Celestia's recently released Fibre data availability protocol has sparked quite a bit of discussion. According to the official statement, this system can achieve a throughput of 1TB per second by using ZODA encoding technology to expand block space. On the path of modular blockchains, Celestia's approach is very clear—separating the consensus layer, execution layer, and data availability layer, each optimizing their own. Theoretically, a performance of 1TB/s far surpasses existing blockchains, making support for AI agents, high-frequency trading, and even large-scale social applications feasible.

But there's a practical issue. The impressive numbers in the white paper are one thing; actually running at that scale is another. To handle 1TB/s throughput, nodes would require extremely strong network bandwidth and storage capabilities. Ordinary nodes simply can't handle it, which could lead to network centralization around a few top nodes, exacerbating centralization issues. Moreover, honestly, no matter how large the block space is, if there aren't enough application demands to fill it, it's just a waste of resources.

However, from another perspective, Celestia's direction is sound. As on-chain applications become more complex, data availability will indeed become a bottleneck. If Fibre can truly be commercialized, it will provide solid infrastructure support for Rollups and L2 networks, which is beneficial for the entire ecosystem's expansion. The key still depends on how far the actual implementation can go.