Grass Network

Thesis

Grass Network represents the purest convergence of DePIN and AI. Users share unused internet bandwidth that is aggregated and sold as training data to AI laboratories. The growth metrics are extraordinary. Monthly active users scaled from 200K to 8.5M in approximately one year, representing 42.5x growth that few crypto projects have achieved at any point in history. The revenue model is real and verifiable. Approximately 20 AI lab clients are paying for bandwidth-sourced training data, and Polychain Capital led a $10M bridge round. This is not a speculative narrative. It is a functioning marketplace with paying enterprise customers at meaningful scale. The demand side is driven by AI labs' insatiable need for diverse, real-world web data to train large language models, and Grass provides this at lower cost than traditional web scraping services. Within the broader DePIN space of 423 projects and $19.2B combined market cap, Grass stands out for its zero-hardware-cost model. Unlike Helium (hotspots), Hivemapper (dashcams), or other DePIN protocols that require physical device purchases, Grass only requires installing a browser extension. This dramatically lowers the barrier to participation and explains the explosive user growth. A user in any country with an internet connection can contribute and earn. The World Economic Forum projects DePIN to reach $3.5T by 2028 across 41.8M connected devices. Grass's pure software model means it can scale faster than hardware-dependent DePINs, though it also faces lower switching costs. Users can easily stop sharing bandwidth if incentives decrease.

Catalysts

• +Enterprise AI lab client expansion beyond ~20 current paying customers
• +Token launch or expanded staking mechanics creating additional demand for GRASS
• +AI training data demand accelerating as LLM development intensifies globally

Risks

• -Bandwidth sharing for AI training faces potential regulatory scrutiny around data provenance and consent
• -Low switching costs. Users can easily migrate to competing bandwidth-sharing platforms
• -AI labs may develop alternative data sourcing methods that bypass decentralized bandwidth networks

Common questions