Cloud

Google Cloud is committed to being a trusted partner for customers who are navigating AI regulations in Europe. We have long understood that this requires a proactive and collaborative approach in order to ensure Europeans can access secure, first-rate AI tools as they become available. 

This week, Google announced that we intend to sign the European Union AI Act Code of Practice (the Code). Google Cloud supports the stated vision of the AI Office to offer a simple and transparent way to demonstrate compliance with the AI Act that offers a streamlined compliance process, with enforcement focused on monitoring their adherence to the Code. We believe that this approach can result in greater predictability and a reduced administrative burden.

By participating, we believe our customers will benefit by being able to compare among the best cloud services and derive their own compliance benefits.

Looking ahead, customers should become familiar with these three compliance documents as they seek to develop and deploy AI in the EU.  

1. The EU AI Act is a legal and regulatory framework that establishes obligations for certain AI systems based on their potential risks and levels of impact. 

2. The General-Purpose AI Code of Practice is a voluntary tool, prepared by independent experts in a multi-stakeholder process, designed to help industry comply with the AI Act’s obligations for providers of general-purpose AI models.

3. Separately, the GPAI guidelines focus on the scope of the obligations for providers of general-purpose AI models laid down in the AI Act, in light of their imminent entry into application on August 2, 2025.

Google Cloud’s approach to helping customers with the Act

A core pillar of our approach to trust in AI, and a critical component of AI Act compliance, is data governance and privacy. Customers control how and where their data is used. We embed privacy-by-design principles throughout our product lifecycle, including in AI development, ensuring architectures include privacy safeguards like data encryption and providing meaningful transparency and control over data use. 

We have delivered on the commitment to European customers we made back in 2020 to help them transform their businesses and address their strict data security and privacy requirements. To date, we’ve invested billions of euros to expand access to secure, high-performance computing capacity with seven data centers in Europe in addition to 13 cloud regions in Poland, Finland, Germany, Italy, Spain, France, Belgium, Sweden, the Netherlands, Switzerland, and more under development. 

Our Sensitive Data Protection service and VPC Service Controls further assist customers in protecting sensitive data and meeting data residency requirements. We are already working to add new features to support data governance in-line with AI Act compliance requirements. 

Supporting your compliance

We were among the first organizations to publish AI principles in 2018, and have published an annual transparency report since 2019. We consistently review our policies, practices, and frameworks, incorporating robust safety and security practices, privacy-by-design, and risk assessments. 

We are committed to providing and regularly updating documentation about our AI tools and services. The Cloud Compliance Center remains the most up-to-date resource for all customer compliance artifacts including Google Cloud’s ISO 42001 AI Management System certification and EU AI Act related documentation. As we prepare for compliance for all new models which will be launched globally, including in the EU, we’ll update all artifacts to ensure timely integration into Google Cloud customer uses.

Our continuously updated Secure AI Framework (SAIF) provides a conceptual framework for securing AI systems across data, infrastructure, application, and model dimensions, emphasizing defense-in-depth and secure-by-design foundations. This ensures early inclusion of prevention and detection controls, adapted to specific product and user risks. 

Of course, operationalizing any industry framework requires close collaboration with others — and above all a forum to make that happen. That’s why last year we worked with industry partners to launch the Coalition for Secure AI (CoSAI) to advance comprehensive security measures for addressing the unique risks that come with AI, for both issues that arise in real time and those over the horizon.

What customers can do to prepare

Customers should work closely with the EU AI Office to understand their legal and regulatory obligations when seeking to modify a foundation model or integrate one into a large system. It will be important to track new guidance and developments released by the AI Office. 

We will continue to follow all legal obligations under the Act and demonstrate how we are fulfilling and supporting compliance requirements — including new forthcoming models which will be subject to the Code. 

We remain committed to providing our enterprise customers with cutting-edge AI solutions that are both innovative and compliant. We have the capabilities and experience, and we will continue to partner with policymakers and customers as new regulations, frameworks, and standards are developed.

Imagine a code review process that doesn’t slow you down. Instead of a queue of pending pull requests, you have an intelligent assistant that provides a near-instant, comprehensive summary of every change. It flags potential bugs, suggests improvements based on best practices, and frees up your human reviewers to focus on the complex architectural decisions that truly require their expertise.

