Bitcoin outlook, 2026-08-23

Most posts explaining that Bitcoin is badly designed deserve the scroll they get. This one is harder to dismiss. On July 7, Eli Ben-Sasson wrote that capping Bitcoin at 21 million coins does not make sense. He is the CEO and a co-founder of StarkWare, a co-inventor of STARK proofs, and one of the founding scientists behind Zcash. He knows exactly which rule he is attacking, and it is the number most people hold Bitcoin for.

He is not asking anyone to print coins without limit. Today's rule caps the stock: no more than 21 million coins will ever exist. His proposal replaces that with a cap on the flow, no more than roughly 4 percent new issuance in any year, forever, a figure he tied to an upper bound on human population growth. He is explicit that Bitcoin should keep a hard, predictable monetary bound; he wants the bound to sit on the rate of creation rather than on the total. Answering that with "21 million is sacred" is not an argument, it is a slogan. If the only defense of a rule is that it must not be discussed, it is not being defended at all.

The idea also has a history. In 2016, a Princeton team (Miles Carlsten, Harry Kalodner, Matthew Weinberg and Arvind Narayanan) published work showing that a Bitcoin funded only by fees is not merely poorer but differently behaved: because fee income arrives with high variance, forking a wealthy block to steal its fees can become the profitable strategy, which changes what miners are paid to do rather than only how much. In July 2022, Bitcoin Core contributor Peter Todd argued in "Surprisingly, Tail Emission Is Not Inflationary" that because coins are continuously lost, a fixed emission per block converges toward a stable accessible supply instead of growing without end. That June, Monero had already shipped its own version, a permanent reward of 0.6 XMR per block. Ben-Sasson is not floating something no serious person has considered.

His first argument is the strongest, and it is simply true. Coins disappear. Someone dies without passing on a seed phrase, a drive goes to the tip, a wallet from 2011 sleeps on a laptop nobody will switch on again. Current research puts roughly 3.1 million coins permanently lost out of just over 20 million mined, and Ben-Sasson pushes it to the limit: as time goes to infinity, all keys will be lost. But losing units is not destroying value. When a coin becomes unreachable, the value does not evaporate, it redistributes; the same total spreads across fewer reachable coins, so every remaining coin is worth mechanically more. Satoshi Nakamoto said as much in the "Dying bitcoins" thread in June 2010: lost coins only make everyone else's coins worth slightly more, and you should think of it as a donation to everyone. Michael Saylor has floated doing it deliberately, describing burning the keys to his own holdings of more than 17,000 coins as a pro rata gift to every other holder.

Even granting that loss is a problem, the remedy assumes a measurement nobody has. Nobody knows the true rate at which coins vanish, and the proposal needs that number to calibrate what it is compensating for. Worse for the argument, the rate looks like it is falling rather than rising: one 2025 study of self-custody attributed about 1.57 million lost coins to that channel, with roughly 98 percent of those losses occurring before 2020, which is what you would expect as hardware wallets, multisignature setups, inheritance planning and institutional custody spread. BlackRock is not going to announce that it mislaid a few hundred thousand coins. So the trade on offer is to correct an unmeasurable and possibly shrinking leak by introducing a permanent and perfectly certain dilution. And the shortage it guards against is not one Bitcoin can suffer: a currency is not consumed like bread, prices adjust, and a single coin already divides into one hundred million satoshis, a unit that could itself be subdivided later without touching monetary policy at all.

The 4 percent deserves its own look. World population growth peaked in 1963 at around 2.2 percent a year and now runs near 0.9 percent; the United Nations projects the population to peak around 2084 at just under 10.3 billion and then decline. A flat 4 percent forever is therefore well above the thing it was meant to track, but the shape matters more than the level. An issuance schedule genuinely indexed to demography would rise, decelerate, and tend toward zero. That is a description of Bitcoin's existing schedule. Taken seriously, the demographic argument points back at the design it was raised against.

The security budget is the one open question here, and it deserves better than either panic or dismissal. Fees have recently run under 1 percent of miner revenue, a ten-year low, while the subsidy of 3.125 coins per block halves again in 2028 and approaches zero long before the last satoshi is issued around 2140. But the pessimistic version assumes everything else stands still, and nothing in Bitcoin does: the price, ASIC efficiency, the cost of energy and above all the demand for block space are all variables, and what the network sells is space in a block, so any use case that bids for that space addresses the problem directly. It also assumes future miners will mine only for direct profit. If Bitcoin becomes a strategic reserve asset, an international settlement layer or collateral inside the financial system, a state, a custodian or a bank could rationally mine at thin or negative margin to keep its own capacity to produce blocks, the way a government maintains an army or a backup grid: not because it pays, but because it must remain available. And permanent issuance does not actually guarantee security, because security depends on the value of what is distributed, not the count. Four percent of a currency that lost its credibility buys less protection than a fraction of a percent of one that kept it.

Which brings us to the objection that actually matters, and it is not economic. Ben-Sasson is right that 21 million has no inherent meaning. It could have been any number. That is the point. A number that must be justified is a truth claim, something that can be correct or incorrect, optimal or improvable. 21 million was never a truth. It is a coordination point, the thing everyone settles on precisely because everyone knows everyone else has settled on it, like driving on the right. Neither side of the road is superior; what matters is that nobody reopens the question at every junction. It does not need to be optimal, it needs to be uncontested. The moment you treat it as a parameter to be tuned, you destroy the property that gave it value, which is that it is not negotiable. Open it for the security budget and it can be opened for lost coins, then for a systemic crisis, then for a large custodian's hack (reissue the stolen coins? restore the balances?), then for a political emergency, and eventually the question of how many coins exist has an address and someone at it. Michael Saylor put the risk plainly in July: the greatest threat to Bitcoin is ambitious opportunists advocating protocol changes. Yesterday it was freezing addresses against quantum risk, this month it was BIP-110, today it is the cap.

There is a human bias underneath all of this that is worth naming, because it explains why brilliant people keep arriving at the same place. Engineering is the discipline of improving systems, and when a parameter looks imperfect the trained reflex is to correct it. That reflex is healthy nearly everywhere else. Bitcoin behaves less like software to be optimized than like a constitution, which does not seek to be optimal but to hold an order stable over time. Changing something feels like control; leaving it alone feels like negligence. Bitcoin's resilience comes partly from its capacity to resist exactly that impulse in the people who care about it most. Meanwhile the market is elsewhere: Bitcoin traded in the mid to high $77,000s on Sunday, consolidating after a run to $79,500 on August 21. Our three horizon views are unchanged. A debate with no code, no proposal in flight and no signaling behind it is a governance story, not a forecast input.

This is general market commentary, not investment advice or a recommendation to buy or sell any asset.