What is Celestia (TIA)?

A rollup has to publish its transaction data somewhere public, so that anyone can reconstruct its state and challenge it. That publication is the single largest cost a rollup carries, and for years the only credible place to put it was Ethereum.

Celestia's proposition is to do that one job and nothing else. It does not execute transactions or run smart contracts — it orders data and makes it available, and it uses data availability sampling so that light nodes can verify data was published without downloading all of it. That is what lets block capacity scale without requiring every node to be a data centre.

The modular thesis behind it is that a blockchain's functions — execution, settlement, consensus, data availability — should be separable, with each layer specialising. It is an intellectually strong argument and Celestia is its purest implementation.

What Matcha changed

Two things at once, in opposite directions on the ledger. Block capacity went from 8MB to 128MB, a sixteen-fold increase in what Celestia can sell. And inflation was cut from 5% a year to 0.25% — one of the most aggressive issuance reductions any live network has made.

Take the second seriously. Most chains talk about reducing dilution and do not, because validators vote. Celestia cut issuance by a factor of twenty. The V9 mainnet in July 2026 then halved block times from six seconds to three, and the Fibre protocol targets another hundred-fold in capacity.

And what it did not change

Celestia's problem was never supply. Daily chain fees have been reported around $200 — not thousand, two hundred dollars — and total value locked in its ecosystem is down more than 95% from its peak. Fusaka made Ethereum's own blob space dramatically cheaper in December 2025, which compressed the price of the exact product Celestia sells, for everyone, permanently.

• Paying for blobspace — the data rollups publish, which is Celestia's entire product.
• Staking to secure the network, which validators are paid for out of issuance.
• Governance over protocol parameters, including the inflation changes made in 2026.
• Gas within the Celestia ecosystem's own chains.

The value capture logic is clean in a way most tokens' is not: rollups buy blobspace with TIA, so more rollup data means more TIA demand. There is no ambiguity about the mechanism. The ambiguity is entirely about the price, and at roughly $200 a day the mechanism is working on a volume that cannot support a large valuation.

This is the honest framing of Celestia: a well-built product in a market whose price collapsed. Data availability turned out to be a commodity, and commodities compete on cost until nobody makes money.

TIA launched in October 2023 with allocations to early backers, core contributors and the ecosystem on multi-year vesting. Those unlocks were the dominant supply pressure through 2024 and 2025 and were widely blamed for the price decline, with some justification.

The Matcha upgrade addressed the other half: annual inflation fell from 5% to 0.25%. In practical terms that removes new issuance as a factor in TIA's supply almost entirely, which is a larger change than most chains have ever made to their monetary policy.

Why it has not been enough

Cutting issuance removes sell pressure. It does not create buyers, and TIA's buyers are supposed to be rollups purchasing blobspace. At around $200 of daily fees, the total demand from the product is immaterial against the token's market capitalisation.

So Celestia in 2026 is a supply story that has been fixed attached to a demand story that has not. Anyone underwriting TIA should model rollup data spend, not inflation — the inflation problem is solved and it was the smaller one.

TIA is staked by delegating to a validator, in the Cosmos SDK model — Celestia is built with the Cosmos stack and inherits its staking mechanics.

Which means the familiar terms apply: an unbonding period of around three weeks with no early exit, and slashing that can cost delegators principal for validator misbehaviour rather than only rewards. Check the current parameters before committing.

The yield fell sharply with the Matcha inflation cut, because staking rewards were funded by that issuance. Stakers earn far less than they did, which is the direct and intended consequence of reducing dilution — the same trade, viewed from the other side.

Data availability became a commodity

Fusaka cut Ethereum's blob costs dramatically in December 2025, and EigenDA, Avail and Ethereum itself all compete for the same spend. A product with no differentiation beyond price, sold into a market where the reference price keeps falling, is a difficult business regardless of how well it is engineered.

$200 a day is the number that matters

Every other metric on this page — capacity, block times, inflation, roadmap — is upstream of revenue that is currently negligible. A sixteen-fold capacity increase does not help if the constraint was never capacity.

Ecosystem contraction

TVL down more than 95% from peak reflects rollups and applications that either did not launch, did not attract users, or moved elsewhere. The modular thesis needs a large population of rollups that need data; the population has not materialised at the scale assumed.

Unlocks and a thin float

Vesting from the 2023 launch dominated the supply picture for two years. The inflation cut helps from here, and it does not undo the distribution that already happened or the concentration it left.

Staking yield fell with the inflation cut

Rewards came from the 5% issuance. At 0.25% there is very little to distribute, so the incentive to stake — and therefore to secure the chain — is much smaller than it was. That is a coherent trade and it is a trade, and it has not yet been tested over a full cycle.