Generative AI and large language models (LLMs) show no signs of slowing. Organizations are hungry to take action and use this transformative technology to outpace their competition. However, developing AI and ML models requires a staggering amount of computing power, preferably using infrastructure purpose-built for AI development and applications.
And that’s precisely what Google Cloud has brought to market with the general release of its A3 instances, which will be available for purchase this month. A3 was announced earlier this year, but Google Cloud chose its recent Next conference as the setting to release news about when the public could get their hands on the tech.
What is the A3 GPU supercomputer?
Google Cloud’s A3 supercomputer is driven by NVIDIA H100 Tensor Core GPUs, which enable unprecedented performance, making it faster and easier to develop and deploy AI apps on the Google Cloud platform. The combination of NVIDIA H100 Tensor Core GPUs and Google’s network advances make the A3 VM a powerful tool. Let’s look at the particulars:
- 10 times more network bandwidth than Google Cloud’s A2 VMs
- 26 exaFlops of performance, slashing the time and associated costs of training models
- 4800 MHz DDR5 DIMMs provide 2TB of host memory
When it comes to developing ML models, speed is everything. A3 VMs accelerate training so organizations can build more complex ML models more quickly and lay the groundwork for developing LLMs.
The other story here is the benefit to the consumer when tech giants join forces. The announcement of the general release of A3 instances comes after NVIDIA was named Google Cloud’s Generative AI Partner of the Year. It represents a vital tech alliance at a time when the frontrunners in generative AI infrastructure are yet to be entirely determined.
There are other elements to the partnership, too, including the general availability of H100 GPUs on Google Cloud’s VertexAI platform, expected within weeks. However, the ultimate aim is to make it easier for customers to build and deploy AI applications on Google Cloud.
“We’re at an inflection point where accelerated computing and generative AI have come together to speed innovation at an unprecedented pace,” NVIDIA founder and CEO Jensen Huang said during a fireside chat with Google Cloud CEO Thomas Kurian at the Google Cloud Next conference. “Our expanded collaboration with Google Cloud will help developers accelerate their work with infrastructure, software, and services that supercharge energy efficiency and reduce costs.”
“Google Cloud has a long history of innovating in AI to foster and speed innovation for our customers,” remarked Kurian. “Many of Google’s products are built and served on NVIDIA GPUs, and many of our customers are seeking out NVIDIA-accelerated computing to power efficient development of LLMs to advance generative AI.”
With demand for AI and ML modeling growing at lightning speed, the business world is looking for products built for purpose. Google Cloud’s A3 VM will certainly enable rapid development and deployment times.