Google Cloud Next ’25 has just kicked off, and from the get-go, it is clear that event will serve as a powerful testament to Google’s unwavering commitment to artificial intelligence and the transformative potential of intelligent agents. The keynote address, viewed by many across the globe, underscored a clear direction: AI is not just an incremental improvement but the core engine driving Google’s innovations and its vision for the future of technology. Google aims to organise the world’s information and make it universally accessible and useful. This is targeting the traditionally harder part of the data continuum; unstructured data, as accessing structured data in a database is considerably easier. Investing in AI is the most important way that Google can advance its mission. This review delves into the key announcements and advancements specifically related to AI and intelligent agents, offering a comprehensive overview for businesses seeking to understand and leverage these cutting-edge developments.
Leading the Way with Cutting-Edge AI Models
Google Cloud Next ’25 also showcased significant progress in Google’s family of foundational AI models. A key highlight was the advancement of the Gemini series, particularly Gemini 2.5. A notable characteristic of Gemini 2.5 is its enhanced “Thinking” ability, referring to the model’s capacity to engage in a more deliberate reasoning process before generating a response. This allows for improved performance and greater accuracy in complex tasks. This enhanced reasoning capability has propelled Gemini 2.5 to the top of the Chatbot Arena leaderboard, a testament to its overall quality and versatility. The introduction of Gemini 2.5 Flash addresses the need for speed and cost-effectiveness in everyday AI applications, making it suitable for high-volume customer interactions and rapid document summarisation. For more intricate tasks, Gemini 2.5 Pro offers a substantial context window of one million tokens, with plans to expand this to two million, alongside robust reasoning and coding capabilities.
Beyond the core Gemini models, Google continues to innovate in generative media. Imagen 3, the text-to-image model, now features enhanced inpainting functionalities for improved image editing. Chirp 3, for audio generation, can now replicate voices from a mere 10 seconds of audio and includes new features for transcribing multi-speaker recordings. Lyria stands out as the industry’s first enterprise-ready text-to-music model, enabling the creation of 30-second music clips across various genres. Veo 2, the leading video generation model, has been upgraded with new features for video creation and editing, such as inpainting, outpainting, and cinematic techniques. Additionally, Notebook LM provides users with a powerful tool for analysing complex datasets, and Gemini 2.0 has seen significant advancements in its multi-modal capabilities, encompassing a wide range of generative media models.
| AI Model | Key Advancement |
| Gemini 2.5 Pro | Enhanced reasoning (“Thinking”), top ranking on Chatbot Arena, large context window (1M+ tokens) |
| Gemini 2.5 Flash | Optimised for low latency and cost efficiency for everyday use cases |
| Imagen 3 | Improved image generation and inpainting capabilities |
| Chirp 3 | Custom voice creation from 10-second audio, new transcription features |
| Lyria | First enterprise-ready text-to-music model |
| Veo 2 | New features for video creation and editing (inpainting, outpainting, cinematic techniques) |
| Notebook LM | Tool for analysing complex information |
| Gemini 2.0 | Advances in multi-modality (generative media models encompassing image, audio, music, and video generation) |
Building the Future of Intelligence with Vertex AI
Vertex AI continues to be a cornerstone of Google Cloud’s AI strategy, serving as a comprehensive and unified platform for developing and managing AI-powered applications and intelligent agents. The platform has witnessed remarkable growth, with a twentyfold increase in usage over the past year. The Vertex AI Model Garden now offers an extensive collection of over 200 foundation models, encompassing Google’s own innovations, open-source contributions, and third-party offerings. A key advantage of Vertex AI lies in its flexibility, allowing users to leverage any data source or vector database across various cloud environments. Google has also introduced enhanced capabilities within Vertex AI, including intuitive dashboards for monitoring usage, throughput, and latency, as well as improved tools for managing custom training and tuning across a wide range of models. The Vertex AI Model Optimizer intelligently directs queries to the most suitable model based on user-defined preferences for quality, speed, and cost. Furthermore, the Live API enables real-time multimodal interactions by allowing streaming audio and video directly into Gemini models. To ensure the reliability and accuracy of AI outputs, Vertex AI offers improved data connectivity and grounding capabilities, including the ability to ground responses with Google Maps and various trusted third-party sources.