This isn’t a future-state prediction; it’s what’s possible today with Gemini Code Assist, integrated directly into your GitHub workflow at no charge. By embedding a powerful AI partner into every pull request, we’re transforming code reviews from a frustrating bottleneck into a fast and painless way to ensure high quality and consistent code, leading to higher code quality and happier developers.

The challenge: Why code reviews are a bottleneck

Code reviews are a non-negotiable part of building quality software, but they are often a major bottleneck in the development lifecycle. This friction slows down delivery velocity, leads to inconsistent code quality, and makes it difficult to enforce best practices.  Research from DORA’s Impact of Generative AI in Software Development report advises that organizations, “Double-down on fast high-quality feedback, like code reviews and automated testing, using gen AI as appropriate.”   DORA research has found that teams with shorter-code review times have 50% better software delivery performance.¹ AI adoption has been shown to increase code review speed by an estimated 3.1% for every 25% increase in AI adoption, and improve reported code quality by 3.4%.²

The solution: An AI-powered partner in your PRs

Gemini Code Assist integrates into your GitHub workflow as an intelligent partner, conducting code reviews to solve these challenges. When a pull request is created, Gemini is automatically assigned as a reviewer and gets to work immediately. Here’s how it helps:

• Near-instant PR Summaries: Provides a comprehensive summary of the changes to help human reviewers get up to speed almost instantly.
• In-depth automated reviews: Identifies stylistic issues, deviations from best practices, and potential bugs, freeing human reviewers to focus on more critical issues.
• Interactive assistance and learning: Allows anyone in the PR to have a conversation with the AI using /gemini commands to ask for alternative implementations or get a fresh review.

The power of Gemini 2.5: A leap in code review intelligence

With our recent general availability release, Gemini Code Assist has been upgraded to run on our latest model, Gemini 2.5. This isn’t just an incremental update—it’s a major leap forward in the quality, accuracy, and intelligence of AI-powered code reviews.

What does this mean for your daily pull requests?

• Deeper insights: You’ll see more insightful suggestions that go beyond simple style fixes. Gemini 2.5 is better at understanding the logic and intent behind your code, helping to identify potential bugs and suggest more efficient implementations.

• More actionable code suggestions: Through customer specific style guides and configurations, the AI-generated code is now more accurate and context-aware. You’ll find that suggestions are often so reliable that you can commit them with confidence after a quick review, speeding up your iteration cycles.

• Improved relevancy We have significantly increased the relevancy and accuracy of suggestions. This means you can trust the feedback you receive and spend less time second-guessing the AI and more time moving your project forward.

These improvements are a direct result of our rigorous internal testing and deliver more comprehensive and actionable feedback to help developers be more productive.

Customer story: How Delivery Hero enhances code quality with Gemini Code Assist 

Delivery Hero is the world’s leading local delivery platform, operating in around 70 countries.  Strengthening the code review process is a key part of their commitment to delivering high-quality, reliable features for the customers, which prompted them to become an early adopter of the Gemini Code Assist GitHub app.  The initial feedback from their developers was overwhelmingly positive, and the app’s recent upgrade to the Gemini 2.5 model brought an even more significant impact to the app development process, particularly on code quality. Their engineers found the AI-generated comments to be highly relevant and valuable, directly contributing to a higher standard of code. The latest developer experience survey after the adoption of the Gemini Code Assist GitHub app showed a clear rise in developer satisfaction with the quality and speed of code reviews—marking a positive turn for this important metric.  

“By integrating the Gemini Code Assist GitHub app into our workflow, we are not just adopting a tool; we are fostering a more robust, efficient, and collaborative engineering culture.  Our close partnership with Google as an early adopter, built on an agile and effective feedback loop, has been a key part of this success.” N. Mert Aydin,  Principal Software Engineer, Delivery Hero 

Get started

• Try it today: Get started with Gemini Code Assist on the GitHub Marketplace.
• Learn more: Dive deeper by reading the official documentation.

<sup>1.  2023 DORA Report
2. 2024 DORA Report</sup>

AI is  evolving beyond single, task-specific agents into an interconnected ecosystem, where autonomous agents collaborate  to solve complex problems, regardless of their underlying platform. To make this transition easier for developers, we are announcing a comprehensive suite of tools that will empower developers to build, deploy, evaluate, and sell Agent2Agent (A2A) agents with Google Cloud.

