On Wednesday, Google held the 2025 edition of its Cloud Next conference in Las Vegas, where the company unveiled new innovations in cloud technology, including enhanced AI capabilities. Google Cloud also confirmed that it will spend approximately $75 billion in capital expenditure in 2025 to further strengthen its AI compute and cloud services.
“We’ve seen over 4 million developers using Gemini and a 20-fold increase in Vertex AI usage. Our goal is to get these AI advances into the hands of both consumers and enterprises,” stated Thomas Kurian, CEO of Google Cloud.
Day 1 of #GoogleCloudNext ✅
Here’s a taste of all the things that we announced today across infrastructure, research and models, Vertex AI, and agents → https://t.co/p6EHb0t7D8
Hint: Ironwood TPUs, Gemini on Google Distributed Cloud, Gemini 2.5 Flash, Lyria, and more. pic.twitter.com/O3jwndzA99
— Google Cloud (@googlecloud) April 10, 2025
At Google Cloud Next 2025, the company’s CEO Sundar Pichai unveiled Gemini 2.5 Pro, Google’s most advanced AI model to date, featuring improved multimodality and native image and audio generation capabilities. Pichai also teased the upcoming Gemini 2.5 Flash, a smaller and more efficient AI model, which—when run on the Ironwood TPU—offers 24 times higher intelligence than GPT-4o and DeepSeek R1.
Google also introduced its seventh-generation Tensor Processing Unit—Ironwood—which offers 3,600 times more performance than the first-generation TPU while being 29 per cent more energy efficient. The company also unveiled an AI hypercomputer, a supercomputer designed to support AI deployment with performance optimization and cost reduction.
Additionally, Meta’s recently introduced open-source model, Llama 4, will be available to Google Cloud users via Vertex AI, along with other popular open-source models. These will include support for building and scaling multi-agent systems via the new Agent Development Kit, built using an open-source framework.
Gemini can now run on Google Distributed Cloud locally, in both air-gapped and connected environments. Google Workspace also received a major update, with new features like “Help Me Analyze” in Sheets, which instantly offers insights from data. Docs can now provide an audio overview of your content, and users can also automate tasks using Google Workspace Flows powered by AI agents.
Google also announced that its private cloud network, dubbed the Cloud Wide Area Network, will soon be available to enterprises (Google Cloud customers) globally. This network is said to be optimized for application performance, offering 40 per cent faster speeds compared to the public internet while also reducing the total cost of ownership by up to 40 per cent.
The company also demoed Chirp 3, a custom voice generation model that turns text prompts into 30-second music clips, along with Lyria, a music generation model, and V2, an industry-leading video generation model.
