How a16z-backed DATA is rethinking its blockchain strategy after $30 million Sui deal

DATA is changing its role in the stack. It will retire the blockchain formerly built under Story Protocol and move Trace to Sui. The goal is to stop spending heavily on running a Layer 1 and focus on recording AI training data’s origins, permissions and licensing activity. Sui Foundation invested $30 million to support the move.

Trace remains the central product. When a participating marketplace uploads a video, voice recording or other file, Trace creates a receipt. That receipt can record when the contribution arrived, what terms the contributor accepted, and later licensing or payment events. DATA therefore keeps building the verification layer rather than the entire blockchain network.

The change reflects a narrower strategy. DATA previously pursued many kinds of intellectual property, but found those markets entrenched. By moving to Sui, it can focus on Trace while avoiding the costs of engineers, validators, audits and infrastructure providers. Existing investors, including a16z crypto, remain involved.

Trace is DATA’s registry for AI training data. It is designed to help buyers and sellers check where data came from and what usage rights exist. That matters because AI developers may otherwise struggle to tell whether a dataset is original, properly licensed or exclusive.

The mechanism begins when someone uploads a file through a participating marketplace. For a video or voice recording, Trace creates a receipt. The record can include when the file was submitted, who contributed it and which consent or usage terms were accepted. Later licensing or payment events can also be added, creating a continuing history.

Trace is not described as a marketplace itself. DATA wants it to serve as infrastructure that marketplaces and their users can consult. Kled supplies much of the activity, adding millions of contributions daily, although DATA has not disclosed how many records have produced licensing revenue.

Kled accounts for much of Trace’s activity. Andrea Muttoni said the marketplace is adding between 3 million and 4 million contributions each day. This gives a striking picture of how quickly data records are entering the system and how large the registry could become.

The figure counts contributions registered through Trace. Those contributions can include items newly submitted by people who provide data, as well as older records that are being brought onto the system. Each record may carry information about its source, accepted terms and later licensing or payment activity.

The number has an important limit. It does not show how much data has actually been sold. DATA also did not disclose how many records have generated licensing revenue. Therefore, the daily total measures registry activity and scale, not marketplace revenue, dataset demand or successful licensing transactions.

DATA began with a broad ambition: put music royalties, patents, trademarks and other intellectual property onchain. The company raised $140 million and built its own blockchain around that goal. But changing established licensing systems proved much harder than expected.

Many intellectual-property industries already had systems in place. Large rights holders also had little interest in making their assets available through permissionless licensing. DATA was therefore spreading itself across many different types of intellectual property while confronting markets Muttoni described as saturated and entrenched.

Its experience with Poseidon revealed a more specific opportunity. AI data buyers often lacked a reliable way to check whether datasets were original, properly licensed or exclusive. Vendors even offered Poseidon supposedly exclusive datasets that were already available elsewhere, including one instance in which the team received its own dataset back. DATA narrowed its focus to solving that verification problem.

When AI developers cannot verify a dataset, they face uncertainty before using it to train models. They may not know whether the material is original, whether contributors gave acceptable permission, or whether a seller truly has exclusive rights. The core problem is weak visibility into origin and usage terms.

DATA’s experience with Poseidon shows how this can happen. Vendors offered Poseidon supposedly exclusive datasets that were already available from other sellers. In one case, the team was given its own dataset back. That example shows why a buyer cannot safely rely on a seller’s description alone.

Trace is designed to provide a checkable record. It can show when data was submitted, who contributed it, what terms were accepted, and whether later licensing or payment events occurred. The article does not claim Trace eliminates every dispute, but it positions the registry as infrastructure for clearer verification between data buyers and sellers.

The article does not mention Walrus or Seal, so it does not confirm that DATA plans to use them. In established Sui infrastructure, Walrus is designed for decentralized storage of large data objects, while Seal provides encryption and controlled access. Those functions could complement a registry that needs to reference files without maintaining every file on its own chain.

A possible arrangement would keep Trace’s receipts and verification records on Sui while storing larger data through Walrus. Seal could help protect restricted files or make access dependent on authorization. The exact design, data-retention rules and permission model are not provided in the article.

The broader reason is clear from DATA’s stated strategy. Operating a blockchain requires engineers, validators, audits and infrastructure providers. Using specialized network services could reduce that burden and let DATA focus on Trace. However, the article only confirms the move to Sui, not Walrus or Seal adoption.

A blockchain is a shared digital ledger maintained across a network rather than controlled by one database operator. Entries are linked and validated through the network’s rules, making past records difficult to alter without detection. This shared history can help different participants rely on the same record.

For digital data, a blockchain can record when an item was submitted, who contributed it and which consent or licensing terms were attached. Later events, such as a license or payment, can be added to the history. A registry such as Trace uses this kind of record to make provenance and usage information easier to check.

The record does not automatically create legal ownership or guarantee that every entry is truthful. It can, however, provide evidence of what was recorded and when. That is useful when buyers and sellers need to compare claims about originality, permission, exclusivity or later data use.