The Mining Subsidy Without Usage: Bitcoin's Uncomfortable Math

The mining subsidy without usage is a problematic counterfactual that people just gloss over. The block subsidy only works as security if bitcoin is actually being used. No real transactions, no real security. The whole thing eats itself.

Sztorc nailed this in his 2019 piece on the security budget¹: the security budget is just what we pay miners, and when it’s low, 51% attacks get cheap. Back in 2018 you could have suppressed BTC with nonstop 51% attacks for about $2.6 billion a year². The US military burns through that in a couple days³.

The Subsidy Has Carried Us. It Won’t Forever.

The subsidy has carried us because price went up faster than the halvings cut it down. The math is simple: as long as BTC appreciates more than roughly 19% a year, it compensates for the halving¹. And historically it’s been way more than that, like 70%+. But come on, price doesn’t go up forever. Even in the most absurdly bullish scenario there’s a ceiling. Bitcoin can’t outgrow the nominal value of all the world’s money⁴. Once price flattens out, the subsidy bleeds toward zero, and fees have to carry the weight. That’s just math.

Users Won’t Pay High Fees When They Have Options

And here’s where the “store of value” crowd doesn’t want to look: users won’t pay high fees when they can just use something else. Altcoin blockspace is a perfectly fine substitute¹. People just want to send twenty bucks, they don’t care if it rides on LTC or DOGE or whatever.

You can see it in the data. BTC crossing the $1/tx mark in mid-2017 lines up exactly with the altcoin explosion⁵, and the fee spike of late 2017 reversed fast, with BTC transactions actually declining in a way we hadn’t seen before. People paid those fees because they were panicking and FOMOing in, not because they thought $28 per transaction was reasonable⁶. Once the panic passed, fees collapsed to nothing. Total fees for the 12 months before early 2019 were around $70 million⁷. That’s a joke for a network supposedly securing hundreds of billions in value.

Mining at a Loss Becomes a Tax on Holding

Now put yourself in the shoes of someone holding millions in BTC. You need the network healthy, your whole stack depends on it. But miners aren’t running a charity. They’re burning electricity, buying hardware, paying rent. The subsidy keeps getting cut in half and nobody’s transacting on-chain to generate fees, so miners start going underwater. Some turn off their rigs. Hashrate drops. Security gets thinner.

And now here you are, a rich bitcoiner sitting on a network that’s getting weaker by the month. If you want it to survive, somebody has to keep mining at a loss. That’s not “storing value” anymore. That’s paying a tax just to hold your own money. All the diamond hands laser eye nonsense falls apart real fast when holding itself costs you money. Nobody posts memes about slowly going broke.

The Difficulty Adjustment Buys Time, Not Solutions

Yeah the difficulty adjustment helps. Every 2,016 blocks it recalibrates⁸, lets weaker miners survive when stronger ones leave. It buys time. But it doesn’t fix anything. Drop the difficulty far enough and suddenly a 51% attack isn’t this theoretical nation-state thing anymore. It’s something a motivated billionaire or mid-size company could pull off. The adjustment smooths the ride down, it doesn’t prevent the destination.

The Death Spiral Nobody Wants to Talk About

The whole thing is a balancing act that only works if people actually use bitcoin. The halving schedule is hardcoded, it doesn’t care about market conditions. Fees have to step in or the security budget just keeps shrinking. And if it shrinks enough you get the spiral: hashrate drops, security weakens, confidence breaks, people sell, price falls, mining gets even worse, repeat.

Bitcoin without actual usage isn’t digital gold sitting safely in a vault. It’s digital gold sitting in a vault where the guards keep quitting because nobody’s paying them. The “hold forever” thing only works if the network is worth defending, and defense costs real money that only comes from real use.

A Counter Worth Taking Seriously

There’s a common pushback to all of the above that deserves direct engagement, because it points at a real flaw in the framing.

The pushback runs like this: hashrate dropping 20% in a year isn’t security dropping 20%. Miners don’t produce security — they produce blocks. Security is a function of cost-to-attack relative to value-at-stake, and that relationship is way less linear than “more hash = more secure.”

When hash drops, what’s actually happening is the market culling its weakest producers. Miners with expensive electricity get priced out. Miners running older-generation rigs get pushed below break-even and shut off. That’s the system working, not failing. The marginal cost of producing a hash converges back toward marginal revenue. Nothing about that math says “the network is dying” — it says “inefficient capital is exiting.”

And hashrate is cyclical. Every 2–3 years a new ASIC generation lands, miners with the new gear come back online aggressively, and total hash blows past the previous peak. The pattern over Bitcoin’s life isn’t a smooth line down — it’s a sawtooth, with hash falling during bad markets and surging on the next equipment cycle. Treating any one downturn as terminal mistakes the trough of a cycle for a death spiral.

So the honest version of the security-budget argument isn’t “subsidy + halving = doom.” It’s narrower: if prolonged low usage drives fees toward zero, and price stops appreciating, and no new ASIC generation pulls efficient hash back online — only then does the budget enter a regime where cost-to-attack falls below what wealthy adversaries can spend. Each of those ifs is real. None of them is automatic.

The version of this argument worth defending is conditional, not deterministic. The death spiral exists as a possible outcome, not a guaranteed one. The asymmetric bet still favors usage-driven security — but the absence of usage doesn’t itself prove the asymmetric bet has failed.

Thanks to Marcel Pechman for the feedback that prompted this counter section.

References