Google Cloud Next'25: Ironwood, Gemini 2.5 Flash and updates in Vertex AI
In addition, over 500 global customer stories were shared, including names such as the Singapore Government, McDonald's, Mercado Libre, Samsung, among others that are already transforming their operations with Google Cloud.
Google Cloud kicked off its annual event today, Next '25, showcasing its momentum in the adoption of business artificial intelligence (AI) and unveiling a wave of innovations in every layer of its fully integrated AI technology products.
In addition, more than 500 global customer stories were shared, including names like the Government of Singapore, McDonald's, Mercado Libre, Samsung, among others who are already transforming their operations with Google Cloud. This progress is supported by the more than 3,000 product advancements in Google Cloud and Workspace that happened last year, with a growing impact of AI in Latin America , where more and more organizations from different sectors are innovating with this technology.
"The opportunity that AI presents is unlike anything we've witnessed before. It has the power to improve lives, increase productivity, and reimagine processes on a scale previously unimaginable," said Thomas Kurian, CEO of Google Cloud.
"Google has been incorporating machine learning into our products for over 20 years, and our investment in AI is deeply rooted in our main mission: to organize the world's information and make it universally accessible and useful. With Google Cloud, we extend this mission, seeing AI as the most powerful catalyst to assist our customers, developers and partners in advancing their mission," Kurian added.
Follow us on: 
Google DeepMind available for Cloud customers based on Google DeepMind research, Google Cloud is offering rapid innovation to its customers with the Gemini model family.
- Gemini 2.5 Models now on Vertex AI: Gemini 2.5 Pro, the model capable of performing deep data analysis, extracting key information from dense documents such as legal contracts or medical histories, is now on Vertex AI in public preview, and today Google Cloud announced that Gemini 2.5 Flash will also be coming to Vertex AI for organizations that need to prioritize speed, low latency, and cost efficiency. Gemini 2.5 Flash is the ideal engine for highly responsive virtual agents and real-time summarization tools where efficiency at scale is key.
- Advancements in Google AI models: In addition to Gemini, Google Cloud is the only company that offers models in every format: image, voice, music, and video. These models continue to advance, for instance, Image 3 now has enhanced capabilities of image generation/filling; Chirp 3 creates custom voices from 10 seconds of audio input; the first enterprise-level text to music model, Lyria, has also arrived, which transforms text prompts into 30-second music clips; and Veo 2 has new capabilities for video generation and editing.
Follow us on: 
Google WAN: A next-generation global network, created for the Gemini era Google announced its global private network, Cloud Wide Area Network, or WAN, is now available for businesses around the world. Google's global network provides performance up to 40% faster compared to the public internet. It spans over 200 countries and territories, offers near-zero latency for services such as Gmail, Google Photos and Google search and is used to train Gemini, Google's most advanced AI model, with which over 4 million developers are already working.
Google Cloud: The Platform for Scientific Research. In order to facilitate and empower new scientific discoveries, Google Cloud is integrating the best of Google DeepMind and Google Research with new infrastructure and capabilities like:
- AlphaFold 3, developed by Google DeepMind, can predict the structure and interactions of all life molecules with unprecedented accuracy and is available for non-commercial uses and deployable via Google Cloud Cluster Toolkit.
- WeatherNext, the set of AI models from Google DeepMind and Google Research that perform fast and accurate weather forecasts, are now available on the Vertex AI Model Garden so that organizations can customize and implement them in scientific research applications and in various industries, accelerating discovery in biology, medicine, and other areas.
AI Hypercomputer: Unleashing unprecedented computational power, Google Cloud's AI Hypercomputer is a supercomputing system comprising of hardware, software, and consumption models. It is the internal foundation for services such as Vertex AI and is designed to power the development and efficient execution of large-scale, low-cost AI applications. Key announcements made at Cloud Next '25 include:
- New TPU, Ironwood: The seventh-generation TPU, Ironwood, will arrive later this year. It is the most powerful chip ever built by Google and will enable the development of the next generative AI models. Ironwood is one of several new components in the Google Cloud AI Hypercomputer architecture and compared to the first publicly available TPU, it achieves a performance 3,600 times better.
- Gemini on Google Distributed Cloud (GDC): In partnership with NVIDIA to bring Gemini to NVIDIA Blackwell systems and with Dell as a key partner, organizations now facing strict regulations, sovereignty, latency or data volume can access the latest AI technology on their premises with Gemini on Google Distributed Cloud (GDC).
