Decentralized artificial intelligence has a lot of ground to make up if it wants to compete with centralized AI providers such as OpenAI Group PBC and Google Cloud, but it’s getting a big boost today with the launch of Singularity Compute Inc.’s first cloud computing cluster.

The company, which serves as the for-profit infrastructure arm of decentralized AI pioneer SingularityNET and the Artificial Superintelligence Alliance, today announced phase one of its Nvidia-based graphics processing unit cluster, which is now up and running at a sustainable data center facility in Sweden.

The deployment is a major milestone for decentralized AI, as Singularity Compute is intended to be the cloud workhorse of one of some of its most influential players. Unlike services such as OpenAI Group PBC’s ChatGPT and Google LLC’s Gemini, decentralized AI services operate on a community owned network that runs on distributed computing resources and data. Because no one owns these services, users retain full control over their data, benefiting from enhanced privacy. It’s also cheaper and incentivizes collaboration among developers and data providers.

SingularityNET is the creator of a decentralized AI network that allows users to buy, sell and share AI models, services, applications and training data, contribute compute infrastructure, storage and more. Its goal is to democratize access to AI via an open marketplace that runs on a peer-to-peer, blockchain-based network.

Along with Fetch.ai and Cudos, it’s also a founding member of the ASI, which is a consortium of decentralized AI companies that’s working towards the goal of Artificial General Intelligence. That’s loosely defined as AI systems that surpass the intellectual capabilities of humans. At the same time, it promotes the adoption of ethical, open-source AI technologies and equitable distribution.

Singularity Compute’s GPU cloud cluster will serve as the backbone of ASI:Cloud, which is one of the flagship projects of the ASI. It’s a cloud-based service that provides access to developer tools and inference capabilities for creating and running AI applications. Developers can access GPU-based virtual machines and bare metal services, Singularity Compute said, as well as dedicated API endpoints for workloads including model training, fine-tuning and inference.

“We are taking a major step toward building the global infrastructure backbone for artificial superintelligence,” said Singularity Compute Chief Executive Joe Honan. “Our enterprise-grade NVIDIA GPUs deliver the performance and reliability modern AI demands, while remaining aligned with our core principles of openness, security and sovereignty.”

Cudo Compute Inc., an official cloud partner of Nvidia, will manage and operate Singularity’s GPU cluster to ensure “enterprise-grade” reliability, uptime and performance that matches better known GPU platforms such as CoreWeave, the partners said. They added that the cluster is just the first step in its plan to build out what will eventually become a global GPU cloud that will span multiple locations across the world.

SingularityNET founder and CEO Ben Goertzel said it’s vital for the industry to provide access to high-performance and ethically aligned infrastructure if it wants to compete with centralized AI providers. “We need powerful compute that’s configured for interoperation with decentralized networks running a rich variety of AI algorithms carrying out tasks for diverse populations,” he said.

Image: SiliconANGLE/Meta AI