Today, we’re excited to announce the release version 0.3 of the A2A protocol, which brings a more stable interface to build against and is critical to accelerating enterprise adoption. This version introduces several key capabilities, including gRPC support, the ability to sign security cards, and extended client side support in the Python SDK, which provide more flexible use, better security and easier integration. 

The A2A protocol is quickly gaining momentum, with support from a growing ecosystem of over 150 organizations that spans every major hyperscaler, leading technology providers, and multinational customers using Google Cloud. Businesses are already building powerful capabilities for their organizations. For example, Tyson Foods and Gordon Food Service are pioneering collaborative A2A systems to drive sales and reduce supply chain friction, creating a real-time channel for their agents to share product data and leads that enhance the food supply chain.

Build: Native support for A2A in the Agent Development Kit (ADK)

We’re releasing native support for A2A in Agent Development Kit (ADK), a powerful open source  agent framework released by Google. This makes it easy to build A2A agents if you are already using ADK and is built upon our previously-released A2A SDKs. For example, with a simple “Hello, World!” style code snippet, developers can now use ADK to:

• Use an A2A agent with an Agent Card and use it as a sub-agent.

• Expose an existing ADK agent to make it discoverable as an A2A agent.

Developers can start building collaborative agents with ADK today.

Deploy: Flexible deployment options with Agent Engine, Cloud Run, and GKE

Once agents are built, they need a robust and scalable home to exist within. We are providing three powerful deployment paths for customers to scale agents into production:

• Deploy to Agent Engine: For a managed, agent-optimized environment, Agent Engine is the choice for many agent builders. We’re adding support for A2A to Agent Engine in the coming weeks so that you can easily deploy an agent written in any framework to Agent Engine and get a production ready, Google-scale, A2A agent.

• Deploy to Cloud Run: For increased flexibility, you can containerize and deploy your A2A agents to Cloud Run, leveraging Google’s serverless infrastructure for massive scale and reliability. Follow the published guide.

• Deploy to Google Kubernetes Engine (GKE): For maximum control, you can deploy agents to GKE, providing the full power of Kubernetes to manage A2A systems at scale.

With support for A2A arriving in the coming weeks, developers will be able to use the `agent-starter-pack` CLI tool to complete CI/CD setup in just one line:
“`
uvx agent-starter-pack create my-agent -a adk@gemini-fullstack
“`

Integrate: Bring your A2A agents to users with Agentspace

Agents need safe and accessible environments to be useful. That’s why we built Agentspace, the destination where agents meet end users. In the coming weeks, partners will be able to make any A2A agent available in Agentspace, transforming it from a standalone tool into a valuable service that people can consume. This includes partner-built agents that are built on partner platforms, giving customers the flexibility to access these A2A agents in multiple locations. 

More than just a hub, Agentspace provides the critical governance, safety, and control features needed for an enterprise-ready agent platform, ensuring that interactions are secure and reliable.

Evaluate and commercialize your A2A systems

Building and deploying agents is just the beginning. To create truly enterprise-grade systems, you need robust evaluation capabilities, which is why we’re extending the Vertex GenAI Evaluation Service to support A2A agent evaluations. See our hands-on guidance.

Discover and sell partner-built A2A agents in AI Agent Marketplace

Partners can now sell their A2A agents directly to customers in the AI Agent Marketplace. This will allow Google Cloud customers to discover and purchase agents published by ISVs, GSIs, and other technology providers. The AI Agent Marketplace provides an important path to market for partners looking to monetize their AI Agents.

Partners can request more information here. 

The A2A ecosystem is growing

We announced the A2A protocol in April to lead the industry toward interoperable agent systems, and in June, we advanced that commitment by contributing it to the Linux Foundation. The industry’s response continues to grow, reflecting a shared belief in vendor-neutral, community-driven standards. Many of Google Cloud’s partners have previously offered agents to joint customers, and they are now enabling these agents with A2A to help future-proof investments for customers. 

