NEAR Protocol Launches Confidential GPU Marketplace for Enterprise AI
NEAR Protocol has launched the Confidential GPU Marketplace, a compute network secured by Trusted Execution Environments (TEEs) for enterprise AI workloads. The platform claims to be three times faster than hyperscalers and provides hardware-level attestation for data privacy and security.
- NEAR co-founder Illia Polosukhin, one of the eight co-authors of the seminal 2017 Google paper "Attention Is All You Need" that introduced the Transformer architecture, is a key figure behind the protocol's push into user-owned AI. His vision stems from early challenges in paying a global workforce for data labeling, which led to the creation of NEAR as a payment and coordination layer. - The platform's "confidential" security is rooted in hardware through Trusted Execution Environments (TEEs) like Intel TDX and NVIDIA Confidential Computing. This isolates computation in a secure enclave within a processor, cryptographically proving that data and models remain private even from the infrastructure operator. - This marketplace is a core component of NEAR's broader strategy to enable an economy of autonomous AI agents. Use cases are already emerging, such as the TravAI platform, where AI agents autonomously handle corporate travel booking, approvals, and settlement on a regulated blockchain layer. - Decentralized GPU networks are primarily targeting the high-volume market for AI inference, data preparation, and fine-tuning, rather than competing with hyperscalers on frontier model training. The value proposition for enterprises is lower cost and access to a distributed global network of hardware, which can reduce latency. - The platform is part of a larger research initiative called Decentralized Confidential Machine Learning (DCML), which aims to create a fully decentralized machine learning cloud for both private and open-source models. - To accelerate development on the platform, venture firm DWF Labs has established a $20 million AI Agent Fund to support teams building autonomous AI x Crypto technologies within the NEAR ecosystem. - The architecture is designed for agent-to-agent commerce through open standards like the agent interaction/transaction protocol (AITP). This allows autonomous agents to discover each other, negotiate agreements, and execute transactions enforced by smart contracts on the NEAR blockchain.
