Google Cloud Platform Technology Nuggets — November 16–30, 2025
Welcome to the November 16–30, 2025 edition of Google Cloud Platform Technology Nuggets. The nuggets are also available on YouTube.
AI and Machine Learning
Google has been named as a Leader in the inaugural 2025 Magic Quadrant for AI Application Development Platforms, positioning it highest for Ability to Execute. Check out the blog post and download a complimentary copy of the 2025 Gartner® Magic Quadrant™ for AI Application Development Platforms report.
Google Cloud has announced the general availability of Anthropic’s most advanced model yet, Claude Opus 4.5, on Vertex AI. This new model delivers frontier performance in coding, complex agentic workflows, and vision tasks while being significantly more efficient and still costing one-third the price of its predecessor, Opus 4.1. Key enhancements include improved tool use with dynamic discovery and a massive context window for deep analysis. You can deploy it today via Vertex AI Model Garden using provisioned throughput for predictable performance. Check out the blog post.
Identity and Security
The second Cloud CISO Perspectives for November 2025 is out. This edition features Phil Venables, Google Cloud’s current strategic security advisor and former CISO, who shares his thoughts on how the role of the CISO is evolving in the AI era, and how organizations should shift their cybersecurity approach from fire stations (reactive) to flywheels (self-reinforcing security).
Data Analytics
Looking to get a quick summary on whats new with Google Data Cloud across Databases, Data Analytics, etc , you can always check out the Whats New with Google Data Cloud. This weeks updates covers the 30-day free trial available for Google Cloud SQL Enterprise databases and highlights key innovations from Customers using Google Cloud databases.
You’ve built your agent using Agent Development Kit (ADK) and now need to understand how the agent is performing? That’s a real challenge but in what is a fine piece of engineering, the new BigQuery Agent Analytics plugin for the Agent Development Kit (ADK) allows developers to stream agent interaction data (tool calls, token usage, and latency) directly into BigQuery with a single line of code. By centralizing this telemetry, you can use BigQuery’s advanced analytics and vector search to pinpoint where users get stuck and more. Check out the blog post.
The BigQuery Data Transfer Service has received enhancements to improve reliability and performance. New features include support for more complex scheduling options, improved error reporting for failed transfers, and higher throughput for transfers from third-party SaaS applications. These updates are designed to make your data ingestion pipelines more robust and easier to manage.
BigQuery AI has seen a good consolidation of advanced data agent and machine learning tools. This simplifies the experience for data practitioners, allowing them to access Gemini models, vector search, and model inference directly using SQL. The goal is to reduce context switching and let you run AI workloads right where your data resides. Check out the blog post.
Infrastructure and Networking
Google Cloud continues to expand its global infrastructure with the announcement of TalayLink, a new subsea cable connecting Australia and Thailand. This cable will significantly improve latency and bandwidth capacity between Southeast Asia and Oceania, supporting the growing demand for digital services and AI workloads in the region.
In addition to TalayLink, is Dhivaru, a new subsea cable connecting the Maldives to the global network.
Managing IP address overlap is a common headache in complex organizations, especially during M&A or when connecting with partners. To ease this process, is a new feature introduced “Private NAT for networks with overlapping IP spaces”, which allows you to map overlapping IP ranges to unique subnets, enabling seamless communication between VPCs without the need for complex re-addressing projects. Check out the blog post.
Network resilience is critical for high-availability applications. A new detailed post explains how Protective Reroute improves network resilience on Google Cloud. This capability automatically detects potential failures or maintenance events in the underlying physical network and proactively reroutes traffic before packets are dropped, ensuring your workloads maintain steady performance. Check out the blog post.
Databases
Google has been named a Leader in the Gartner Magic Quadrant for Cloud Database Management Systems, recognized for its “AI-native Data Cloud” vision. Check out the blog post and download the complimentary 2025 Gartner Magic Quadrant for Cloud Database Management Systems report.
As a developer who is looking to try out databases on Google Cloud, one of the areas of friction has been the cost of these databases. Recognizing this, Google Cloud has launched a dedicated 30-day free trial for Cloud SQL, giving you a risk-free way to test enterprise-grade features. The trial provisions an Enterprise Plus instance with 8 vCPUs and 64GB of RAM. This is perfect for developers wanting to benchmark high-availability configurations, Data Cache performance, or simply explore the platform without upfront costs or commitment. Check out the blog post.