• Adobe: A leader in generative AI, Adobe is leveraging the A2A protocol to make its rapidly-growing number of distributed agents interoperable with agents in Google Cloud’s ecosystem. The A2A protocol enables Adobe agents to collaborate in the enterprise to create powerful new digital experiences, streamline workflows that optimize the content creation process, and automate multi-system processes and data integrations. 
• S&P Global Market Intelligence: S&P, a provider of information services and solutions to global markets, has adopted A2A as a protocol for inter-agent communication. This strategic alignment enhances interoperability, scalability, and future-readiness across the organization’s agent ecosystem.
• ServiceNow: As a founding partner of A2A, ServiceNow empowers customers with its AI Agent Fabric, a multi-agent communication layer that connects ServiceNow, customer, and partner-built agents. This provides enterprises with the greater choice and flexibility needed to unlock the full potential of agentic AI, resulting in faster decisions, fewer handoffs, and more scalable solutions.
• Twilio: Twilio is using A2A protocol for implementing Latency Aware Agent Selection. By extending the A2A protocol, individual agents now broadcast their latency, enabling the system to intelligently route tasks to the most responsive agent available and also adapt gracefully – for example, playing a filler prompt or adding typing sounds, if a high-latency agent is the only option.

Developers can review more about past releases in the release notes, learn about what’s coming in the future in our roadmap, and join the community to help evolve the protocol moving forward.  The community has also released great tooling around A2A with the launch of A2A Inspector and Technology Compatibility Kit. 

Get started

We’re excited to partner across the industry to build the future of artificial intelligence. Here’s how you can start:

• Explore Agent Development Kit (ADK) with A2A and build your first collaborative agent.

• Deploy it on Cloud Run.

We can’t wait to see what you build.

We’re thrilled to announce a significant expansion of our C4 virtual machine series, with the general availability of 28 powerful new shapes. This expansion introduces C4 shapes with Google’s next-gen Titanium Local SSD, C4 bare metal instances, and new extra-large shapes, all powered by the latest Intel Xeon 6 processors, Granite Rapids. We’re excited to be the first leading hyperscaler to bring Xeon 6 to customers.

C4 VMs with Xeon 6 deliver performance gains of up to 30% for general compute and up to 60% for ML recommendation workloads, and up to 35% lower access latency on Titanium Local SSD shapes. If you already use the C4 machine series, this means an easy and powerful path to the latest hardware without needing to migrate to a new machine series. You can take advantage of your existing committed use discounts (CUDs) and enjoy capabilities like managed instance groups and Google Kubernetes Engine (GKE) custom compute classes.

C4 VMs on Intel Xeon 6

Initially launched with Intel 5th generation Xeon processors (Emerald Rapids), C4 VMs provided advancements in performance and control for enterprise and mission-critical workloads, especially AI, gaming, databases, and data analytics. And now with Xeon 6 (Granite Rapids), C4 VMs power even more demanding workloads, delivering the highest frequency of any Google Compute Engine VM — up to 4.2 GHz, the most vCPU and RAM of any Intel-based comparable product, larger L3 cache size, and 1.35x higher maximum memory bandwidth. The C4 machine series offers enhanced maintenance controls with a 30-day uptime window between planned maintenance events, and scalable Hyperdisk storage with up to 500k IOPS and 10 GB/s, including features like Hyperdisk Storage Pools.

Inference workloads on C4 with Xeon 6 demonstrate up to 60% better performance per core compared to prior-generation C3. C4’s new shapes support FP16-trained models with Intel AMX-FP16, making it a great choice for accelerating machine learning inference. Also, a new larger half- and full-host C4 shape (144 and 288 vCPU) guarantees memory isolation, for highly predictable performance. The largest C4 shape enables up to 30% more performance per core on the estimated SPECrate®2017_int_base benchmark versus C3.

New C4 standard, highmem, and highcpu with 144 and 288 vCPU shapes are powered exclusively by Xeon 6:

Turbocharge your storage with Titanium Local SSD 

We’re also excited to announce Local SSD support for C4 VMs, available with new -lssd machine types, exclusively on Xeon 6 processors. These new local SSD shapes leverage the latest Titanium SSDs, delivering impressive I/O performance for workloads like high-performance databases (e.g., Cassandra, MongoDB, SQL Server tempDB and pagefile workloads), big data processing (e.g., Spark, Hadoop), media rendering and transcoding, and caching layers.

C4-lssd offers up to 7.2M max read IOPS, more than three times higher than comparable options from other leading hyperscalers. This translates to dramatic improvements in read/write latency. Titanium SSDs on C4 enable up to 35% lower access latency compared to previous generation SSDs.

