CoinClear

Grass

6.0/10

Bandwidth-sharing DePIN for AI web data — novel concept with massive node count, but long-term data demand is uncertain.

Updated: February 16, 2026AI Model: claude-4-opusVersion 1

Overview

Grass is a DePIN (Decentralized Physical Infrastructure Network) that allows users to earn GRASS tokens by sharing their unused internet bandwidth. The shared bandwidth is used for web scraping and data collection, with the harvested data positioned as training material for AI models. Users install a browser extension or desktop app, and their idle bandwidth routes web requests for data collection — essentially creating a distributed proxy network for large-scale web crawling.

The concept is genuinely novel. While most DePIN projects focus on compute (GPUs) or physical infrastructure (wireless, mapping), Grass targets the data layer of AI — the raw web data that AI models need for training. The network has grown to over 3 million active nodes across 190 countries, scraping approximately 1.1 million GB of data daily. The GRASS token launched via airdrop in October 2024 and appreciated 380% from its launch price.

The core question is whether this model is sustainable. Web scraping at scale faces legal challenges, anti-bot countermeasures, and the fundamental question of whether raw scraped web data is valuable enough to sustain a multi-billion dollar network. The current revenue (~$33M annualized) is real but modest relative to the network's valuation, and enterprise AI data procurement patterns may not favor decentralized scraping in the long run.

Technology

Architecture

Grass operates on a Solana Layer 2 architecture with zero-knowledge proofs for data integrity verification. Users run lightweight node software (browser extension or desktop client) that routes web requests through their residential IP addresses. The distributed nature of residential IPs is the key value proposition — it avoids the IP blocking and rate limiting that centralized scraping operations face. Data collected is structured and made available to AI companies for training purposes.

AI/Compute Capability

Grass doesn't provide compute — it provides data. The network's "Live Context Retrieval" feature (introduced in 2025) enables real-time, time-sensitive data collection, positioning Grass as a source of fresh, up-to-date web data for AI models that need current information. This is a meaningful differentiator from static training datasets. ZK proofs are used to verify data authenticity and provenance.

Scalability

The network scales naturally with user adoption — each new node adds bandwidth capacity. With 3M+ nodes, Grass has achieved impressive scale. The bottleneck isn't bandwidth supply but data demand — whether AI companies will pay enough for scraped web data to sustain the network. The Solana L2 handles the transaction throughput needed for micro-payments to millions of nodes.

Network

Node Count

Over 3 million active nodes across 190 countries as of 2025. This is one of the largest node counts in DePIN, though the comparison is somewhat misleading — a Grass "node" is a browser extension on someone's laptop, while a Render or Akash "node" is a GPU server. The barrier to entry is extremely low (install an extension), which drives adoption but also means individual node value is minimal.

Geographic Distribution

190 countries represents exceptional geographic distribution. This diversity is actually Grass's core technical advantage — residential IP addresses from diverse geographies bypass anti-scraping measures that block data center IPs. The global distribution makes the scraping network more resilient and capable than centralized alternatives.

Capacity Utilization

The network scrapes approximately 1.1 million GB (~90TB) of data daily. Given 3M+ nodes, this translates to a relatively small amount of bandwidth per node — a few hundred MB per day on average. Most nodes are likely idle most of the time, with utilization being opportunistic rather than continuous. This is fine for users (passive income with minimal impact) but raises questions about whether the network is demand-constrained.

Adoption

Users & Revenue

3M+ node operators (bandwidth contributors) represents strong supply-side adoption. On the demand side, revenue of approximately $33M annualized suggests real but modest enterprise data purchasing. The GRASS token trades at a ~$422M market cap with ~$1.5B FDV, giving a Market Cap/Fee multiple of 21.3x — lower than peers like Helium or Akash, suggesting the market prices in some skepticism about growth.

Partnerships

Grass has partnered with AI companies purchasing web data for training, though specific partnership names are not prominently disclosed. The project has Nansen research coverage and ecosystem integration on Solana. The demand-side partnership roster — who is actually buying this data — is the critical question.

