key Management, which lets organizations retain full control of encryption keys in a tamper-proof environment using Confidential Compute, and Gemini on Google Distributed Cloud, which enables deployment of Gemini models in air-gapped or on-premises environments. These are backed by NVIDIA Blackwell GPUs, new VM families, and enhanced storage with 6x capacity and 10x performance per zone to support sovereign AI workloads.
What to do next
These cross-cloud infrastructure innovations provide the foundation for enterprises to run AI agents at machine speed while maintaining security, compliance, and control. Teams should evaluate their agentic workloads against the new compute, networking, data, and sovereignty capabilities to determine where to pilot or migrate workloads for optimal performance and cost efficiency.
Source: Cross-cloud infrastructure innovation for the agentic enterprise (GCP).