C4 with Local SSD is available in standard and highmem configurations starting from 4 vCPUs, exclusively on Xeon 6:

Unlock direct access with C4 bare metal shapes

For workloads that demand direct access to CPU and memory resources, we’re introducing C4 bare metal shapes. These instances are ideal for commercial and custom hypervisors, applications not traditionally supported in virtual machines, and those with special performance monitoring or licensing requirements. C4 bare metal is SAP-certified, delivering an impressive 132,600 aSAPs — the highest of any comparable machine, providing peak performance and compliance for your critical SAP application servers.

Customers in financial services, SaaS/PaaS security, with dev/test environments, and private cloud platforms will find C4 bare metal types essential for meeting their stringent performance goals. Compared to the previous generation C3 bare metal instances, C4 bare metal offers up to a 35% performance improvement.

New C4 bare metal instances are available in standard-metal and highmem-metal, powered exclusively by Xeon 6:

Get started with C4 on Intel Xeon 6 today

The expanded C4 machine series with Intel Xeon 6 is available today in 19 zones, with more expansion to come. Check the latest regional availability on our regions and zones page. Experience the leading performance, predictability, and control that C4 VMs deliver today! To learn more about C4 and its new capabilities, visit the C4 documentation.

Managing complex SAP landscapes can feel like navigating a maze. You’re juggling application performance, database health, infrastructure stability, and more, all while striving to ensure seamless business applications. Today, we’re excited to announce new observability and monitoring functionality in Workload Manager to provide SAP customers with purpose-built tools that bridge gaps in data and provide a unified view of SAP environments across all layers of their cloud environments. 

Workload Manager is not intended to replace your existing SAP monitoring tools, but rather complement them by providing a broader view of your SAP systems and enabling you to more easily correlate application-level insights with infrastructure performance and events.

Expanding visibility in complex SAP environments

Traditional SAP monitoring excels at providing detailed information at the application and database level. Metrics about transaction performance, user activity, and application logs are readily available. However, understanding the relationship between this application data and the underlying infrastructure can sometimes require additional effort. For example, a slowdown in transaction processing might be due to a database issue, a network bottleneck, or even a problem with the underlying virtual machine. While existing tools provide valuable insights, correlating application performance with infrastructure health can be a time-consuming process or may require toggling between different monitoring platforms and manual efforts.

Creating a system-centric, unified view with Workload Manager

After enabling the necessary functionality in Google Cloud’s Agent for SAP, you can navigate to the main Observability dashboard to see an overview of your SAP systems and the overall health status of each system and sub-layer. The health status is calculated using a variety of metrics from SAP Netweaver instances, SAP HANA databases, pacemaker clusters, and the availability of the underlying infrastructure hosting them.

Clicking on a system will take you to the system Overview page, where you can visualize the components of the system and see the current health status for each instance included. The Health Insights table will summarize any health issues that are currently detected, and the Maintenance Events table displays any upcoming maintenance for supported machine types.

The “Applications” and “Databases” tabs behave similarly, and help you correlate key performance indicators from your SAP applications with metrics from your VMs, storage, network, and other Google Cloud services such as the following:

• Availability: GCE Instance Availability, SAP NetWeaver and SAP HANA instance status, SAP NetWeaver and SAP HANA process status and Pacemaker Cluster nodes and resource status. 

• Performance: GCE Instance CPU and memory utilization, operating system processes by CPU and memory and information about swap I/O operations on the operating system. 

• Storage: SAP related file systems usage, disk space utilization and usage, disk throughput and disk IOPS.

• Networking: Information about network traffic and network packages. 

• Netweaver: CPU and memory utilization by SAP NetWeaver process, response time by processes in the SAP NetWeaver instance, SAP NetWeaver sessions and RFC connections by type, SAP NetWeaver process utilization per work process type and information for each SAP NetWeaver process type for each respective queue in the system.

• HANA: SAP HANA System Replication status, SAP HANA System Replication latency, memory usage in the SAP HANA Database by type, information about idle and running connections in the SAP HANA database and top schemas by record count in the SAP HANA database.

• Backups: Information related to Backint backups including Backup/Recovery success rate, Backup/Recovery average MBps for all operations and historical data about both Backup/Recovery status and throughput. 

