Inspired by: A Vision of Ethereum (2025). Please consider this as a work of hard… | by Polynya | Medium
Celestia is the global schelling point for data availability, acting as the technological and cultural mesh between competing crypto tribes. Put another way, Celestia has cemented itself as the default choice for data publishing the same way Tendermint/CometBFT is the default choice for rollup consensus today.
There’s a cultural pluralism of rollups (some based, some not) building on top of Celestia. Many of these communities choose to think and settle in tia. But just as many choose to think and settle in a currency of their choice. Leading L2s and L3s from across the Bitcoin, Ethereum, and Solana ecosystems choose to use Celestia for DA -as a way of reducing costs while making the minimal amount of tradeoffs to security.
This dual ability to foster a fiercely loyal tia community and be credibly neutral towards communities who prefer to think and settle in other currencies has allowed for a cultural richness of sovereign communities – each with their own unique identities, values, beliefs – to flourish.
We’re firmly on the path to 1 billion people running light nodes. some consciously, but most without even realizing it. Since light nodes are embedded directly inside wallets – users are verifying the integrity of the system without even needing to think about it.
Celestia is on the path to 1GB blocks (more light nodes means a higher maximum safe block size) with more than 100,000 rollups, ~1000 consensus nodes distributed across the globe; CometBFT’s block propogation speed has been pushed to new heights and core developers have worked closely with the leading liquid staking and restaking protocols to ensure that a large % of these nodes are maximally geopolitically distributed and that the underlying protocols are governance minimized (with the only non-algorithmic degree of freedom being the curation of the underlying set). These staking protocols act as an important counterweight to the centralization pressures exerted by exchanges, ETFs, and the BlackRocks of this world, ensuring that no compromises are made to Celestia’s resilience and censorship-resistance.
What types of rollups?
-
rollups launching today as L2s on other ecosystems (to save costs while minimally compromising on security) e.g Eclipse: Celestia is going to meaningfully outpace Ethereum on DA bandwidth in the long term. In theory there’s nothing Ethereum can do that Celestia can. In practice, it’s really hard to steer a $500 billion ship (look how long it took to ship 4844). Ethereum will always be playing catch up in this respect. There’s a pretty meaningful path dependence here. Assuming danksharding (fullDAS with 32 MBs/slot) takes 2 or 3 years to fully roll out, the game will have shifted meaningfully by that point. By the time fullDAS is fully operational, Celestia will have cemented itself as a high security chain that users and communities trust, so it probably won’t make sense for chains like Eclipse to migrate back. Some version of this can also said to be true for Bitcoin L2s thinking of launching with Celestia today.
-
rollups who wish to use tia as their native ecosystem / gas token or as trust-minimized collateral (post zk-accounts) across celestia rollups e.g Forma
-
rollups who need to post a lot of data (i.e need to rely on sub-cent transactions) e.g. fully on-chain games, central limit order books, DePin networks, certain types of social networks (think how much smoother the friend.tech experience would have been with more abundant blockspace)
-
rollups on Solana that care deeply about user verifiability, and/or for whom posting to Ethereum is not deemed politically feasible/acceptable: Solana turbine has a lot of DA throughput, but their erasure encoding is not suitable for DAS. Making the erasure coding DAS friendly would detract from their north star: performance (maximize throughput, minimize latency) since 2d erasure encoding is fundamentally less efficient
-
rollups who care deeply about sovereignty and/or wish to escape from alignment politics: yes all rollups, like all blockchains, are potentially sovereign in the sense that forking state essentially boils down to social consensus; the caveat though, is that the chain you choose to roll up to is built on top of great myths that permeate through its social layer. and some of those myths may not be particularly conducive to the notion of sovereignty (in the broadest sense of the term)
Economic sustainability
With big blocks and an average cost of less than 1 cent per transaction, Celestia has achieved economic sustainability by providing high quality blockspace at scale, not through artificial scarcity or via restricting access to highly contested state. Even though Ethereum’s danksharding has fully rolled out, demand for blockspace has proven to be insatiable. And Celestia has been able to take advantage of this via Neilson’s law.
Thanks to both a dynamic base fee and a fee burn mechanism – validators have set a market price such that it’s not unprofitable for them to provide the service that they’re providing.
Celestia – the base layer
At this point, Celestia has moved to a fraud-proof-free DA checking scheme (possibly STARKed data availability roots). Light clients are succinctly verifying L1 state-transition finality (thanks to lightweight state transition proofs), meaning that users no longer need to trust the consensus for execution correctness.
Celestia supports a rich diversity of sharded or partial nodes – nodes that only fully download a part of the total block data and sample the rest – participating in fraud proof creation (pre zk) and block reconstruction (pre and post zk). Some of these are namespaced, while others are universal, some of these store data, while others prune it – for example namespaced partial storage nodes store the data for a set of specified namespaces, and allow themselves to advertise themselves for that namespace – effectively taking responsibility for storage of historical data blobs for a particular rollup. Equally-sharded partial storage nodes have been implemented, unlocking a protocol for pay-to-play historical blob storage providers.
We might even see some experiments with namespace programmability, leveraging zero-knowledge proofs for intra- and inter-rollup credible commitments, and perhaps going so far as to enable out-of-the-box account abstraction natively on the base layer, or even a native privacy-preserving keystore rollup.
—–
There’s a lot more I’m dreaming about but I’ll stop here. Where do you think Celestia will be in 2030? 