
Singapore, 8 July 2025: Pundi AI, a protocol for collecting, labeling, and verifying AI training data on-chain, has announced the launch of Data Pump, a breakthrough feature that enables contributors to transform their datasets into tradable, tokenised BEP-20 assets. The platform has achieved significant milestones to date, featuring 122,097 datasets from 28,152 data owners, and over 4.5 trillion data tokens issued. This innovation further strengthens Pundi AI’s position as a leading decentralised ecosystem for high-quality AI training data.
With the appetite for AI models and agents growing rapidly, Pundi AI’s Data Pump feature addresses the lack of qualitative training datasets, a challenge that has become central to AI’s mass adoption. Data Pump will additionally address a critical imbalance: tech giants profiting from AI models, while contributors who generate and curate the underlying data receive minimal compensation with no ongoing stake in the value they create.
Through its bonding curve model, dataset owners can tokenise their AI datasets into tradeable BEP-20 tokens, raise community funding, and deploy an AI Agent that showcases the data’s utility. This turns a static dataset into a living on-chain asset that earns for its creator, and which can be traded as tokens on decentralised exchanges like PancakeSwap.
“The AI training dataset market is projected to grow to over $9.58 billion by 2029, yet its contributors are locked out of the value chain,” said Zac Cheah, Co-founder & CEO at Pundi AI. “Decentralised AI has the potential to rise above tokenised outputs, and enable community-driven valuation of the inputs. With Data Pump, we’re turning passive data into active, liquid assets, giving ownership and upside back to the people who make generative AI as we know it possible.”
The token’s pricing will be defined by a bonding curve pricing akin to the Pump Fun model, which rewards early supporters while ensuring continuous, market-driven valuation without traditional exchange requirements. The system includes MEV protection and a dual-contract deployment that prevents front-running and ensures fair launch conditions.
Getting Started as a Contributor
To get started, contributors can easily connect their wallets and verify ownership of a dataset NFT on Pundi AI’s web platform. Then, they can mint an BEP20 token tied to that dataset, with custom allocation and metadata, and open it up for community support. Once the funding goal is met, the token launches on a decentralised exchange, with bonding curve pricing applied; Contributors can then deploy the linked AI agent and make their data an active, accessible on-chain asset.
Why Now?
Pundi AI’s Data Pump launch comes as industry giants race to secure high-quality training data. In June 2025, Meta invested over $14 billion in Scale AI, a company whose valuation hinges on massive human input, yet offers minimal upside to the data contributors powering it. Data Pump’s decentralised system ensures the value of AI data flows back to its source, creating sustainable income streams rather than exploitative labour relationships.
Apart from long-term recurring income and IP ownership rights for dataset contributors, Pundi aims to bring transparent, community-audited datasets with clear provenance to AI developers. By inspecting data quality through integrated AI agents before purchase, Pundi AI expects to build auditable models that support ethical AI development and fair compensation.