Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

New AWS service lets customers rent Nvidia GPUs for quick AI projects | TechCrunch

Nov 01, 2023 - techcrunch.com
AWS has launched Amazon Elastic Compute Cloud (EC2) Capacity Blocks for ML, a service that allows customers to reserve access to GPUs for a specified amount of time. The service, aimed at companies running large language models, offers access to NVIDIA H100 Tensor Core GPUs instances in cluster sizes of one to 64 instances with 8 GPUs per instance. Customers can reserve time for up to 14 days in 1-day increments, up to 8 weeks in advance, and the instances will shut down automatically when the timeframe ends.

The new product provides cost certainty for customers, as they will know upfront how long the job will run, how many GPUs they’ll use, and how much it will cost. The pricing for access to these resources will be dynamic, varying depending on supply and demand. The service is available starting today in the AWS US East (Ohio) region.

Key takeaways:

  • AWS has launched Amazon Elastic Compute Cloud (EC2) Capacity Blocks for ML, a service that allows customers to reserve GPU instances for a specific amount of time.
  • The service provides access to NVIDIA H100 Tensor Core GPUs instances in cluster sizes of one to 64 instances with 8 GPUs per instance, which can be reserved for up to 14 days.
  • The pricing for this service will be dynamic, varying based on supply and demand, and users will know the total cost upfront, providing cost certainty.
  • The new feature is available starting today in the AWS US East (Ohio) region.
View Full Article

Comments (0)

Be the first to comment!