Google Cloud Next 2025: Ambition to lead AI agent ecosystem

Reimagining a classic with AI

Google Cloud Next 2025 kicked off with an unforgettable moment -- a sneak peek into a collaboration between Google Cloud, Google DeepMind and the Sphere, an immersive music and entertainment arena in Las Vegas. The project? A reimagining of The Wizard of Oz.

The vision: Expand the original film beyond the frame, wrapping it around viewers to fully immerse them in the story. This ambitious feat was made possible by generative AI media models Imagen and Veo, along with Google Gemini. As the first scene unfolded, you could feel the emotion in the room. It wasn't just nostalgia -- it was awe. This reimagined Oz is set to debut in late August, and it might just redefine what storytelling means in the AI era.

The project's genesis began with a question: What if the original filmmakers weren't limited by the technology of the late 1930s? Using AI-powered "outpainting," Google teams extended the world of Oz, training multiple AI models on archival material including scripts, imagery and vision documents to craft scenes that go beyond the borders of the original film.

The project signals a broader shift in media and entertainment. AI isn't just helping retell stories; it's helping filmmakers immerse audiences in stories in ways never even remotely possible before.

Google Cloud's big bet on AI agents

Google Cloud is placing serious bets on AI agents. CEO Thomas Kurian laid out the company's vision with new products that could have big implications:

• Gemini 2.5. AI models now with enhanced reasoning capabilities for powering intelligent agents.
• Agent Development Kit (ADK). A developer kit built on Vertex AI for creating custom AI agents.
• Agent2Agent protocol. Perhaps the most critical piece to the long-term success of AI agents, this open source interoperability protocol lets agents communicate seamlessly, across platforms and providers. It's not just infrastructure; it could become the connective tissue for multi-agent ecosystems.

For AI agents to truly evolve and scale, they need to talk to each other, quickly and reliably. Think of near-zero latency across the cloud. That means hyperscalers like Google Cloud won't just be hosting AI -- they'll become the highways for agent communication.

Security, reasoning and the business of AI agents

For multi-agent systems to work in the enterprise world, they need airtight controls -- permissions, access management and guarantees of safety. Google is thinking about how to tackle this, but how well security works and how much enterprises will trust AI agents to be secure remains a big question.

Recent research from Enterprise Strategy Group, now part of Omdia, shows how important security is for AI. In the report, 27% of respondents said concerns over data privacy, protecting intellectual property and security are the biggest challenges they've encountered in implementing AI.

And then there's reasoning. AI models that can reason are more accurate, but that reasoning introduces latency.

A new Google Cloud tool helps IT teams evaluate models based on what matters to them: speed, accuracy and cost. Our research confirms that enterprises' focus is on costs; 32% of respondents said the high costs associated with AI implementation are their top challenge. It's all about tradeoffs, and this tool puts those decisions into the hands of developers.

Another key advantage for Google Cloud in the AI agent ecosystem is Google Search. AI agents powered through Agentspace can now tap into enterprise data with the power of Google Search. Think about that -- the world's most successful search engine technology can now be leveraged to hunt down the right proprietary data that AI agents need to perform their roles. That's a serious edge.

Google Cloud is all in on AI agents

As the curtains closed on Google Cloud Next, one thing was crystal clear: Google Cloud has positioned itself as the AI agent platform to beat.

The company has shown that it thought through interoperability with Agent2Agent protocol, security for enterprise trust, the need for powerful reasoning-capable models with Gemini 2.5 and low-latency infrastructure. It's also working on affordable AI computing.

AI agents aren't just another AI trend -- it's a platform shift. Google's focus on building the ecosystem around them shows that it understands both the technical depth and the business reality.

Mark Beccue is principal analyst, AI at Enterprise Strategy Group, now part of Omdia.

Enterprise Strategy Group is part of Omdia. Its analysts have business relationships with technology vendors.