5 Takeaways from Google Cloud Next 2025

I was fortunate to be part of a strong Silk contingent in Vegas at “Google Cloud Next 2025”.  Being there, I had the opportunity to witness first-hand the impressive rate of change in IT being driven be the rapid adoption of AI across the industry. Silk was proud to see both our products and our customers featuring in a variety of sessions.  Two of our joint customers Wayfair and JB Hunt were both highlighted for the innovation they are delivering with AI. The conference was very well attended, every seat in the Michlob Ultra Arena was taken for the keynote and the Allegiant Stadium was packed for the Next at Night concert.  The energy and enthusiasm of so many attendees was truly infectious. 

We were also overwhelmed not just by the number of attendees or the view over the Bellagio Fountains, but by the quality of the conversations we had at the Silk joint event with Eviden at the Vegas Eiffel Tower with fantastic views of the Strip. 

Here were my biggest technical highlights from the conference as I reflect upon the week. 

AI Agents Take the Spotlight

Agents are taking over everywhere. Google introduced a suite of autonomous AI agents throughout its platform, capable of executing complex tasks beyond simple conversations. They operate across various platforms, thanks to the new Agent2Agent protocol, enabling inter-agent communication. AI Agents that talk to each other are slightly scary and highlight the need for proper planning for oversight or potential performance impact or security concerns.  These concerns around multi-agent architectures are becoming an ever-stronger use case for customers deploying Silk when they need to expose production data. 

Vector embedding advances are driving AI into relational database platforms

One of the most strategic shifts covered was the deep integration of vector embeddings into traditional databases. Google announced native support for vector search within AlloyDB Omni—its downloadable, self-managed PostgreSQL-compatible database—bringing the power of AI-driven search and semantic understanding directly into relational workflows. This move eliminates the need for separate vector databases in many use cases, allowing developers to unify structured and unstructured data handling. Silk was pleased to be featured as the “Ultimate Performance Layer for AlloyDB Omni”.  

Google Cloud Infrastructure Enhancements

Google announced a wave of infrastructure enhancements to accelerate AI workloads. Key highlights included the launch of new high-performance processors “TPU v5p” and the new custom Arm-based chip “Axion,” designed for high-performance AI training and inference. New VM shapes became public including C4, H4, M4 and Z3 storage optimized varieties.   Google announced storage infrastructure upgrades including enhanced performance for Filestore, Turbo Mode for Persistent Disk and some data management features improving tiering and cost optimization. We are looking forward to continuing to incorporate advances in the Google Cloud stack into the Silk platform as they become widely available. 

Enhanced Enterprise AI Integration

The conference highlighted Google’s commitment to integrating AI across their platforms especially into their enterprise solutions. Possibly the biggest impact might be seen with the introduction of Gemini AI models into Google Workspace to enhance productivity tools like Gmail and Docs. They also announced a new unified security platform, GUS, that leverages AI for advanced threat detection and response, bolstering organizational cybersecurity. ​  

The Value of Real Time Data for AI

Real-time data took centre stage at Google Cloud Next 2025 as a critical enabler for next-gen AI applications. This featured in many of Google’s enterprise application advancements. To me, this is a great use case for Silk’s ability to instantly autonomously clone databases, especially for use cases like fraud detection, personalization, and predictive maintenance. By minimizing the gap between event occurrence and AI response, Google is underscoring a simple but powerful truth: real-time data isn’t just faster—it’s smarter. 

To summarise my thoughts from a very valuable few days: from the conference, it is clear, that nearly all businesses are already realising true value from AI in many ways. I was surprised everyday by some of the new capabilities I saw demonstrated. 

On a personal level, it has been a minute since I made it to Vegas and as a city, it is unique and still eye opening.  The event was a reminder that the value of meeting customers, partners, colleagues and friends should never be underestimated and even makes the jet-lag worthwhile.