NEAR AI staking model draws 500,000 NEAR for confidential compute

NEAR AI has launched a staking-based model for confidential AI workloads, with more than 500,000 NEAR locked within roughly two weeks of the July 30 rollout. Users stake tokens to earn recurring compute credits rather than spend principal, with 500 NEAR generating about $5 in monthly credits and 50 NEAR needed to activate the first IronClaw agent. The system runs inference inside Trusted Execution Environments using Intel TDX and NVIDIA GPU-based TEE technology, building on NEAR Protocol's earlier Intel-verified confidential inference push. The launch followed IronClaw 1.0's July 27-28 debut and benchmark results in PinchBench and OfficeQA.

The information on this website is generated using AI and we cannot guarantee its accuracy. Please use it as reference information only.