Neutral

Grayscale highlights data, credit, and compute models behind AI tokens

На русском
3 min
Grayscale highlights data, credit, and compute models behind AI tokens

Grayscale’s October 23 research note evaluates distinct crypto-and-AI infrastructure roles: who supplies training data, who finances hardware, and who receives payment when computing capacity is used. The analysis cites DATA, USD.AI and Pearl as examples. These are Grayscale’s selections and expectations, not sector-wide proof.

TL;DR

  • DATA, USD.AI and Pearl illustrate three roles: data provision, GPU-backed lending and compute networks.
  • Hardware financing adds credit risk alongside blockchain mechanics.
  • Service utility and token value are separate: rights differ by token design.

Three business models sit behind one broad AI label

In the note, research head Zach Pandl outlines: DATA’s contributed training data, USD.AI’s GPU-secured lending and Pearl’s combination of AI computing with blockchain rewards. Each model serves a different customer: a data network requires buyers for licensed datasets, a financing protocol requires borrowers who repay debt, and a compute network requires useful workloads with measurable delivery. Grouping them under an AI-token label obscures material differences.

USD.AI’s documentation formalizes the credit structure: loans secured by GPUs and cash flows, recourse generally limited to collateral with defined exceptions. Its yield-bearing instrument draws on loan payments and reserve-related income.

This raises underwriting questions delivered through a new distribution format: what if equipment devalues faster than modeled, how concentrated are end customers paying for capacity, and who controls collateral recovery and liquidation on borrower failure. A blockchain record can improve claim transparency without resolving these risks.

The operating-versus-financing split also appears in Bitdeer’s mining and AI disclosures: machines, funded capacity and realized customer revenue are related but not interchangeable metrics.

The token must connect to the underlying activity

Infrastructure success does not automatically transfer to every associated token. Governance tokens, income claims and utility tokens confer different rights and exposures.

USD.AI’s depositor guide separates instruments: USDai as non-yield-bearing, sUSDai as income-accruing, and CHIP for governance. Naming a protocol does not specify the holder’s exposure.

For data networks, paying contributors can expand supply, but durable demand depends on whether buyers view the data as lawful, relevant and difficult to source elsewhere. Token incentives can bootstrap activity without proving customers will keep paying once incentives decline.

Tokenized-asset growth must be assessed through the rights embedded in each token. AI-facing consumer applications still require users who return because the product is useful.

Grayscale’s report frames business models to evaluate. Evidence of durability will come from adoption metrics, unit economics and enforceable claims.

Written by the News Desk and edited by Samuel Rae.

Original publication

Authors credited by the source: NewsBTC Editorial Team

Published by Holder based on an external source.