In a move that brings powerful forecasting to your operational data, TimesFM models are now available in BigQuery and AlloyDB. TimesFM (Time Series Foundation Model) is Google’s breakthrough model for time-series forecasting. By integrating this directly into AlloyDB and BigQuery, you can now run accurate forecasts on sales, inventory, or traffic data using simple SQL queries, eliminating the need for complex external ML pipelines. Check out the blog post.
Containers and Kubernetes
Google Cloud engineers have pushed the limits of scalability and demonstrated successfully running a 130,000-node GKE cluster, doubling the previous limit. This feat was enabled by architectural innovations like Kueue for job queuing and a distributed storage backend based on Spanner. The cluster sustained a throughput of 1,000 pods per second and managed over 1 million objects. Check out the blog post for more details that provides interesting information of this was achieved across phases.
Developers and Practitioners
Gemma 3, Google’s latest multimodal open model, presents a great opportunity for developers looking to use open models for their own domains. The key thing is also how do we deploy these models on Google Cloud. There are a couple of paths available and to make that journey easier, two codelabs have been published to help developers work their way through the steps. The two paths are:
- A serverless approach using Cloud Run with vLLM for instant APIs
- A platform approach using GKE Autopilot for scalable, orchestrated inference
Check out the blog post for more details.
As AI agents become more autonomous, ensuring they behave as expected is a challenge. What you need as the blog post suggests is a a methodical approach to Agent Evaluation, which outlines a framework for testing non-deterministic agent flows. It proposes a three-pillar framework:
- Success & Quality (did it do the job?)
- Process & Trajectory (did it use tools correctly?)
- Trust & Safety (is it robust against attacks?).
The post covers metrics for reasoning steps, tool usage accuracy, and safety checks, providing a rigorous way to validate your agents before deployment.
Fine-tuning open models for specific domains is a high-value task and here is a detailed step-by-step guide to fine-tuning MedGemma for breast tumor classification. The tutorial walks you through preparing the medical imaging dataset, configuring the fine-tuning job on Vertex AI, and evaluating the model’s performance, showcasing the power of specialized models in healthcare.
Vertex AI Studio (in Google Cloud Console) has a good set of capabilities that lets you work visually with Gemini3 and many other foundational models. Vertex AI Studio has received a major overhaul, introducing Agents as tools. New commands like “/Prompt” for refining instructions, “/Evaluate” for assessing quality, and “/Build” for generating application code, streamline the development loop. The update also adds collaboration features, allowing teams to share prompts and track version history, making it easier to a move to a team engineering workflow. Check out the blog post.
Business Intelligence
The Gemini CLI continues to grow in capability, now adding Looker extensions, which allow you to access Looker and Looker Conversational Analytics, directly from the command line. You can now run natural language queries against your Looker data models or manage Looker resources without leaving your command-line workflow. Check out the blog post.
Learning Path : AI prototype to production
If you have been developing AI applications and are still grappling with what it means to deploying these applications in production, a structured path is a good starting point. Google Cloud has launched a comprehensive, free learning path designed to bridge the gap between AI prototype and production. The curriculum (10 modules) covers everything from MLOps and model monitoring to security and governance. It is designed to equip developers with the skills needed to take GenAI applications from a notebook to a scalable, secure production environment.
Get started on this journey. Bookmark the post too, since not all modules are currently available and the post will get updated as they get published.
Write for Google Cloud Medium publication
If you would like to share your Google Cloud expertise with your fellow practitioners, consider becoming an author for Google Cloud Medium publication. Reach out to me via comments and/or fill out this form and I’ll be happy to add you as a writer.
Stay in Touch
Have questions, comments, or other feedback on this newsletter? Please send Feedback.
If any of your peers are interested in receiving this newsletter, send them the Subscribe link.
Google Cloud Platform Technology Nuggets — November 16–30, 2025 was originally published in Google Cloud – Community on Medium, where people are continuing the conversation by highlighting and responding to this story.
Source Credit: https://medium.com/google-cloud/google-cloud-platform-technology-nuggets-november-16-30-2025-411ce1a9089f?source=rss—-e52cf94d98af—4
About the Author