New Event Annotations built specifically for SAP allow you to surface system events directly on monitoring dashboards to help empower you to quickly identify the root cause or see the impact of events on your system’s performance and health. These events are also now available outside of Workload Manager in Cloud Monitoring dashboards. 

SAP Availability:

• Pacemaker cluster:

• Pacemaker cluster node status

• Pacemaker cluster resource status

• SAP HANA instance status

• SAP HANA service status

• SAP NetWeaver instance status

• SAP NetWeaver service status

SAP Operations:

• SAP HANA System Replication status

• SAP HANA Backup status

• SAP HANA Data backups

• SAP HANA Log backups

For example, in the following screenshot we can see how a SAP HANA database failover impacted the database availability and see the series of events as the Pacemaker cluster moved the database operations to the secondary node.

Other key features and benefits:

• Comprehensive metrics: The solution collects a wide range of metrics from your SAP applications and Google Cloud infrastructure, but if you are looking for more, you can create your own custom queries and add the metrics to your custom dashboards inside Workload Manager Observability.

• Customizable dashboards: Modify the dashboards to create alternative views to help visualize what matters most to your business and allow for faster identification of critical issues

• Alerting and notifications: The dashboards are natively integrated with Cloud Monitoring and Alerting. You can set up alerts for specific metrics and receive notifications when thresholds are breached

• Automated system discovery: The Agent for SAP automatically identifies and reconciles your SAP instances and resources when enabled, saving you valuable time and avoiding manual tagging or labeling

Getting started

Ready to experience the benefits of a single-pane-of-glass for SAP observability on Google Cloud? You can view detailed documentation and instructions on how to get started and the supported architectures here. The observability service in Workload Manager is available at no additional cost, but the underlying metrics and logs are subject to costs associated with Cloud Monitoring and Cloud Logging. You can view an estimated cost for the required features that must be enabled here.

If you would like assistance onboarding or would like a demo or more information, please contact your account representative or Google Cloud support and we will be in touch.

Looking ahead

We are committed to continuously improving SAP observability on Google Cloud and plan on adding new features and more correlation capabilities and insightful visualizations.

Six months into 2025, we’ve already published hundreds of posts here on the Google Cloud blog. We asked ourselves, why wait until the busy end of the year to review your favorites? With everything from new AI models, product launches, emerging cyber threats, company news, certifications and customer stories, here is a mid-year recap that will get you up to speed on the latest from Google Cloud and the rapidly emerging cloud and AI landscape. 

25. How Google Does It: Making threat detection high-quality, scalable, and modern

Published January 7, 2025

Google and Alphabet run the largest Linux fleet in the world, with nearly every flavor of operating system available, and see a steady stream of malicious system and network activity. Learn how our threat detection and response team detects, analyzes, and responds to threats on a vast scale.   

Read the blog. 

24. Cloud Run GPUs are now generally available

Published June 2, 2025

More and more organizations are turning to Cloud Run, Google Cloud’s serverless runtime, for its simplicity, flexibility, and scalability. And now, with the general availability of NVIDIA GPUs on the platform, developers can choose Cloud Run for applications that require powerful graphics processing, like machine learning models.

Read the blog. 

23. BigQuery emerges as autonomous data-to-AI platform

Published April 10, 2025

This is not your grandfather’s data warehouse. BigQuery is now an AI-native, multimodal, and agentic data-to-AI platform. The blog post provides an overview of the many new features and capabilities that went into this new designation, including new data preparation, data analysis, code generation and management and troubleshooting capabilities. 

Read the blog. 

22. Announcing Gen AI Toolbox for Databases. Get started today

Published February 6, 2025

Tired of building custom plumbing to connect your AI apps to your databases? This article announces the public beta of the Gen AI Toolbox for Databases, an open-source server built with LangChain that provides a secure, scalable, and manageable way to connect your generative AI applications to your data.

Read the blog. 

21. Ghost in the router: China-nexus espionage actor UNC3886 targets Juniper Networks

Published March 11, 2025

After discovering in 2024 that threat actors deployed custom backdoors to Juniper Networks’ Junos OS routers, Mandiant worked with Juniper to investigate this activity and observed that the affected routers were running end-of-life hardware and software. Learn more about the threat and how to remediate it in your environment. 

Read the blog.

