Singularity Compute, the infrastructure division of AI layer SingularityNET, has announced the launch of its first enterprise-grade NVIDIA GPU cluster. The facility, deployed in Sweden in partnership with sustainable data center operator Conapto, realizes Singularity’s goal of establishing a global hardware backbone for the Artificial Superintelligence (ASI) Alliance.
The launch will fortify the infrastructure behind decentralized artificial intelligence while giving enterprises on-demand access to compute for AI workloads. This dual purpose deployment provides high-performance computing power for commercial enterprise workloads and supports the ambitious research projects taking place within the ASI ecosystem.
A foundation for decentralized intelligence
Singularity Compute’s deployment arrives at a critical juncture in the AI industry, where the demand for raw computing power is outstripping supply. The new facility offers flexible access models including bare metal rentals, virtual machine (VM) based rentals, and dedicated inference API endpoints. This flexibility allows businesses to handle everything from training and fine-tuning heavy models to running R&D workloads.
Beyond raw compute, the Swedish cluster will serve as the engine for ASI:Cloud, an AI model inference service developed in collaboration with CUDOS, the Web3 arm of CUDO. The service aims to provide a scalable path for developers, offering OpenAI-compatible APIs that allow projects to move smoothly from serverless inference to dedicated clusters.
The infrastructure itself is managed by CUDO, a long-standing NVIDIA cloud partner, ensuring the reliability and uptime required by enterprise clients. Its deployment directly addresses the primary challenges faced by businesses that require access to GPU compute, including vendor lock-in and spiraling costs due to high demand.
This “GPU squeeze” means that getting access to high-performance NVIDIA GPUs can involve lengthy waitlists or signing multi-year contracts with providers. The setup offered by Singularity Compute will also benefit teams running projects that require ultra-high performance compute.
The majority of public cloud GPU instances are virtualized, which entails a slight reduction in performance compared to running directly on the hardware. By offering bare metal access, Singularity Compute’s solution is ideal for training massive AI models in which every 1% of efficiency gained can translate to thousands of dollars in electricity and time saved.
Advancing ethical AGI
Dr. Ben Goertzel, CEO and Founder of SingularityNET and the ASI Alliance, views the physical infrastructure as a prerequisite for the development of Artificial General Intelligence (AGI), explaining: “We need powerful compute that is configured for interoperation with decentralized networks running a rich variety of AI algorithms… The new GPU deployment in Sweden is a meaningful milestone on the road to a truly open, global Artificial Superintelligence.”
Joe Honan, CEO of Singularity Compute, described the launch as “a major step toward building the global infrastructure backbone for Artificial Superintelligence,” emphasizing that the cluster combines enterprise-grade performance with the company’s commitment to openness and data sovereignty.
Early enterprise customers are already being onboarded, with additional capacity expansions and new geographic locations planned in response to demand. The Swedish rollout is the first phase of a broader global strategy that aims to make powerful, sovereign AI compute widely available to both enterprises and decentralized AI projects. In the process, it adds another infra player to the global GPU race in which enterprises are competing for the computational resources they need to succeed.
