Multicoin Capital invested in Grass via both its hedge fund and venture fund.
VC & Fundraising ·
Multicoin Capital announced dual investments in Grass through both its hedge fund and venture fund, signaling confidence in the project's approach to serving machine intelligence infrastructure. Grass operates as a network where contributors share underutilized internet bandwidth in exchange for token incentives, creating a pool of resources that AI labs can tap for data retrieval and parsing. Over eighteen months, the platform has moved beyond early-stage experimentation to secure positions within the data pipelines of frontier AI research organizations.
The investment thesis rests on distinguishing between two phases of AI data demand. Grass initially addressed pretraining, where it solved the challenge of accessing geographically distributed or rate-limited web content by routing requests through millions of opted-in residential connections—avoiding the hardware acquisition or device infiltration practices of prior proxy networks. Multicoin's analysis emphasizes that the larger opportunity lies ahead: inference-time retrieval, where deployed models repeatedly search and fetch live information as agents act and reason, potentially orders of magnitude larger than one-time pretraining datasets.
The structural advantage Grass claims hinges on the capital intensity of building a web index capable of handling the recursive growth of agent-driven queries—a task historically confined to players of Google and Microsoft scale. How quickly agent systems adopt live retrieval at inference time, and whether Grass's existing lab relationships and technical infrastructure prove sufficient to capture that market, remain open questions.