20. What’s new with AI Hypercomputer?

Published April 9, 2025

It’s a platform, it’s a system, it’s AI Hypercomputer, Google Cloud’s fully managed supercomputing system for running AI and HPC workloads. As discussed at Google Cloud Next 2025, AI Hypercomputer supports all the latest and greatest compute, networking and storage infrastructure, and its software layer helps AI practitioners and engineers move faster with open and popular ML frameworks. Finally, there’s a full suite of workload management and observability tools to help you manage the thing.

Read the blog. 

19. Ipsos research shows why cloud certification matters — get certified with Google Cloud

18. Connect globally with Cloud WAN for the AI Era

Published April 9, 2025

With 202 points of presence (PoPs), powered by over 2 million miles of fiber, 33 subsea cables, and backed by a 99.99% reliability SLA, Google’s backbone network is, how do we put it? Vast. And with Cloud WAN, enterprises can now use it for their own wide area network (WAN) architectures. 

Read the blog. 

17. Expanding generative media for enterprise on Vertex AI

16. Suspected China-nexus threat actor actively exploiting critical Ivanti Connect Secure vulnerability

Published April 3, 2025

Threat actors continue to target edge devices globally, leveraging deep device knowledge and using both zero-day and now n-day flaws. This activity aligns with the broader strategy that the Google Threat Intelligence Group has observed among suspected China-nexus espionage groups, who invest significantly in exploits and custom malware for critical edge infrastructure.

Read the blog. 

15. Defending against UNC3944: Cybercrime hardening guidance from the frontlines 

Published May 6, 2025

Who is UNC3944? A financially-motivated threat actor characterized by its persistent use of social engineering and brazen communications with victims. Mandiant provides guidance and strategies for hardening systems and defenses against the cybercrime group, offering practical steps to protect against their specific attack methods.

Read the blog. 

14. MCP Toolbox for Databases (formerly Gen AI Toolbox for Databases)

Published April 22, 2025

Ready to build AI agents that can actually use your data? This article announces that our MCP Toolbox for Databases now supports the Model Context Protocol (MCP), making it easier than ever to connect your generative AI agents to enterprise data. With new support for the Agent Development Kit (ADK) and LangGraph, you can build powerful, stateful agents with intuitive code and connect them to your databases securely.

Read the blog.

13. Formula E’s AI equation: A new Driver Agent for the next era of racing

12. Google Agentspace enables the agent-driven enterprise

Published April 9, 2025

Do you want to search all your company’s information in a few clicks, or generate ideas with built-in agents that already know your company’s style? Google Agentspace now includes a no-code agent designer, a gallery for discovering agents, and two new expert agents for deep research and idea generation, all integrated directly into Chrome.

Read the blog.

11. Announcing Veo 3, Imagen 4, and Lyria 2 on Vertex AI

10. Adversarial misuse of generative AI

Published January 19, 2025

In the security realm, large language models (LLMs) open a world of new possibilities, from sifting through complex telemetry to secure coding, vulnerability discovery, and streamlining operations. However, some of these same AI capabilities are also available to attackers, leading to understandable anxieties about the potential for AI to be misused for malicious purposes.

Read the blog.

9. Ivanti Connect Secure VPN targeted in new zero-day exploitation

Published January 8, 2025

Ivanti kicked off the year by disclosing two new vulnerabilities impacting its Ivanti Connect Secure (ICS) VPN appliances. Mandiant identified UNC5221, a suspected China-nexus espionage actor that previously exploited two other Ivanti vulnerabilities as early as December 2023, as the threat actor targeting the new zero-days. Successfully exploiting one of the vulnerabilities could result in downstream compromise of a victim network.

Read the blog. 

8. Google announces agreement to acquire Wiz

Published March 18, 2025

Google Cloud shares a vision with Wiz to improve security by making it easier and faster for organizations of all types and sizes to protect themselves, end-to-end, across all major clouds, and this post announces Google’s agreement to acquire the cloud security startup.

Read the blog.

7. Veo 3 available for everyone in preview on Vertex AI

Published June 26, 2025

You dream it, Veo creates it. This post announces Veo 3, our most powerful text-to-video model yet, is now open for everyone to try in public preview on Vertex AI. Create stunning, near-cinematic videos with synchronized sound, and join the next wave of creative storytelling, now available to Google Cloud customers and partners.