The Rise of Intelligent Agents
A significant theme at Google Cloud Next ’25 was the emergence and advancement of intelligent agents. These are defined as sophisticated systems capable of reasoning, planning, remembering past interactions, and utilising various tools to achieve specific goals. Agents can think strategically, anticipating future steps and leveraging tools, including software and systems, to execute tasks autonomously under user supervision. Their role is to work collaboratively with employees, enhancing efficiency, supporting decision-making processes, and fostering innovation. To facilitate the development of these advanced systems, Google introduced the Agent Development Kit (ADK), an open-source framework designed to simplify the creation of complex multi-agent systems while providing precise control over agent behavior. The ADK has broad applications, enabling the development of Customer Agents to enhance customer service, Data Agents to streamline data management and analysis, and Creative Agents to empower creative teams. Google also unveiled the Agent2Agent (A2A) protocol, an open standard designed to enable seamless communication and collaboration between different AI agents, regardless of their underlying technology. Complementing these developments is the Agent Garden, a repository of ready-to-use agent samples and tools accessible within the ADK, allowing agents to connect to a wide array of pre-built connectors, custom APIs, and data sources.
Gemini Integrates Seamlessly into Google Workspace
The power of Gemini is being deeply integrated into Google Workspace, bringing advanced AI capabilities directly to the tools millions use daily. A new feature called “Help me Analyse” in Google Sheets empowers users to gain deeper insights from their data through AI-driven analysis and visualisation suggestions. Audio overviews in Docs provide a novel way to interact with documents by generating podcast-style summaries, allowing users to consume information more efficiently. Workspace flows introduce powerful automation capabilities, enabling users to automate repetitive tasks and make more informed decisions by leveraging contextual information. Furthermore, Gemini can now assist in Meet by summarising meeting content and even helping with scheduling follow-ups.1 These integrations are designed with security in mind, ensuring that user data within Workspace remains protected.
Google Cloud AI: The Preferred Choice for Enterprises
Google Cloud positions itself as the premier platform for enterprises looking to harness the power of AI for several compelling reasons. The platform offers an optimised AI environment characterised by leading price-performance metrics, precision, and overall quality. It provides a robust foundation for building and managing sophisticated multi-agent systems. Recognising the diverse IT landscapes of modern enterprises, Google Cloud offers an open, multi-cloud platform that facilitates the adoption of AI agents while seamlessly connecting them with existing infrastructure, including databases, document stores, and various ISV applications. This interoperability extends to interacting with models and agents from other providers. Moreover, Google Cloud delivers an enterprise-ready AI platform built for deep integration while addressing critical concerns around data sovereignty, privacy, security, and compliance with evolving regulatory requirements.
Real-World Impact: How Leading Companies are Leveraging Google Cloud AI
The announcements at Google Cloud Next ‘25 were brought to life through numerous examples of how leading global organisations are already utilising Google Cloud’s AI and agent technologies to drive tangible business outcomes. Toyota relies on Google Cloud’s robust AI Hypercomputer infrastructure. Global financial institution Schroder is among the many brands leveraging Google Cloud’s infrastructure. Airbus has strategically migrated its IT infrastructure to Google Cloud, enhancing visibility and ownership of its complex systems. Global law firm Freshfields has embarked on a strategic AI collaboration with Google Cloud, implementing Gemini within their Workspace environment and developing bespoke AI agents to transform legal processes, including using Gemini for their Dynamic Due Diligence tool. Food industry giant Kraft Heinz has dramatically accelerated its marketing campaign creation timelines using Google’s AI tools. L’Oréal is leveraging Imagen 3 and Gemini in their dedicated AI-powered content lab, CREAITECH, and utilising Veo for video content creation. McDonald’s is deploying intelligent assistants powered by Gemini on Vertex AI in its restaurants to help shift leaders identify issues quickly and to provide restaurant managers with real-time alerts for predictive maintenance. Verizon has integrated Google Cloud’s generative AI into its customer operations, achieving impressive answerability rates for customer inquiries. KPMG is actively developing an AI Advisor and Agent for the banking sector using Google Vertex AI and Gemini, and is also planning to leverage Agentspace internally. Reddit was highlighted as a key customer during the Google Cloud CEO’s; Thomas Kurian’s keynote. Mercedes Benz is integrating the Automotive AI Agent into its MBUX Virtual Assistant, creating highly personalised multimodal experiences for drivers and passengers through Gemini on Vertex AI. Home Depot is tapping into Google Cloud’s AI capabilities to enhance the customer experience and optimise its operational efficiency. Honeywell is actively developing AI agents for the industrial sector, utilising Google’s Gemini platform on Vertex AI to drive autonomous operations. Finally, Wayfair is deploying Google AI, specifically Gemini on Vertex AI, to significantly scale and enrich its extensive product catalogs and to improve employee productivity through Google Workspace. These diverse applications across industries underscore the transformative power and wide-ranging applicability of Google Cloud’s AI and agent technologies.
