Case Study: Reducing cloud computing costs for Generative AI company

An early stage startup focused on Generative AI using LLMs was using a GPU cloud server consisting of 8x NVIDIA 32GB V100s with a 48C Intel Xeon Platinum 8268 CPU @ 2.90GHz, 768 GB RAM and 1TB storage. The GPUs were hosted in a data center outside the US.

They approached VALDI for help with lowering their cloud computing costs.

Given their specific budget constraints, the technical requirements of their application, and the relative efficiency of the V100 chip they were using, VALDI’s solution architecture team proposed the following configuration:

• 4x A6000 48GB / 256 GB RAM / 40 Xeon Gold 6148 vCPUs

An A6000 can be up to 2x faster than a V100 and comes with 50% more VRAM, which allows for training larger models.

In addition to the improvements in compute power, VALDI’s solution includes:

• A highly reputable and secure data center in the continental United States
• Guided onboarding with a VALDI technical account manager
• Lowering of overall industry-standard costs by almost 50%
• Flexible subscription options for monthly, quarterly and annual usage
• Ability to scale up on-demand

Join VALDI here: https://www.valdi.ai/ to access to over 30,000 GPUs and lower your costs by over 50% using a simple, all-you-can-compute monthly subscription model.

Docs: https://docs.valdi.ai/