Grayscale Highlights Pearl’s 580% Gain in AI Infrastructure Review

Grayscale has highlighted three cryptocurrency projects targeting artificial intelligence infrastructure, including Pearl, which the asset manager reported had risen 580% over three months. Its assessment connects blockchain applications with the data, financing and computing resources needed to deliver AI services.

The October 23 research note, written by Grayscale head of research Zach Pandl, examines Pearl alongside DATA Network and USD.AI. The reported price increase uses CoinGecko data through October 22, 2026, making it a historical performance figure rather than a measure of future returns.

Pearl seeks to let graphics processing units run AI workloads while also helping secure a proof-of-work blockchain. In Grayscale’s description, computing providers could receive PRL cryptocurrency rewards alongside the revenue they earn from supplying AI computation.

An earlier announcement from Together AI offers a concrete example of that approach. On May 15, the company announced a partnership with Pearl Research Labs and introduced an inference endpoint that it said was discounted by more than 25%, with the reduction offset by the future value of cryptocurrency issuance.

Together AI described the underlying mechanism as proof of useful work, using proofs of matrix multiplication performed during AI training and inference. The company said the endpoint was its first product integrating Pearl and outlined plans to broaden the offering and eventually let customers receive some currency emissions directly.

That arrangement illustrates an important distinction between computing demand and token-based support for pricing. Together AI explicitly connected the discount to the anticipated value of newly issued cryptocurrency, meaning the economics described in its announcement involve both the AI service and the value attributed to its associated rewards.

The other projects address different infrastructure needs. Grayscale describes USD.AI as arranging financing backed by GPU equipment and related cash flows, while DATA Network develops systems for collecting, licensing and tracking human-contributed training material, including images, video and voice recordings.

DATA’s infrastructure plans received additional detail in an October 12 announcement from Sui Foundation and DATA Foundation. The organisations said DATA Network and its Trace registry would join the Sui stack, with Walrus becoming the default storage layer for applications operated by DATA Foundation.

According to that announcement, the arrangement is intended to give AI laboratories a place to obtain training data with documented usage rights and provenance. Trace records information about the material, while Walrus provides storage, separating the management of permissions and records from the files themselves.

Taken together, the examples show distinct attempts to connect blockchain systems with commercial AI activity rather than a single shared business model. Grayscale’s outlook remains conditional: sustained adoption and competitive economics will determine their relevance, and its performance disclosure warns investors not to expect exceptional historical returns to repeat.