Read the blog. 

6. Vertex AI offers new ways to build and manage multi-agent systems

5. Techniques for improving text-to-SQL

Published May 16, 2025

Even though it’s been around for a long time, not all developers speak fluent SQL. English, on the other hand, is pretty well-known. In this technical deep dive for developers working with natural language processing and databases, get the insights and techniques you need to enhance the accuracy and performance of your text-to-SQL conversions.

Read the blog.

4. Firebase Studio lets you build full-stack AI apps with Gemini

Published April 9, 2025

For over a decade, developers the world over have relied on Firebase’s backend cloud computing services and application development platforms to power their web applications. And with the new Firebase Studio, they can now use it to develop full-stack AI applications, integrating with the Gemini AI model.

Read the blog.  

3. Multiple Russia-aligned threat actors targeting Signal Messenger

Published February 19, 2025

As part of the ongoing Russian-Ukrainian conflict, Signal Messenger accounts are of great interest to Russia’s intelligence services for their potential to deliver sensitive government and military communications. Google Threat Intelligence Group has observed increasing efforts from several Russia state-aligned threat actors to compromise Signal Messenger accounts used by individuals of interest to Russia’s intelligence services.

Read the blog.

2. New Google Cloud certification in generative AI

One of the top questions we hear is “how do I get ahead”? This isn’t just another certification in a sea of technical qualifications. The Generative AI Leader certification is specifically focused on generative AI, and designed for visionary professionals like you — the managers, administrators, strategic leaders and more who understand that AI’s impact stretches far beyond code.

Read the blog.

1. 601 real-world gen AI use cases from the world’s leading organizations

Thank you for being a part of the Google Cloud blog community! We look forward to bringing you lots more blogs for you to devour in the second half of the year.

Pluto7, an early Google partner, has integrated Agentspace into its Planning in a Box platform Pi Agent to address inventory and demand planning challenges.

Agentspace provides an Agent Development Kit (ADK) for building specialized agents and an A2A (Agent-to-Agent) communication protocol for agent collaboration. These tools facilitate a shift from static workflows to dynamic, adaptive business systems.

Pluto7’s Planning in a Box Pi Agent is designed to complement and offer an intelligent AI layer for autonomous planning. It integrates with various enterprise systems like SAP, Oracle, and Salesforce, consolidating structured and unstructured data into a Master Ledger via Google Cloud’s Cortex Framework. This effectively creates a real-time digital twin and control tower, providing a unified view and synchronized decision-making across the supply chain.

Pi Agent functions as a real-time supply chain planning assistant, mirroring a ride-share system for inventory. It senses real-time signals, reallocates resources, and makes proactive decisions. Examples include agents for demand analysis (Ron), inventory optimization (Kassy), financial balancing (Alex), and defect flagging (Bob), all collaborating through the A2A protocol. This allows for significantly faster decision-making and a transition from reactive problem-solving to proactive planning.

Practical impact 

A case study featuring a LatAm CPG manufacturer demonstrates the practical impact. By implementing Planning in a Box + Pi Agent (now powered by Google Agentspace), the manufacturer gained real-time inventory visibility, improved order fulfillment, and is on track to reduce excess inventory by 15% while cutting manual reporting by 70%. It also has natural language access to inventory data: “How much inventory do I have for Faucet at Location Warehouse?”

The combination of Agentspace + Planning in a Box – Pi Agent creates an agentic command center for enterprises, enabling autonomous workflows, seamless ERP (SAP,Oracle,Netsuite) integration, and deployment within the user’s Google Cloud tenant delivered  in the form for a service-as-a-software. This setup offers rapid time to value, with initial use cases deployable in weeks.

Over years, Pluto7 has seen customers getting on the platform adoption journey see significant business outcomes, including a 10–20% improvement in forecast accuracy, up to a 50% reduction in inventory carrying costs, over 50% safety stock reduction, faster planning cycles, and a 10%+ increase in margin all of which is articulated in Pluto7’s 2:10 rule of revenue growth with planning error reduction.

Looking ahead 

Agentic AI isn’t the future – it’s already here redefining supply chain planning and enabling Planners to become superheros. Businesses can explore how Google Cloud’s Agentspace and Pluto7’s Planning in a Box can enable intelligent orchestration and real-time responsiveness in their supply chains.