Google Cloud Next 2025: Big AI Updates Announced on Day 1

At its big annual cloud event, Google Cloud Next ‘25, Google announced a massive wave of new AI tech aimed at helping businesses work faster, cheaper, and more creatively. Here’s an overview of the big news from the opening keynote today.

Over the past year, Google says it rolled out 3,000+ updates to its cloud services. This includes improvements to Google Workspace (such as Gmail, Docs, etc.) and its AI tools. More than 4 million developers are already using its Gemini AI models, and companies are now building apps with tools like Vertex AI, Imagen (for images), and Veo (for videos).

A New Kind of Network

Google also said its cloud network which moves at “Google speed” is now available for billions of users globally at near zero latency. Dubbed Cloud WAN, Google says it is the new global internet backbone for the enterprise. Basically, it lets businesses plug into Google’s private internet network, making things run up to 40% faster and cheaper compared to old-school setups.

Next-Gen AI Hardware: Meet Ironwood

Google revealed its seventh-get Ironwood TPUs – powerful computer chips made for running large AI models. These chips are over 10 times more powerful than previous versions, giving businesses a major upgrade in computing power for AI tasks like image generation, chatbots, and more.

They’re also rolling out new GPUs from NVIDIA, faster storage systems, and smarter software to help all this tech run smoothly and at lower cost.

Smarter AI Models

Google launched Gemini 2.5 Pro and Flash, the newest versions of its Gemini AI. Pro is great at deep thinking tasks like medical reports or coding, while Flash is built for fast, low-cost tasks like customer service chats or quick summaries.

They also showed off Imagen 3, which makes high-quality images from text; Chirp 3 for creating custom voices with just 10 seconds of audio; Lyria for making music from text prompts; and Veo 2, which generates and edits videos with cool features like background removal and screen resizing.

AI for Everyone

Google introduced Agentspace, a tool that helps everyday employees use AI at work—no coding required. Workers can ask AI to summarize reports, generate ideas, or find documents across their company.

Agentspace has new improvements, such as Agent Designer, a drag-and-drop tool that lets anyone create custom AI helpers, and Deep Research, an agent that digs into complex topics for you and returns an easy-to-read report.

More Use Cases: From Doctors to Developers

Google shared some practical examples of how its AI agents are making big moves in the real world for clients:

• Healthcare & Retail: Doctors can search patient data more easily. Retailers like Lowe’s are using Google AI to recommend products.
• Call Centres: Google’s AI now understands emotions and can even see live video to help respond better.
• Creative Work: Brands like Kraft Heinz and Agoda are using Google’s AI tools to make ads and travel videos faster and more cheaply.
• Coding: Google says 25% of its own code is now written by AI. New tools can help with debugging, planning, and coding in teams.
• Data & Security: Their new “alert triage agent” can figure out if a cybersecurity alert is real or not, and explain why.

Google’s big push this year is to make powerful AI tools more accessible to everyone—from companies dealing with sensitive data (on local servers) to employees using a spreadsheet. The focus is on speed, cost-saving, creativity, and flexibility.

Last year, we saw Google announce Axion at its Google Cloud Next event, its first Arm-based CPU for data centres.

Check out a replay of the nearly 2-hour keynote below–there’s a lot to cover: