Key Takeaways
- • Grayscale Head of Research Zach Pandl assigned three crypto projects distinct roles in an AI infrastructure value chain in an October 23 analysis: DATA Network (training data), USD.AI (GPU financing) and Pearl (GPU computation).
- • DATA Network's planned migration to the Sui blockchain, announced October 12 by the Sui Foundation and DATA Foundation, would retire its independent chain and make Walrus the default storage layer; no completion date has been set.
- • DATA's Trace product has registered more than 250 million contributions since its June launch and processed an average of more than 3 million receipts per day, according to the foundations.
- • USD.AI's CHIP token is a utility and governance token that does not represent equity, ownership, or a claim on protocol assets, per the analysis.
- • Pearl's AI-computation claims rest on unproven adoption: a June 2026 preprint under review found no inference code in dominant mining software across 8,012 workers analyzed.
- • The research announced no Grayscale fund addition, token purchase, or price target.
Summary
Grayscale Head of Research Zach Pandl, in an October 23 analysis, grouped three crypto projects into separate roles within a potential AI infrastructure chain, covering data collection, GPU financing and computing. The analysis, as reported by TokenPost, framed the projects as addressing different layers rather than competing for the same function.
DATA Network (DATA) targets human-contributed material for AI training, with tools to collect, license and track items such as photos, videos and voice recordings. On October 12, the Sui Foundation and DATA Foundation announced a planned migration of DATA's protocol onto Sui. Under the plan, DATA would retire its independent chain after the upgrade and Walrus would become the default storage layer for DATA Foundation applications. The foundations have not set a completion date. They also said Trace, DATA's product, has registered more than 250 million contributions since its June launch and processed an average of more than 3 million receipts per day.
USD.AI (CHIP) occupies the financing role, providing loans backed by graphics processing units that supply computing capacity. The loans are designed to fund GPU deployments. CHIP serves utility and governance functions and, according to the analysis, does not represent equity, ownership, or a claim on protocol assets.
Pearl (PEARL) proposes using GPUs for AI computation alongside proof-of-work mining, with miners earning network rewards under a model that uses matrix multiplication for AI workloads. The analysis cited a June 2026 preprint, still under review, by Abhinaba Basu. Basu reported finding no inference code in the dominant mining software among the 8,012 workers he analyzed, and 44 pool-accepted shares from a miner submitting random matrices. The analysis notes these findings do not establish Pearl's performance across the network today.
Taken together, the proposals address different parts of a potential AI service chain: assembling training material, financing computing equipment and assigning GPUs to computation. However, the analysis states that the projects' adoption, revenue and competitive positions remain unestablished. Pandl wrote, "Their long-term relevance will depend on adoption and competitive economics." The research announced no Grayscale fund addition, token purchase or price target.
DATA Network (DATA) targets human-contributed material for AI training, with tools to collect, license and track items such as photos, videos and voice recordings. On October 12, the Sui Foundation and DATA Foundation announced a planned migration of DATA's protocol onto Sui. Under the plan, DATA would retire its independent chain after the upgrade and Walrus would become the default storage layer for DATA Foundation applications. The foundations have not set a completion date. They also said Trace, DATA's product, has registered more than 250 million contributions since its June launch and processed an average of more than 3 million receipts per day.
USD.AI (CHIP) occupies the financing role, providing loans backed by graphics processing units that supply computing capacity. The loans are designed to fund GPU deployments. CHIP serves utility and governance functions and, according to the analysis, does not represent equity, ownership, or a claim on protocol assets.
Pearl (PEARL) proposes using GPUs for AI computation alongside proof-of-work mining, with miners earning network rewards under a model that uses matrix multiplication for AI workloads. The analysis cited a June 2026 preprint, still under review, by Abhinaba Basu. Basu reported finding no inference code in the dominant mining software among the 8,012 workers he analyzed, and 44 pool-accepted shares from a miner submitting random matrices. The analysis notes these findings do not establish Pearl's performance across the network today.
Taken together, the proposals address different parts of a potential AI service chain: assembling training material, financing computing equipment and assigning GPUs to computation. However, the analysis states that the projects' adoption, revenue and competitive positions remain unestablished. Pandl wrote, "Their long-term relevance will depend on adoption and competitive economics." The research announced no Grayscale fund addition, token purchase or price target.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.