Premium
This is an archive article published on April 9, 2024

Google Cloud doubles down on AI with Nvidia’s cutting-edge hardware

In a deepening partnership, Google Cloud will offer Nvidia's powerhouse Grace Blackwell platform and H100 GPU instances, turbocharging its AI capabilities for customers and partners.

google cloud nvidia featured Nvidia is bringing more AI hardware muscle to Google Cloud. (Image: Nvidia Newsroom)

Google Cloud is going all-in on AI. With new hardware partnerships, an expansive model library, and a booming ecosystem of enterprise solutions and startups, the company is trying to build the world’s most extensible AI development platform. Here’s everything announced as part of the Google Cloud Next ‘24 event.

Deepened partnership with Nvidia

Google Cloud is turbocharging its AI infrastructure by deepening its partnership with chipmaker Nvidia. The cloud giant will be among the first providers to offer Nvidia’s powerful new Grace Blackwell platform to customers and partners. Blackwell was announced last month and it’s designed to be the world’s largest GPU for datacenter-scale generative AI, and has 25 times better energy efficiency than Nvidia’s prior-generation GPU chips.

nvidia blackwell (Image: Nvidia)

In addition, Google is also rolling out support for Nvidia’s A3 Mega instances powered by H100 GPUs by month’s end, boasting double the interconnect bandwidth of the previous A3 instances.

Story continues below this ad

This expanded selection of beefy AI hardware is no surprise, as Google Cloud revealed that 90% of generative AI “unicorns” (startups valued over $1 billion) are customers.

New AI models

In addition to hardware, Google is focused on delivering a slew of AI models across its Vertex AI Model Garden. The curated library now includes over 130 foundation models spanning Google’s own offerings, open-source gems like LLaMA and Gemma, and partner models from Anthropic, Cohere, Mistral, and more.

Google recently made Anthropic’s latest Claude language models generally available. It has also partnered with Hugging Face to allow speedy deployment of hundreds of thousands of models on Google Cloud infrastructure.

For enterprises, customers and partners can tap into Google’s capable Gemini models through platforms like Salesforce, SAP’s new Generative AI Hub, and others.

Story continues below this ad

Nvidia partnership for startup acceleration

In a new partnership between the Google for Startups Cloud Program and Nvidia Inception, early-stage AI startups will gain fast-tracked access to Google Cloud infrastructure, Nvidia GPUs, and software for testing and production on Google Cloud.

Expanding AI implementation services

Recognising the importance of expert services and delivery, Google Cloud announced several initiatives to support its ecosystem of partners:

– A new Generative AI Services Specialisation for partners demonstrating technical proficiency in Google Cloud’s generative AI solutions, with initial partners including Accenture, Capgemini, Cognizant, Datatonic, Deloitte, Quantiphi, Searce, TCS, and Wipro.

– New Generative AI Delivery Excellence and Technical Bootcamps to equip partners with tools, technology, and expertise for AI-driven business growth.

Story continues below this ad

– An enhanced Delivery Readiness Portal for simplified access to important resources, ensuring optimal delivery quality for customers.

Latest Comment
Post Comment
Read Comments
Advertisement
Loading Taboola...
Advertisement