Lambda Launches 1-Click Clusters for On-Demand Nvidia H100 GPU Access
Lambda, a leading GPU cloud company, is set to launch its innovative 1-Click Clusters, offering customers on-demand access to Nvidia H100 GPUs and Quantum 2 InfiniBand clusters.
In collaboration with Nvidia, Lambda provides flexible cluster configurations ranging from two to 64 nodes, allowing businesses to utilize only the computing power they require, particularly if 24/7 GPU access is unnecessary.
AI Agents: The Future of Machine Learning
Robert Brooks, co-founder and vice president of revenue at Lambda, emphasized the service’s versatility for companies developing their own AI models. “Training presents significant computational challenges, both hardware and software,” Brooks noted. “Starting an AI venture today typically demands substantial investment to build a large GPU cluster, and it can take time to initiate that process.”
Lambda alleviates many logistical hurdles related to GPU rentals, such as complicated negotiations with providers for server access. By preparing Nvidia H100 GPUs and InfiniBand networking for immediate deployment, Lambda streamlines the training process for new model projects.
Many companies prefer leveraging pre-trained AI models, fine-tuning them with their own data due to the high costs of developing foundational models. Nvidia GPUs can be prohibitively expensive, and limited access to suitable data centers can hinder smaller firms.
To access necessary computing power, smaller AI companies often opt to rent GPU space or partner with cloud providers. While mainstream cloud services like AWS and Microsoft Azure provide Nvidia GPU access, Brooks highlighted that their primary focus is on inference—deploying the AI model—rather than training. Additionally, contracts for GPU clusters can span a year or longer.
Lambda’s 1-Click Clusters cater to organizations that need short-term GPU access without the burden of long-term contracts. Companies can reserve the desired number of nodes for a minimum duration of two weeks, with pricing varying based on node count and reservation length.
Founded in 2012, Lambda recently secured $320 million in funding, boosting its valuation to $1.5 billion.