GPU-less, Trust-less, Limit-less: Reimagining the Confidential AI Cloud
AI Engineer World's Fair 2025 · 43:41
Confidential AI infrastructure
Super Protocol provides a confidential AI cloud and marketplace for AI developers and data providers to collaborate, deploy workloads, and commercialize models and datasets. Its infrastructure uses trusted execution environments to protect data during processing. Customers can deploy applications and databases, train machine learning models, and store data on clusters of Ubuntu-based confidential virtual machines connected through a virtual network.
Founder and CEO Nukri Basharuli describes software verifiability and data privacy as foundational to the company’s approach. Its architecture documented in 2024 combined confidential computing with blockchain-based orchestration. A trusted loader verifies workload integrity and publishes remote attestation reports, while confidential tunnels run inside trusted execution environments to connect users to deployed services. These mechanisms address both private execution and verification of the software handling sensitive data.
The marketplace supports a workflow in which developers lease proprietary base models, adapt them with private datasets, and deploy commercial AI agents. In the documented arrangement, model owners and compute providers receive hourly usage payments. Customer applications include TytoNyx, which uses Super Protocol infrastructure to deliver auditing services for medical-device quality management systems.
AI Engineer World's Fair 2025 · 43:41
Affiliations reflect their AIE appearances, not necessarily current employment.
The demonstrations brought the infrastructure discussion to application workflows through an n8n deployment and encrypted confidential virtual machines.
Affiliations reflect each recorded session, not necessarily current employment.