Cross-Cloud Infrastructure for the Agentic Enterprise

Summary

At Google Cloud Next '26, Google introduced new cross-cloud infrastructure innovations for the Agentic Enterprise. These updates focus on compute, secure connectivity, unified data layers, and digital sovereignty. As AI agents become more common in business workflows, enterprises need infrastructure that can support dynamic workloads, secure agent traffic, and work across cloud environments.

Why Cross-Cloud Infrastructure Matters

Most large organisations do not depend on a single system or cloud platform. Their data, applications, workloads, and users are often spread across:

• Public cloud platforms
• Private data centres
• SaaS applications
• Hybrid environments
• Regional infrastructure
• Third-party business systems

As AI adoption grows, agents may need to access information and complete tasks across these environments. This makes cross-cloud infrastructure important for performance, governance, security, and business continuity.

The Rise of Agentic Workloads

Agentic AI introduces a new kind of workload. Unlike traditional applications, AI agents can generate many internal requests, interact with other agents, call models, access tools, and work through reasoning loops.

This can increase demand on:

• Compute systems
• Networks
• Databases
• Storage platforms
• Identity systems
• Security controls

Traditional infrastructure may not be enough for this level of dynamic activity. Enterprises need systems that can adapt quickly and securely.

Four Key Areas of Google's Cross-Cloud Infrastructure Updates

Google highlighted four main areas of innovation for cross-cloud infrastructure at Cloud Next '26.

1. Fluid Compute

Fluid compute is designed to help Google Compute Engine and Google Kubernetes services work together for high-speed, cost-effective AI agents and enterprise workloads.

Agentic workloads can be unpredictable. They may suddenly increase compute demand when agents start multiple tasks or interact with other systems.

Fluid compute helps support this by enabling infrastructure to adapt more dynamically.

Key updates include:

• New CPU families
• GKE capabilities
• Hyperdisk block storage capabilities
• Support for secure agent execution
• Better handling of demand spikes

For businesses, this means better support for both traditional workloads and AI-driven operations.

2. Secure Cross-Cloud Connectivity

AI agents may communicate with applications, databases, APIs, other agents, and cloud services. This creates new security and visibility challenges.

Google introduced updates such as Agent Gateway, which acts as a control point for enterprise agentic traffic. It helps govern and orchestrate agent interactions and supports protocols such as MCP and A2A.

Secure connectivity is important because enterprises need to know:

• Which agents are communicating
• What systems they are accessing
• Whether the traffic is safe
• Whether permissions are being followed
• How agent activity is monitored across clouds

This helps businesses manage AI agents more safely in complex environments.

3. Unified Data Layer

AI agents need trusted data to make useful decisions. Google's cross-cloud infrastructure updates include capabilities such as Smart Storage and Knowledge Catalog to help transform data into a more active reasoning layer.

A unified data layer helps agents:

• Find relevant data
• Understand business context
• Respect data permissions
• Work across multiple data sources
• Reduce dependency on manual data search

This connects closely with Google's Agentic Data Cloud approach, where data becomes a foundation for business action.

4. Digital Sovereignty

Many businesses need to meet regional, industry, and compliance requirements around data location, access control, and encryption.

Google announced updates such as Confidential External Key Management and new capabilities in Google Distributed Cloud. These are designed to bring Google's AI models and enablers closer to where enterprise data lives.

This is especially important for organisations in regulated sectors such as finance, healthcare, government, telecom, and public services.

Why CPUs Still Matter in the AI Era

While GPUs and TPUs are important for AI training and inference, Google also highlighted the role of CPUs in agentic workflows. CPUs are useful for branch-heavy logic, orchestration, secure execution sandboxes, small language model inference, retrieval-augmented generation, and control flows.

This is important for enterprises because not every AI workload needs the same type of processor. A balanced infrastructure strategy can help improve performance and cost efficiency.

GKE Agent Sandbox

Google introduced GKE Agent Sandbox to help secure agents with trusted gVisor isolation. It is designed to handle demand spikes and launch up to 300 sandboxes per second, per cluster.

For businesses, this can help improve the security of AI agent execution by isolating agents and reducing risk when workloads scale quickly.

Business Benefits of Cross-Cloud Infrastructure

Cross-cloud infrastructure can help enterprises in several ways:

• Support AI agents across complex environments
• Improve workload scalability
• Strengthen security and governance
• Reduce infrastructure bottlenecks
• Improve visibility into agent traffic
• Support hybrid and multicloud strategies
• Help meet compliance and sovereignty requirements
• Enable faster AI adoption across business units

These benefits are useful for companies that want to move from AI pilots to production-scale AI operations.

What Enterprises Should Consider

Before scaling agentic AI, organisations should review their infrastructure readiness.

Key questions include:

• Are workloads running across multiple clouds or environments?
• Can current systems handle dynamic AI agent activity?
• Is agent traffic visible and governed?
• Are identity and access controls ready for non-human agents?
• Can data be accessed securely across platforms?
• Are compliance and sovereignty requirements clearly defined?
• Can compute and storage scale based on demand?
• Is there a monitoring strategy for agent behaviour?

Answering these questions can help businesses build a safer and more reliable AI foundation.

Resources / References

• Cross-cloud infrastructure innovation update
• Google Cloud Next '26 keynote and announcements
• Agentic Data Cloud announcement

Credits

This blog is based on official Google Cloud announcements shared during Google Cloud Next '26.

Conclusion

Cross-cloud infrastructure is becoming essential for the Agentic Enterprise. As AI agents work across applications, data platforms, cloud environments, and business systems, enterprises need infrastructure that is secure, scalable, adaptive, and governed. Google Cloud's updates around fluid compute, secure connectivity, unified data layers, and digital sovereignty show how cloud infrastructure is evolving for the next phase of enterprise AI.