Vertex AI: The Comprehensive Platform for AI Innovation The use of Vertex AI has experienced explosive growth, multiplying by 20 in the past year, resulting in thousands of AI applications created by our customers. Hence, new advancements are announced optimizing the ability of customers to manage their AI initiatives and promote a multi-agent ecosystem within Vertex AI.
- Over 200 models in Vertex AI Model Garden: Customers can access Google's models; third-party models from companies such as Anthropic, AI21, Mistral, CAMB.AI and Qodo; and the complete portfolio of open-source models from The Allen Institute on Vertex AI.
- Grounding with Google Maps: Vertex AI now offers grounding (the ability to connect a model's output with reliable databases) on Google Maps to ensure that agent responses that depend on location context are objective and up-to-date; as well as additional third-party sources such as Cotality, Dun & Bradstreet, HG Insights, S&P Global and ZoomInfo.
- Other innovations include: Vertex AI Model Optimizer, which automatically selects the best performing model based on quality, speed, and cost preferences; Live API for real-time conversational interactions through streaming audio and video directly to Gemini models; Agent Development Kit (ADK), an open source framework for easily building and managing agents; Agent2Agent (A2A) Protocol, developed to promote secure communication between agents built with different technologies and providers; and Agent Garden, a collection of samples and tools for simplifying the development and integration of agents with cloud data and APIs, ensuring the interoperability of agents created with multiple frameworks and their connection with business applications.
Google Agentspace: Empowering Every Worker with AI. Google Agentspace, which was launched in December 2024, brings together capabilities similar to Google's search engine, but adapted to a business context, conversational AI, Gemini, and third-party agents to enable users within an organization to find and synthesize information, converse with AI agents, and perform actions in other business applications. It also allows conversation with AI agents, such as NotebookLM, which is already used by more than 100,000 companies, and take action with their business applications. The Agentspace announcements include:
- Integration with Chrome Enterprise: allows employees to search for and access all their business resources directly from the search box in Chrome, streamlining workflows and increasing productivity.
- Agent Gallery: provides employees with a unique view of the available agents throughout the company, including those from Google, internal teams, and partners, making it easier to discover and use the agents.
- Agent Assembler: A no-code interface for creating customized agents that automate daily work tasks or enhance knowledge. It helps employees, regardless of their technical experience, to adapt the agents to their workflows and individual needs.
- Idea Generation Agent: uses a "tournament" to effectively rank ideas according to the criteria defined by the employees, and can help employees to refine or generate new ideas. Brainstorming and problem solving.
- Deep Research Agent: explores complex topics and provides findings in a comprehensive and easy-to-read report.
- Google Agentspace Search at GDC: In public preview starting from the third quarter of 2025, Google Agentspace search provides all knowledge workers with ready-to-use business capabilities that unify access to all their data in a secure and permission-conscious manner.
Google Workspace: AI-driven productivity Google Workspace harnesses all the capabilities of Gemini in popular applications like Gmail, Documents, Drive, Meet - providing more than 2 billion monthly AI assists to business users. The innovations of Workspace to further empower users with AI include:
- Help me analyze: This function transforms Google Sheets into a personal business analyst, identifying data insights without the need for explicit prompts, allowing you to make data-driven decisions with ease. Help me analyze will arrive to Sheets later this year.
- Google Documents audio overview: it will now be possible to create high-quality audio readings with a human-like sound or podcast-style summaries of documents.
- Workspace Flows: helps to automate daily work and repetitive tasks such as managing approvals, analyzing customer information, organizing your email, summarizing the daily schedule, and much more.
- Gemini in chat conversations: By typing @gemini in any chat conversation, a detailed summary and the next steps to complete can be provided. It is a simple way to eliminate noise and ensure that the team is aligned and advancing. It will be available in Labs in the coming weeks.
Google Security: driving secure innovation with AI. The new innovations to help businesses improve their critical cyber defense capabilities include:
- Google Unified Security (GUS): GUS makes every aspect of the professional's experience more efficient with Gemini. It creates a unique, scalable, search-capable security data fabric across the entire attack surface; provides visibility and detection and response capabilities in networks, endpoints, clouds, and applications; and automatically enriches security data with the latest Google threat intelligence for more effective detection and prioritization.