io.net expands AI toolkit with Walrus integration for encrypted model storage

io.net is bolstering its decentralized AI infrastructure by integrating Walrus’ encrypted storage protocol, enabling developers to securely train and deploy custom models without relying on traditional cloud providers.

According to a press release shared with crypto.news on June 17, decentralized cloud provider io.net has partnered with Walrus, a Sui-based storage protocol, to provide a robust infrastructure for AI and machine learning applications.

The partnership enables developers to securely store proprietary models on Walrus’ tamper-proof decentralized network while leveraging io.net’s globally distributed GPU clusters for training and inference.

At the core of io.net’s integration with Walrus lies a trifecta of innovative features aimed at addressing the shortcomings of conventional, centralized cloud services. The Bring Your Own Model framework breaks the mold of restrictive AI ecosystems, giving developers the freedom to deploy custom models without being limited to pre-approved options.

This feature is paired with Private Compute Execution, a function that enables encrypted models stored on Walrus to be seamlessly pulled into io.net’s GPU clusters while remaining cryptographed even during processing.

And in perhaps the most practical departure from Big Tech’s playbook, the pay-as-you-go model eliminates the predatory pricing traps that have made AI development prohibitively expensive for all but the best-funded teams.

The timing of io.net’s integration with Walrus couldn’t be more relevant. As demand for AI/ML infrastructure continues to grow, the industry’s dependence on centralized providers has become a major choke point — introducing risks not only around cost, but also data ownership and censorship resistance.

By decentralizing both storage and compute, io.net’s partnership with Walrus offers a credible alternative that aligns with the ethos of Web3 while delivering practical, developer-first features.