The Scale of Google Cloud’s AI Innovation
The sheer scale of Google’s investment and the adoption of its AI technologies are evident in several key statistics. Google’s global infrastructure includes an impressive 2 million miles of subsea cables and fibers, forming the backbone of its high-performance network. The developer community actively engaging with Google’s AI is substantial, with 4 million developers currently utilising Gemini models. The adoption of Google’s AI development platform, Vertex AI, has seen explosive growth, with a twentyfold increase in usage. Within Google Workspace, AI-powered features are being utilised extensively, generating 2 billion AI workplace assists. Furthermore, Google’s Distributed Cloud and Cloud Wide Area Network (Cloud WAN) are optimised for application performance, delivering over 40% faster performance compared to the public internet and reducing the total cost of ownership by up to 40%. These figures highlight the extensive reach and significant impact of Google’s AI innovations across its cloud platform and productivity suite.
The Foundation: AI-Optimised Infrastructure
At the heart of Google’s AI strategy lies a robust and meticulously engineered infrastructure. The introduction of the AI Hypercomputer marks a significant leap forward in this domain. This revolutionary supercomputing system is designed to streamline the complexities of AI deployment, enhance performance metrics, and optimise the financial aspects of training, fine-tuning, and deploying AI workloads. A cornerstone of the AI Hypercomputer is the unveiling of the Ironwood 7th generation Tensor Processing Unit (TPU). This latest iteration of Google’s custom-designed AI accelerator boasts a performance increase of over tenfold compared to its predecessor and an astounding 3600 times the performance of the initial TPU. Furthermore, Ironwood demonstrates remarkable energy efficiency, operating at 29 times the efficiency of the first TPU and twice the power efficiency of the previous Trillium generation. Each Ironwood pod is equipped with over 9,000 chips, delivering a staggering 42.5 exaflops of compute power, specifically optimised for the demanding requirements of AI inference. Complementing this hardware innovation is the Cluster Director, a new feature that enables organisations to deploy and manage vast clusters of accelerators as a unified computing resource, maximising both performance and operational efficiency. These infrastructural advancements underscore Google’s dedication to providing the computational muscle necessary for the next wave of AI innovation and deployment.
Conclusion: Embracing the AI-Powered Future with Google Cloud
Google Cloud Next ’25 unequivocally demonstrated that AI and intelligent agents are at the forefront of technological innovation. The advancements in AI models, the robust capabilities of Vertex AI, the emergence of sophisticated intelligent agents, and the seamless integration of AI into Google Workspace collectively paint a picture of a future where AI is deeply embedded in business operations and daily workflows. These developments offer significant opportunities for organisations to enhance efficiency, drive innovation, and deliver superior customer experiences. The emphasis on an open, multi-cloud approach, coupled with a strong commitment to security and compliance, further solidifies Google Cloud as a leading platform for enterprises navigating the evolving landscape of artificial intelligence. Businesses looking to stay competitive and leverage the transformative power of AI would be well-served to explore the comprehensive suite of AI and agent technologies unveiled at Google Cloud Next ’25.