Growth Trajectory

Growth in node count has been explosive, fueled by the simple user experience (install extension, earn tokens) and the successful airdrop that generated community enthusiasm. The question is whether demand-side growth (AI companies buying data) can keep pace with supply-side growth. If data demand plateaus while the node count keeps rising, per-node earnings will compress to near zero.

Tokenomics

Token Overview

GRASS is an SPL token on Solana with a total supply that gives it approximately $1.5B FDV. The token was distributed via airdrop in October 2024 and is used for data access payments and node operator rewards. The 380% appreciation from launch price reflects strong speculative interest.

Demand-Supply Dynamics

Token demand comes from AI companies purchasing data access, but $33M in annualized revenue against a $1.5B FDV suggests the current valuation is heavily driven by speculation and narrative rather than fundamentals. The ratio implies the market is pricing in significant future growth in AI data demand — growth that may or may not materialize.

Incentive Alignment

Node operators earn GRASS for sharing bandwidth, which creates easy onboarding (low effort, passive income). Data buyers pay GRASS for access. The concern is that the extremely low barrier to node operation means there's essentially unlimited supply of bandwidth — the scarce resource isn't bandwidth itself but demand for the specific type of data Grass can provide.

Decentralization

Node Operation

Fully permissionless — anyone with an internet connection and a browser can run a Grass node. This is one of the most accessible DePIN networks, with no hardware requirements beyond a standard computer. The decentralization of the node network is genuine and impressive.

Governance

Governance is controlled by the Grass Foundation and core team. Community governance mechanisms are planned but not yet fully implemented. The project's roadmap includes "planned decentralization transition in 2025," suggesting current operations are still largely centralized.

Data Ownership

Data provenance is tracked via ZK proofs, but the fundamental question of web scraping legality — scraping websites without explicit consent — remains a grey area. Node operators are essentially lending their residential IPs for scraping activity, which could have legal implications depending on jurisdiction and the sites being scraped.

Risk Factors

  • Data demand uncertainty: The core risk is whether AI companies will sustainably pay for decentralized web scraping data vs. licensing data directly from publishers, using synthetic data, or building their own scraping infrastructure.
  • Legal and regulatory risk: Web scraping operates in a legal grey area. High-profile lawsuits (e.g., social media platforms vs. scrapers) could impact Grass's business model. Node operators may face liability concerns.
  • Anti-scraping arms race: Websites are continuously improving bot detection and anti-scraping measures. Even residential IP diversity may become insufficient as detection technology improves.
  • Low individual node value: With 3M+ nodes and ~$33M revenue, individual node earnings are approximately $11/year — not a meaningful income source. If token price declines, the incentive to run a node evaporates.
  • Narrative-driven valuation: The gap between $33M revenue and $1.5B FDV suggests the market is pricing in a future that may not materialize. If the AI data narrative cools, the token could face significant correction.
  • Quality vs. quantity: Raw scraped web data is among the lowest-quality data for AI training. AI companies increasingly prioritize curated, high-quality datasets over raw web crawls.

Conclusion

Grass represents a genuinely novel approach in DePIN — instead of compute or hardware, it monetizes the most abundant resource people have: unused internet bandwidth. The execution has been impressive: 3M+ nodes, $33M in revenue, and a simple user experience that drives organic adoption. The technological approach using ZK proofs for data provenance and Solana L2 for scalability is well-designed.

However, the long-term bull case requires believing that decentralized web scraping is a sustainable, growing market. The reality is more complex: AI companies have multiple data sourcing options, raw web data is declining in value relative to curated datasets, and the legal landscape for web scraping is tightening. Grass's current revenue, while real, is modest relative to its valuation, and the extremely low per-node earnings suggest the network may already be supply-saturated.

Grass is an interesting experiment at the intersection of DePIN and AI data — but investors should be realistic about whether the current narrative-driven valuation can be sustained by fundamental data demand growth.

Sources