2 min read
Add as a preferred source on Google

Shelby Enters Private Beta With Three AI and Data Workloads

The decentralized hot-storage project’s first customers are using it for enterprise AI, distributed 3D rendering and physical-world data.

Storage module, GPU enclosure and action camera on a workbench / TokenPost.ai
Storage module, GPU enclosure and action camera on a workbench / TokenPost.ai

Shelby, a decentralized hot-storage protocol co-developed by Aptos Labs and Jump Crypto, has entered private beta with three customer workloads spanning AI, 3D rendering and physical-world data.

Shelby was live in private beta by Oct. 7, 2026. The first customers are Teepin, Pictor Network and PathPulse AI.

Teepin is testing the protocol as part of an enterprise AI system that works with private datasets and open-source models. Pictor Network is developing distributed 3D-rendering infrastructure that can use GPU capacity across locations and has integrated Shelby into its development environment. PathPulse AI is using real-world video to build spatial-intelligence technology.

The beta is testing distributed data and computing across providers, regions and hardware types. That setup is intended for workloads in which data and computing resources are spread across locations, potentially reducing the need to copy, stage and synchronize data.

Shelby is designed for data requiring frequent, rapid access rather than long-term archival storage. Its target applications include AI pipelines, streaming, gaming and user-generated content.

The protocol combines decentralized storage with Aptos-based coordination, cryptographic verification and data-access incentives. Its architecture is intended to reduce reliance on centralized cloud storage while supporting the performance requirements of cloud-based applications.

Listed Shelby capabilities include sub-second reads, pay-per-view access and real-time monetization. The private-beta announcement did not provide measured results for workload volume, throughput, latency or cost.

Aptos introduced Shelby in June 2025 with stated specifications and targets including sub-second reads, 600-millisecond end-to-end Aptos finality, capacity of 30,000 transactions per second and a $0.000005 transaction fee. Those figures describe the project’s stated capabilities and targets, not measured performance from the private beta.

Simon Yoon

Reporter

Simon Yoon reports on blockchain technology for TokenPost. Send corrections or tips to info@tokenpost.com.

Loading…