Hirys legend. @irys_xyz is not just a data storage or DA (Data Availability) layer, but a project aiming to be a 'programmable data chain that transforms data into executable assets.' This network combines storage, validation, and execution within a single L1 structure, designed so that the data itself can operate like code. In other words, it builds a 'persistent and manipulable data layer' that can continuously operate, rather than a structure that temporarily posts data to support the scaling of rollups like existing Celestia or EigenDA. The core innovation of Irys is the integrated model of storage and execution. When developers upload data, they can register specific logic through the SDK, and directly call this data in smart contracts using the EVM-based precompiled function (readBytes()). As a result, AI models can trigger automatic execution based on historical data, NFTs can embed royalty rules in their metadata, and DeFi protocols can reference permanent price data without external oracles. This represents a fundamentally different approach from the temporary DA structures of Celestia or EigenDA, which only allow for short-term data posting. The consensus structure is also unique. Irys combines useful proof of work (uPoW) with stake staking, allowing block creators to mine blocks through actual data storage proofs rather than simple calculations. Validators operate storage partitions of 16TB and are verified for data availability and replication status through continuous sampling. This provides an economic guarantee that "stored data does not disappear," clearly distinguishing it from the Tendermint structure of Celestia, which merely 'posts' data. Efficiency has also been secured in the cost structure. Irys's storage cost is a one-time payment structure of about $0.05 per GB, making it approximately 72% cheaper than AWS S3 and 16 times cheaper than Arweave for long-term preservation. While Celestia and EigenDA require ongoing posting costs, Irys is based on the premise of permanent data storage, making it suitable for data where deletion and permanence are crucial, such as AI training data, on-chain knowledge bases, and legal contracts. In terms of market position, Celestia holds a 45% share of the DA market, integrated with over 20 rollups, and EigenDA provides security through over 20 million ETH in restaking, while Irys is still in the pre-TGE stage. However, its operational history of processing over 600 million transactions on the Bundlr Network and over 855 million testnet records supports its reliability. Irys focuses on the intersection of AI, DeFi, and DePIN. In the AI sector, it collaborates with @Mira_Network to reach consensus on AI output verification on-chain, and is experimenting with AI-generated NFTs that embed royalty logic in partnership with @wardenprotocol. In the DePIN area, it has showcased a structure that verifies and monetizes real-world sensor data on-chain in collaboration with 375ai. This form of 'programmable data' is difficult to implement with Filecoin or Arweave, which only allow for simple storage. The ecosystem expansion is also rapid. Starting with about 20 partners in January 2025, it now collaborates with over 80 projects, and 74 real-use dApps have been built. The scope is broad, including AI-based networks (Morpheus, Beyond Network), infrastructure-level integrations (Polygon, Base, Aptos, Solana), and data marketplaces (Codatta). Technical development is ongoing in both Rust and TypeScript, with about 55 repositories on GitHub and consistent activity of 2-5 commits per week. However, there are clear risks that need to be addressed. First, there is market saturation. Celestia and EigenDA have already dominated the L2 ecosystem, and there is a strong trend for most rollups to revert to their own DA following Ethereum's blob expansion. Second, there is technical complexity. An L1 that processes storage and execution simultaneously has a high risk of bottlenecks, and if large-scale AI data traffic concentrates, it could burden network stability. Third, there is an imbalance in decentralization. The large storage requirement (16TB) raises the entry barrier, potentially leading to a concentration of large validators. Fourth, there is regulatory risk. The 'permanence' of data conflicts with the GDPR's right to deletion clause, which could lead to legal controversies in the European market. Nevertheless, Irys accurately identifies the direction of evolution for Web3 data infrastructure. While previous blockchains have guaranteed 'transaction immutability,' Irys adds 'data executability' on top of that. This structure could serve as the foundation for next-generation on-chain services such as AI model verification, IP rights automation, DePIN data settlement, and institutional RWA audits. Ultimately, the success or failure of Irys hinges on the 'balance of permanence and scalability' and 'how much programmable data is actually used in the market.' If it successfully passes the mainnet launch and TGE in Q4 2025, and the initial ecosystem proves real data can be permanently stored and utilized, Irys will establish itself as the most important Web3 data infrastructure after Celestia. If it fails, it risks becoming "a technically beautiful architecture that the market is not ready for." In summary, Irys continues to experiment at the intersection of AI×Data×Modular Web3 under the narrative of being "the first chain to intelligently handle data that lasts forever." At a time when Web3 is evolving from a simple transaction layer to a data-driven automation system, Irys is evaluated as a fundamental infrastructure candidate to support that change.