Wall Street holds the coins, a few pools mine the blocks, and Core decides the code. A plain-English look at where Bitcoin really centralized, and why a small group of users forking it in 2026 is the best evidence the original idea still works.
Bitcoin was sold to the world as money nobody controls. Sixteen years later, an honest observer has to admit the picture is more complicated. A handful of asset managers custody a growing share of all coins through ETFs. Two or three mining pools produce most blocks. One software project’s release decisions ripple across the entire network. None of that is secret; it is the natural result of Bitcoin becoming big, valuable, and institutional.
This article does two things. First, it looks squarely at where Bitcoin has centralized and why, without the usual hand-waving. Second, it explains why the most encouraging news of 2026, at least to us, was a small community of users forking Bitcoin over a principle. If Bitcoin were truly captured, that could not have happened. The fact that it did is the system working as designed.
When spot Bitcoin ETFs launched in 2024, they solved a real problem: they made Bitcoin exposure available inside brokerage and retirement accounts with one click. They also created a structure Bitcoin’s designers would have found strange. A single custodian, in most cases one company, holds the coins behind the largest funds. Millions of investors own a claim on Bitcoin but have no keys, no ability to move it, and no vote if that custodian’s terms, jurisdiction, or regulators change.
We laid out the tradeoffs in Bitcoin ETFs vs. real Bitcoin, and the same logic applies to the corporate treasury companies that hold coins on shareholders’ behalf. None of this is nefarious. But the more Bitcoin sits in a few vaults, the more those vaults become a point of leverage for anyone who wants to influence the network. Convenience and sovereignty pull in opposite directions, and for the last two years convenience has been winning.
Bitcoin mining was meant to be permissionless, and technically it still is. In practice, block production has concentrated in a small number of pools, because pooling is the only way a miner earns steady income. We explained the mechanics in our mining-pools guide. The important detail is that in most pools, the pool operator, not the individual miner, chooses which transactions go into a block. A few operators therefore shape what the network enforces, day to day, far more than the thousands of machines plugged into them.
This is where the 2026 story starts. Those pool operators have a commercial incentive that ordinary users do not share: high-fee data transactions, the kind that embed images and tokens into blocks, pay them well. When users asked the network to restrict that data, the pools’ answer was, in effect, no.
Bitcoin has no CEO, but it does have a reference implementation, Bitcoin Core, and its defaults become the network’s defaults simply because most people run it. That is a soft kind of power, and for years it was exercised cautiously. In late 2025, Core version 30 removed its long-standing default limit on OP_RETURN data outputs, the main channel for arbitrary data. Reasonable people disagreed about that choice. What is not debatable is that a small group of maintainers made it, and the rest of the network was expected to follow.
A centralized bank is still a bank. A centralized Bitcoin is a contradiction. The entire value proposition, the reason it is worth anything at all, is that the rules cannot be changed by whoever happens to be powerful this year. Every step toward concentration, whether in custody, mining, or code, chips at that promise. Investors who bought Bitcoin as a hedge against institutions should care deeply when Bitcoin starts to look like one.
Which is why the question matters: when push comes to shove, who really decides what Bitcoin is? For years it was a theoretical debate. In August 2026 it got a practical answer.
The short version, which we tell in full in our BIP-110 explainer: a community of node operators, largely running Bitcoin Knots, activated a rule set called BIP-110 that restricted non-financial data in blocks. The mining pools ignored it. The chain stalled after eight blocks. And then, rather than accept that pools could veto users, the community changed Bitcoin’s proof-of-work algorithm to BLAKE2b, an algorithm the pools’ hardware cannot run, and kept going without them.
Set aside for a moment whether you agree with the money-only philosophy. Look at what the event demonstrates about each of the three centralization pressures above.
In other words, the fork is the stress test Bitcoin’s centralization critics have been asking for, and the answer was: the exits still work. Users can still leave, and they can take the entire history, every coin, and every key with them.
It is worth understanding why BIP-110 supporters cared enough to do something this drastic, because it is not really about images in blocks. It is about who Bitcoin serves. Every full node must store everything that goes into a block, forever. Cheap, small, money-only blocks mean a node can run on modest hardware in a bedroom, which means thousands of ordinary people can independently verify the ledger, which is the only thing that keeps custodians, pools, and developers honest. Expensive, data-heavy blocks push node-running toward companies with budgets, which quietly hands verification back to institutions.
Seen that way, the BLAKE2b chain’s small blocks and data limits are not an aesthetic preference. They are a direct response to the centralization this article opened with. Whether the approach succeeds is an open question. That someone tried it, and that the design allowed them to, is the encouraging part.
Nothing dramatic. But a few practical conclusions follow.
Enthusiasm should not become salesmanship. As of early September 2026 the BLAKE2b chain had no exchange listing and no established price, a hash rate small enough that a well-funded actor could reorganize it, and a name most of the industry does not accept. Its supporters are a minority, and its data limits are scheduled to expire in 2027 unless renewed. It may grow into a durable network or remain a principled experiment. None of that changes the point of this article: a Bitcoin that can still be forked by its users over a principle is not a captured Bitcoin.
Centralization is real, and it is worth naming plainly. But the response to it is also real, and it happened this year, on the record, at block 961,640. We think that is the most hopeful thing to happen to Bitcoin in a long while.
Get the Bitcoin Wallet Security Checklist — the 7 rules every beginner should follow to keep their savings safe.
We’ve sent the Bitcoin Wallet Security Checklist to your email. Should arrive in under a minute.