Aevir, a new decentralized collective intelligence protocol, has officially launched on mainnet, entering the crowded and competitive Decentralized Physical Infrastructure Network (DePIN) sector.
The project aims to tackle one of the crypto industry’s most persistent technical challenges of “useful proof-of-work” by converting the computational power of mining directly into fuel for AI model training. In a clear break from the VC-dominated landscape of 2025, Aevir has committed to a “100% fair distribution” model, following Bitcoin’s allocation model.
According to the project’s whitepaper, there is no private sale, no pre-mine, and no token allocation reserved for the founding team or investors. Instead, all tokens will be distributed via its proprietary “Proof of Intelligent Contribution” (PoIC) consensus mechanism to users who provide computational power or verified data.
Aevir’s launch strategy centers heavily on a hardware-first approach. In partnership with AI computing provider Motus, the protocol has released its first product: the NEU-X AI Station, a dedicated AI computing server that doubles as mining rigs. It uses AMD Ryzen AI Max 300 series processors to power operations. The company claims the device can reach an AI engine performance of 126 TOPS (Trillions of Operations Per Second), positioning it as a node capable of fine-tuning Large Language Models (LLMs) locally.
This sets Aevir apart from DeAI projects like ASI, Bittensor, and newer players such as io.net, all of whom are racing to aggregate distributed GPU power to service the insatiable global demand for AI compute. Aevir’s bet is that a dedicated, standardized hardware device will offer more stable and reliable performance than a network of disparate consumer GPUs.
A Bold Experiment in a VC-Dominated Era
Aevir’s entry is a high-stakes gamble on both economic and technical fronts. By choosing a true fair launch with zero team allocation, it directly appeals to a crypto community exhausted by high-FDV, low-float, VC-backed tokens. The model echoes early Bitcoin and Monero, aiming to build a genuine community of individual knowledge experts, AI developers, and miners. The trade-off: no large VC war chest for marketing or partnerships.
The concept of useful proof-of-work, which uses mining energy for actual computation rather than just random hashing, is the holy grail of blockchain efficiency, but it is notoriously difficult to implement securely. The challenge lies in verification: how does the network trustlessly prove that a node actually trained an AI model correctly without re-running the entire computation.
Aevir’s “PoIC” mechanism attempts to solve this by quantifying knowledge contributions via community voting and scoring new models objectively on test sets, offering a transparent solution in theory. However, its real-world resilience will only be proven under live and adversarial conditions.