Grass Data Network Scales Capacity to Meet AI Demand
The network reached 8 million nodes in 2026, forcing a shift toward owned data centers to satisfy rising AI training needs.
Updated on Oct. 29, 2026 in Data Centers
Live Poll
Do you trust decentralized networks that utilize your home internet connection for AI data collection?
By mid-2026, the decentralized data network Grass scaled to 8 million active nodes, facilitating the delivery of 1 billion GB of public web data to AI labs including OpenAI and Anthropic. The project has begun shifting from residential crowdsourcing to its own data centers to manage capacity strain.
Why it matters
AI developers require massive volumes of high-quality, real-time training data, forcing providers to rapidly industrialize their infrastructure. This move toward owned data centers marks a pivot from residential peer-to-peer routing toward traditional, centralized architecture.
Grass has delivered 1 billion GB of data across its network of 8 million nodes, with peak daily scraping hitting 4 PB. The project reports monthly operating expenses between $2 million and $3 million while projecting 2026 revenue between $65 million and $75 million.
The players
Grass
A decentralized data network that leverages residential internet connections and proprietary infrastructure to provide web data for AI model training.
OpenAI
A leading AI research and deployment company that uses large-scale web datasets for training frontier large language models.
Anthropic
An AI research lab focused on developing large-scale models that utilizes high-quality web data to improve model inference and training accuracy.
The details
Grass functions by routing web-scraping traffic through residential IP addresses, which allows the network to bypass blocks often applied to corporate data center traffic. To meet AI training demand, the project is scaling its infrastructure, shifting to owned data centers that include GPU processing capacity. The platform currently requires 30 to 50 times its existing infrastructure to satisfy current AI laboratory requests.
Timeline
2025: Grass generated $17 million in revenue.
H1 2026: Grass generated $17 million in revenue.
Mid-2026: The network reached 8 million active nodes.
The Tech Race
As AI laboratories exhaust available public web data, the competition has shifted toward securing specialized, real-time datasets. Grass is positioning itself to scale infrastructure faster to stay relevant against traditional corporate data providers.
Users who participate in the Grass network by sharing their bandwidth are seeing a transition in how their connection contributes to the system's processing capacity. The shift to owned GPU centers suggests that the network is moving toward a more professionalized service model to meet sustained AI demand.
The takeaway
Grass is scaling its infrastructure 30 to 50 times to meet current AI demand for high-quality, real-time web training data. Observers should track the company’s ability to meet its 2026 revenue projection of $65 million to $75 million as it completes its transition to owned data centers.
Further reading
Learn more about the evolving landscape of infrastructure in Data Centers.
Source note: This article includes information reported by Crypto Briefing.
Live Poll
Do you trust decentralized networks that utilize your home internet connection for AI data collection?
