🟢Background

With the development of Web3 data infrastructure, the underlying file storage protocol has gradually been adopted by many Web3 developers. However, since the underlying distributed storage protocols, such as Arweave and IPFS, can only store static files, developers only get a storage file ID. This cannot meet the requirements of structured storage, query and data modification.

Here are some pain points identified in the Web3 market for AI and decentralized applications:

- Lack of Decentralized Vector Databases for AI

As more AI projects integrate with Web3, a significant pain point is the absence of decentralized vector databases. This gap hampers the efficient storage and retrieval of high-dimensional data critical for AI tasks, forcing projects to rely on centralized solutions that undermine the core principles of decentralization.

- High Storage Cost and Lack of Built-in Querying Capabilities

As a database middleware on the top of storages, Glacier uses batch querying to reduce data access and storage costs. Combined with Bundlr, it can reduce costs by up to 90%. File storage networks are designed for immutable and unstructured data, but they’re not designed for storing data that needs to be updated frequently and the lack of built-in querying capabilities makes data retrieval difficult.

- Uninteroperability of Smart Contracts and Public Chains

Dapps of different blockchains are like isolated islands due to multi-chain and multi-storage ecosystems. By leveraging modular design DDBs, DDB can unlock data composability, interoperability and ownership across different Dapps and protocols.

Glacier’s mission is to build this programmable, modular and scalable blockchain infrastructure for storing, indexing and querying data, supercharging AI and DePIN.