DATA Network Sui Deal Puts AI Training Data Onchain

Sui Foundation and The DATA Foundation said the DATA Network Sui plan will put Trace, their onchain registry for rights-cleared AI training data, on the Sui stack and make Walrus the default storage layer for DATA Foundation apps, according to the official Sui blog and a matching Decrypt release dated October 12.

What happened

Trace has registered more than 250 million contributions since a June launch and processes more than three million receipts a day on average, both outlets said. Applications writing to Trace include Kled, described as an opt-in human data marketplace onboarding more than 1.5 billion records, plus Numo and Oto, which already license data to frontier labs and robotics firms. Under the DATA Network Sui upgrade, the independent DATA chain will be retired, $DATA will continue one-to-one on Sui with holder economics and governance rights at least equivalent to today, and locked or vesting positions will carry over unchanged.

The commercial package is a $30 million long-term strategic investment from Sui Foundation, which becomes the largest institutional stake alongside PIP Labs and the DATA Foundation. SY Lee, chairman of Poseidon and chief executive of PIP Labs, joins Sui Foundation as a strategic adviser. Mysten Labs co-founder Adeniyi Abiodun said the open internet has largely been scraped and that the next frontier is consented datasets with provenance regulators will accept. DATA Foundation chief executive Andrea Muttoni said joining Sui lets the project move beyond an audit trail toward encrypted reference-file storage and, later, verifiable environments for evaluations and fine-tuning.

Why it matters

AI labs face procurement pressure from the EU AI Act and training-data litigation to show where files came from and that they are cleared to use them. Trace records origin, rights and contributor payment; Walrus is meant to hold an encrypted, access-controlled copy so hashes can be regenerated from the file itself rather than from application metadata. That combination is the practical pitch of the DATA Network Sui deal: one place to source consented real-world data at machine scale, with storage and ownership tracking in the same ecosystem.

It also fits a wider week in which onchain infrastructure keeps attaching itself to AI workflows, from our coverage of Anthropic internal evaluations to the Arena valuation story in model benchmarking. For Sui, Trace is a flagship data supply already used by labs; for DATA holders, the question is execution of the chain retirement and the still-unpublished transition timetable.

Provenance is becoming a procurement checkbox, not a research nice-to-have. The Sui blog argued that EU AI Act duties and training-data lawsuits push labs toward datasets with consent trails they can show counsel. Trace's pitch is that a nurse's night note, a Jakarta dashcam clip or a Bengali support call can be registered, paid for and later proven unaltered. Walrus extends that by holding the encrypted bytes so a buyer can re-hash the file. The DATA Network Sui migration is therefore as much about evidence as about throughput: Sui's parallel execution and sub-second settlement are meant to keep up with millions of daily registrations without turning the registry into a bottleneck.

Governance and token mechanics will decide whether holders treat the move as continuity or as a soft relaunch. One-to-one migration with locked positions carried over is designed to reduce that friction, but the still-unpublished timeline leaves a window where the old chain and the new protocol could overlap. Competitors in AI data marketplaces will watch whether Kled, Numo and Oto keep licensing volume rising after the cutover. For Sui, success looks like Trace receipts that stay high and Walrus objects that actually get queried by lab pipelines—not only a press-cycle alliance.

Hardware and software claims in the announcement should be read as roadmap language. Walrus "becomes the default" for DATA Foundation apps only after engineering work ships; Trace's ability to hold encrypted reference copies is described as a result of the integration, not as a feature already live on every receipt. Adeniyi Abiodun's comment that Sui and Walrus were built to store, license and pay for AI data at frontier scale is a product thesis that will be tested against latency, cost and lab security reviews. SY Lee's move into a Sui Foundation advisory seat aligns incentives but does not by itself migrate bytes.

What's next

Both statements said the existing DATA Network will keep running until the upgrade is complete and that a detailed timeline and holder plan will be published beforehand. Watch for Walrus integration milestones, whether Trace receipt volume keeps rising after the move, and how licensing apps such as Kled, Numo and Oto expose Sui-native objects to buyers. Until that plan lands, the DATA Network Sui announcement is a strategic map rather than a completed migration. This article is for information only and is not investment advice.