Title: Google Cloud Amplifies AI Capabilities at ‘Next’ Annual Conference
Source: Datanami
Key Takeaways:
– Google Cloud introduced new processor options for AI training.
– New GenAI capabilities were unveiled in Vertex, alongside the AI model for generating videos.
– The company launched AI-powered security features and a new security-focused browser, Chrome Enterprise Premium.
– The TPU v5p and A3 Mega were announced for faster machine learning and AI operations.
– Google Cloud’s first ARM-based CPU, Axion, was introduced.
Google Cloud, a leading player in the cloud computing industry, recently made a series of announcements in its ‘Next’ annual user conference. The conference was a showcase of hardware improvements, AI innovations, enhanced storage functionalities, and advancements in data analytics.
**Revolutionizing Hardware for AI Training**
A standout revelation was the Tensor Processing Unit, TPU v5p, and A3 Mega – two new processor options designed to enhance AI training. The TPU v5p is capable of training language models three times faster than its predecessor. The A3 Mega, built on Nvidia H100 GPUs, offers GPU-to-GPU networking bandwidth that’s two times faster than before, further enhancing AI training and inference, as explained in a blog post by Mark Lohmeyer, VP and General Manager of Compute and ML Infrastructure at Google Cloud.
Google Cloud also unveiled Axion, its first ARM-based CPU designed to handle general-purpose workloads. This CPU boasts up to 50% better performance and 60% higher energy efficiency compared to comparable X86 based instances.
**Boosting AI Hypercomputer Storage and Vertex AI**
Google Cloud augmented its AI hypercomputer storage with a block storage offering named Hyperdisk ML. This addition enables up to 12 times faster model load times over conventional alternatives, effectively speeding up AI training and inference processes.
Furthermore, Google Cloud added new capabilities to Vertex AI, its AI development platform. The additions include new models such as Gemini 1.5 Pro, Imagen 2.0, and open-source models Mistral 7B, Mixstral, and Code Gemma among other notable entries.
**Powering Analytics with GenAI through BigQuery and Looker**
Google’s GenAI capabilities are now available in the data analytics warehouse, BigQuery, facilitating users to fine-tune their Gemini models with stored data. In Looker, a data analytics and BI product, Gemini enables business users to chat with enterprise data, thus generating visualizations and reports.
Google is extending GenAI to MySQL and Postgres databases. This development will assist in SQL code-assist in Database Studio, manage customers’ database fleets in Database Center, and provide aid in Database Migration Service.
**Bolstering Video Creation and Security with AI**
Google Vids, an AI-powered video creation app, was officially launched. Integrated with Vertex AI, the company aims to simplify the process for teams to collaborate, create, and share content.
Enhancing security, Google introduced new features blending AI and security. They include an investigation feature incorporated in Gemini and also adopted Gemini in Threat Intelligence. To further fortify organizational defenses, Google launched Chrome Enterprise Premium, a security-focused web browser.
**Conclusion**
With these cutting-edge enhancements and unparalleled innovation, Google Cloud continues to revolutionize its AI capabilities. AI could likely become an even more integral part of all tech-savvy industries in the near future, thanks to constant advancements and integrations like these.
