What Is BIP 110? Bitcoin's Data Limit Fight, Explained for Hashrate Buyers

Here is a question Bitcoin had to answer this year. If people start paying to store pictures and tokens inside a money ledger, should the network stop them? BIP 110 was one loud answer. It proposed one year of strict rules to squeeze that data out, it needed 55 percent of the network's mining power to say yes, and it won barely 2.5 percent. A tiny breakaway chain that enforced it anyway mined two blocks and stalled. The story sounds finished. The question underneath it is not. And if you buy hashrate, a small piece of the answer quietly belongs to you.

Start from zero. What is Bitcoin actually agreeing on?

Strip everything away and Bitcoin is a list of transactions that thousands of computers keep in identical copies. Every ten minutes or so, one miner wins the right to add a page to that list. The page is a block, and a block has limited space. That is the whole machine. Scarce space, an open auction for it, and a rulebook every computer checks before accepting a new page.

Now ask, who writes the rulebook? There are really two of them, and the difference decides everything that follows. Consensus rules are the hard ones. Break them and every honest node rejects your block, no matter how much computing power stands behind it. Relay policy is the soft one. It is each node choosing which transactions to pass along, like a bouncer with personal taste. Policy can be walked around, because anyone can hand a transaction straight to a miner. Consensus cannot. So when people disagree about what belongs in a block, the argument eventually climbs to the consensus layer. BIP 110 is that climb, written down in December 2025 by a pseudonymous author called Dathon Ohm.

Why would anyone store pictures in a ledger?

Because permanent, uncensorable space is valuable for more than money. Since 2022, inscription protocols have used clever corners of the transaction format to embed images and tokens directly on the chain, and the Ordinals project turned that into a real market. The people doing it pay real fees, and fees are income for miners, so pools kept accepting the transactions. Then Bitcoin Core version 30 shipped in late 2025 and relaxed its default limit on one data carrying field. Part of the community read that as policy picking a side. If the soft rulebook would not hold the line, only the hard one could. That conviction became BIP 110, the Reduced Data Temporary Softfork, backed mostly by the Bitcoin Knots ecosystem and the Ocean mining pool, and opposed by voices as different as Michael Saylor and Adam Back.

What exactly did it propose?

The name carries both key facts. Temporary, because the rules expire on their own after 52,416 blocks, about one year. Reduced data, because every rule narrows a field the data protocols rely on.

• New output scripts are limited to 34 bytes, with OP_RETURN outputs allowed up to 83 bytes
• Individual data pushes are capped at 256 bytes, with carve outs for witness scripts and redeem scripts
• Taproot annexes are prohibited and Taproot control blocks are capped at 257 bytes
• Spending through undefined witness versions and OP_SUCCESS opcodes is blocked while the rules apply

Coins created before activation were grandfathered, so nothing already on the chain would become unspendable. And because the whole package expires by itself, a network that merely disliked it could simply wait it out. As soft forks go, it was built to be an easy yes. So why did almost nobody say yes?

How does a network without a boss say yes?

It counts blocks. Miners signal readiness inside the blocks they mine, which makes activation a vote weighted by hashrate. Taproot activated in 2021 after 90 percent of blocks signaled. The older standard was 95. BIP 110 asked for just 55 percent, and it added a sharp edge called mandatory signaling. In the final counting period, software enforcing BIP 110 would reject any block that failed to signal, whatever the earlier tallies said. Supporters called that decisiveness. Critics saw a design that turns low support into a guaranteed split instead of a quiet failure.

The count was not close. Of the 2,016 blocks before the deadline, 51 signaled. That is 2.53 percent against a 55 percent bar. The mandatory phase still began, at block 961,632 on August 8, 2026. AntPool mined that block without the signal. Nodes running BIP 110 rejected it, and the Ocean pool built the alternative block that a small breakaway chain followed instead.

Then arithmetic took over. A minority chain inherits the difficulty of the chain it leaves, and difficulty only adjusts every 2,016 blocks. With about 2.5 percent of the network's power, the breakaway chain found two blocks in roughly eight hours while the main chain added dozens. At that pace its next difficulty adjustment sits nearly a year away. The fork did not lose an argument. It ran into math. Barring a surprise, the whole deployment expires on September 1, 2026, unactivated.

What does any of this mean if you rent hashrate?

Here is the part most coverage skipped, and the reason we wrote this up. When you buy hashrate and point it at a pool, the pool's node builds the block template and the pool decides what to signal. During the mandatory window, identical machines pointed at Ocean and at AntPool sat on opposite sides of a chain split. Renting hashrate means lending your consensus vote to whichever pool receives it. That is not a reason for alarm. It is a reason to know where your power points, because Bitcoin will argue about something again.

There is a money thread too. Data stored on the chain pays fees, and fees are one of the two ingredients of hashprice, the daily revenue a unit of hashrate earns. A rule that caps data transactions caps a source of fee demand, which is one reason miners were never likely to volunteer for it. The honest scale check, from mempool.space on our publication date: fees were about 0.66 percent of total block reward over the trailing day. Neither the proposal nor its failure moves today's number much. The deeper question, what kind of fee demand Bitcoin should welcome, just went unanswered at the consensus level.

And a quieter thought for solo miners. Point hashrate at your own node and the template is yours. What your candidate blocks contain, and what they signal, nobody decides on your behalf. The moving parts behind this story are all things we keep live on the site.

• The live hashprice index, where fee revenue shows up
• Solo mining, where you control the block template
• Network difficulty, the math that stalled the fork
• Bitcoin quantum readiness, our standing protocol analysis

So what should you actually do?

Honestly, almost nothing. The main chain's rules did not change, no balances or payouts were touched, and the breakaway chain holds no meaningful hashrate. There is no setting to change on IonGrid and no urgency worth manufacturing. The useful work is calibration, not reaction.

• Know your pool's posture. Signaling is visible in every block header on any explorer, and your pool's choice is effectively your choice
• Watch fee share rather than headlines. Consensus fights about data are ultimately fights about future fee revenue
• Judge any fork by difficulty math first. A minority chain that inherits mainnet difficulty is measured in blocks per day, not in conviction

The disagreement itself is not over. Luke Dashjr, who maintains Bitcoin Knots, wrote early code for a proof of work change years ago, and developer Chris Guida recently rebased it as a contingency, with no date attached. The people behind BIP 110 have not changed their minds about data on the chain.

So we end where we began, with a question. When the next argument arrives, and it will, do you know which way your hashrate votes? Our hope is a simple one. That Bitcoin keeps settling its fights the way it settled this one, in the open, with visible signals and unforgiving math rather than closed rooms. And that everyone holding a sliver of the network's power, including you, holds it knowingly. That is the whole reason explainers like this live on our site.

Sources

Primary references, current as of our research date of August 11, 2026.

• BIP 110, Reduced Data Temporary Softfork, the full specification
• CoinDesk on the breakaway chain stalling after two blocks
• Cointelegraph on the start of mandatory signaling
• CoinDesk on the governance fight around the proposal
• mempool.space mining data, source of the live fee share figure