On-chain activity
Grass Network
Decentralized network leveraging idle bandwidth/storage for AI data validation, rewarding with GRASS points.
Grass news, features & analysis
Matched from published articles, podcasts, and talks using the project name, token name, or token symbol.
-
The Intersection Of Crypto & AI | Andrej Radonjic
The Grass Network: A Revolutionary Approach to Web Scraping ... The Grass Network, developed by Wind Labs, has emerged as a groundbreaking solution to the ethical and technological challenges of web scraping.
-
Solana Ecosystem Call: February 2024
Grass Protocol: Decentralized Data for AI ... Andre from Grass Protocol introduced their innovative approach to decentralized data collection for AI training.
Grass
Grass
Grass is a decentralized physical infrastructure network (DePIN) built on Solana that turns idle residential and commercial internet bandwidth into a structured data supply for AI developers. Developed by Wynd Labs and launched in 2023, the platform lets anyone install a lightweight node client—available as a browser extension or desktop and mobile app—and earn GRASS tokens by contributing bandwidth to a network that scrapes publicly available web data at scale. By mid-2025 the network had grown to over 8.5 million users and was processing roughly 3 petabytes of web data per day.
The Problem Grass Solves
Large AI model builders require enormous quantities of clean, current web data to train and fine-tune their systems. Historically this data was gathered by a handful of centralized web crawlers operated by major technology corporations—a model that concentrates both the cost and the economic value of that activity. Grass argues that ordinary internet users already subsidize this infrastructure indirectly: corporations route web scraping through residential IP pools without users' knowledge or compensation.
Grass formalizes and democratizes this arrangement. Participants knowingly opt in, retain control of their personal data, and receive direct economic rewards for the bandwidth their nodes relay. The result is a marketplace that connects AI data buyers with a globally distributed supply of verified residential bandwidth—one that is both cheaper for buyers and more transparent than legacy data broker pipelines.
How the Network Works
Grass describes its architecture as a "Sovereign Data Rollup" and organizes it into five layers:
Grass Nodes are the user-facing entry point. The node client uses the device's spare bandwidth to relay web scraping traffic on behalf of the network. Personal browsing activity and private data are not accessible to the network; the extension only routes outbound scraping requests.
Routers sit one level above nodes. They accept batches of relayed web requests from nodes, bundle that traffic, and forward it to the validator layer. Routers are incentivized to maintain high uptime and low latency; they earn a commission on the rewards earned by the nodes staked to them.
Validators receive data payloads from routers and verify the authenticity and completeness of the web content returned before batching it for cryptographic processing.
The ZK Processor is the cryptographic backbone of the system. It aggregates validity proofs from all web requests handled by validators and submits them to Solana, creating an on-chain record of every scraping event. This zero-knowledge proof architecture makes it possible to verify data provenance without re-downloading the underlying content—critical for AI buyers who need confidence that training data was not manipulated.
The Grass Data Ledger is an immutable on-chain repository that links each dataset to its corresponding ZK proof. Edge embedding models then clean, normalize, and format the raw web content into structured datasets ready for AI consumption, completing the pipeline from raw bandwidth contribution to AI-ready data product.
GRASS Token
The GRASS token has a fixed supply of one billion tokens and serves three functions: paying for access to the network's web scraping services, staking to routers to earn a share of bandwidth rewards, and participating in governance decisions through the Grass DAO.
Node operators accumulate Grass Points based on uptime, geographic location, and volume of verified bandwidth served. Those points determine eligibility and allocation size for periodic airdrop distributions. Stakers delegate GRASS to routers through the official staking interface, with rewards accruing continuously and a seven-day unbonding period applying to withdrawals. Staking creates a shared-risk, shared-reward dynamic: routers with more stake process more bandwidth and generate proportionally higher returns for both themselves and their delegators.
Grass launched its inaugural Airdrop Season 1 on October 28, 2024, distributing 100 million GRASS tokens to more than 2.8 million eligible users. Nearly 1.5 million Solana addresses claimed tokens—making it the most widely distributed airdrop in Solana's history at that time. Airdrop Season 2, underway from mid-2025, expanded the allocation to 170 million tokens, reflecting the network's growth and the team's intent to deepen decentralization.
Growth and Funding
Grass attracted early backing from Polychain Capital and Tribe Capital, among others, raising $1 million in a pre-seed round in July 2023 and $3.5 million in seed funding in December 2023. The team subsequently raised a further $10 million bridge round, also co-led by Polychain Capital and Tribe Capital, to expand infrastructure and accelerate commercial data partnerships with AI foundation model labs.
The network reached 2.5 million active nodes across 190 countries at its 2025 peak, with more than one million concurrent user connections. The scale of the node fleet, combined with ZK-attested data provenance, positions Grass as one of the largest verifiable sources of freshly scraped public web data available to AI developers outside the traditional data broker ecosystem.
Looking Ahead
Grass has outlined "Live Context Retrieval" as a near-term capability—real-time data scraping that would enable AI models requiring time-sensitive inputs, such as market data or current news, to query the Grass network on demand rather than relying on periodic batch datasets. This would extend Grass's addressable market from foundation model pre-training toward inference-time retrieval-augmented generation (RAG) applications, where freshness of data is a competitive differentiator.
Contents
Note: inclusion in Solana Compass directory does not indicate a recommendation or endorsement of this project, its token(s) or its products. Data sourced with thanks from The Grid to aid in building these pages.Solana Token Markets