CEO Eric Park shared with Reuters that the organization plans to establish clusters of Nvidia AI chips in Texas, Virginia, and Washington. While some clusters are already operational, the complete deployment of the 24,000 Nvidia chips is anticipated by around February.
The initiative aims to provide accessible, low-cost AI computing for both short-term and long-term use, seeking to mitigate the ongoing shortage of AI chips. Voltage Park, headquartered in San Mateo, California, operates as a subsidiary of McCaleb’s non-profit Navigation Fund, actively working to address the shortage of AI chips by offering leasable large machine learning capacity.
Recently, on October 29, the organization launched a new cloud that will be available for hourly, monthly, or long-term access. Any profits generated by Voltage will be directed to Navigation, where McCaleb is not directly involved in operational or board matters.
CEO Park emphasized that their motivation stems from a belief that the current machine learning ecosystem is flawed, aiming to address this through their efforts. “It’s our belief that basically the current ecosystem for machine learning is broken,” Park said.
“With around 24,000 NVIDIA H100 GPUs, the Voltage Park cloud is one of the most powerful collections of cutting-edge ML compute in the world. Our clusters consist of 80GB H100 SXM5 GPUs fully interconnected with 3.2T InfiniBand. We currently offer bare-metal access for large-scale users that need peak performance. We will add support for short-term leases and hourly billing soon as we spin up our infrastructure along with support for familiar tools like Slurm, Kubernetes, and Mosaic for easy integration into existing training frameworks,” Voltage Park said in a blog post.
In addition, Voltage Park is conducting auctions for contracts lasting one to three months involving 1,560 GPUs. In an announcement, the company said that its overarching mission is to democratize access to machine learning infrastructure, aiming to assist startups, researchers, and larger AI labs facing challenges in obtaining the latest chips for ML training.
“The market for cutting-edge ML compute is broken. Startups, researchers and even big AI labs are scrambling to buy or rent access to the latest chips for ML training. […] We’re on a mission to make machine learning infrastructure accessible to all.”
McCaleb, a prominent figure in the cryptocurrency realm, notably founded Mt. Gox in 2014 and later co-founded Ripple Labs and the Stellar blockchain after his departure from Ripple. He also ventured into the space industry with a startup collaborating with SpaceX.
This comprehensive effort by Voltage Park underscores a pivotal step towards addressing the current challenges in AI computing, offering hope for greater accessibility and democratization of machine learning infrastructure.