Taproot’s Fee Paradox: Who Pays Bitcoin’s Bills in 2026?

On the morning of September 25, 2026, sending a Bitcoin transaction into the next block cost about one satoshi per virtual byte. The mempool, the waiting room where unconfirmed transactions queue for space, sat close to empty. Fees added almost nothing to what miners collected on top of the block subsidy. And yet, by most measures, more was being built on top of Bitcoin than at any point in its history: dollar-denominated tokens moving over the Lightning Network, virtual UTXOs settling payments away from the chain, and a live zero-knowledge rollup posting proofs back to the base layer. Almost all of it traces back to a single upgrade.

Taproot, which activated at block 709,632 in November 2021, is the technical foundation for the application layer Bitcoin has grown over the five years since (CoinDesk). It is also, in 2026, at the center of a question the anniversary retrospectives tend to skip. The apps Taproot made possible are busy. The fee market that is supposed to pay Bitcoin’s miners, and to secure the entire system once the block subsidy fades toward nothing, is not. This is the story of that paradox: how the most consequential Bitcoin upgrade in a decade helped build a thriving app layer that pays the base chain almost no rent, and why that gap grows more serious with every halving.

The upgrade that made Bitcoin programmable

Taproot arrived as a bundle of three Bitcoin Improvement Proposals. BIP 340 introduced Schnorr signatures, a cleaner scheme whose signatures can be aggregated; BIP 341 defined the Taproot output type (the addresses that begin with bc1p) along with a structure called MAST that lets a coin commit to many possible spending conditions while revealing only the one actually used; and BIP 342 defined Tapscript, the scripting dialect for those spends (bitcoin/bips). The pitch at activation was modest: smaller, cheaper, more private transactions, and multisignature arrangements that look identical to ordinary single-key spends on the chain.

That understatement is why so much of the coverage since has called Taproot a quiet upgrade. Under the hood, though, Schnorr and MAST handed developers a set of primitives that turned out to be unusually powerful. Signature aggregation made complex custody cheap. Adaptor signatures, a trick Schnorr enables, made it possible to tie a payment to the revelation of a secret, which is the machinery behind discreet log contracts and the newer point time-locked contracts on Lightning. And because MAST lets a spender reveal a single script branch out of many, contracts that would once have bloated the chain became compact and private.

None of that was an accident, but what came next largely was. The same features that made custody elegant also made Bitcoin a far more capable place to store data and to define assets. Within about a year of activation, that capability collided with a speculative mania, and Bitcoin’s fee market has not looked the same since.

How Taproot accidentally started a fee boom

In January 2023, a developer named Casey Rodarmor released Ordinals, a scheme for numbering individual satoshis and attaching arbitrary data to them. The mechanism leaned directly on Taproot. Inscriptions tuck their data into the witness portion of a Taproot script-path spend, and because SegWit gives witness data a fee discount, Taproot made it economically viable to stuff images, text, and eventually entire token standards into Bitcoin blocks. Without the script-path flexibility and the witness structure that Taproot formalized, the inscription craze would have been far more expensive and far smaller.

The result was a surge in on-chain demand that Bitcoin had never seen from anything other than plain payments. BRC-20, a token experiment built on top of Ordinals inscriptions, sent thousands of transactions a day competing for block space. Taproot usage, which had been a rounding error, climbed from roughly 1% of transactions in early 2023 toward the top of its range within a year (Glassnode). For miners the timing could not have been better: a new, price-insensitive bidder for block space had arrived just as the community was bracing for the next halving to cut the subsidy in half.

It is worth being precise about what was Taproot and what was not, because the distinction matters for the fee story. Ordinals inscriptions and the BRC-20 tokens layered on them are Taproot-native. The Runes protocol that followed is not; it stores its data in a Bitcoin OP_RETURN output, a different and older mechanism. But the two belong to the same wave, and they peaked together on a single, remarkable day.

April 2024, the day fees beat the subsidy

On April 20, 2024, Bitcoin mined block 840,000 and executed its fourth halving, cutting the block subsidy from 6.25 BTC to 3.125 BTC. Rodarmor had deliberately scheduled the launch of Runes for that same block, framing it as a way to hand miners a fresh source of fee revenue exactly as their subsidy was cut. The market obliged in spectacular fashion. The average Bitcoin transaction fee hit $127.97 that day, more than seven times the previous day’s level and roughly double the prior all-time record, while total miner revenue reached a single-day record of $107.8 million (CoinDesk).

For the first time in Bitcoin’s history, transaction fees briefly overtook the block subsidy: fees reached about 75% of per-block miner revenue at the peak, an inversion of the usual arrangement in which newly issued coins pay for almost everything. Runes swallowed the mempool whole. As independent Bitcoin developer Jimmy Song put it to CoinDesk in the days after launch, “The Runes asset issuance has overridden almost every other use case at the moment.”

That day became the reference point for a hopeful thesis: that Bitcoin’s long-run security could be funded by demand for block space rather than by inflation, and that tokenization would be the demand. Taproot, in this telling, had not just made Bitcoin programmable; it had shown that a busy Bitcoin could pay its own way. The problem is that the boom did not last, and the recovery has not come.

The bust, from 90% of fees to under 2%

Runes captured more than 90% of Bitcoin network fees in the hours around its launch. One year later, that share had collapsed to under 2% (BlockEden). The pattern repeated across the inscription economy. The speculative frenzy that had bid up block space in 2023 and 2024 cooled, then faded, and the fee spikes went with it.

Taproot’s own usage tells the same story from a different angle. After peaking above 40% of transactions in early 2024 on the back of inscriptions and Runes, Taproot’s share of on-chain activity drifted down to roughly 15% to 20% by early 2026 (Glassnode). Some analysts read the decline as a sign of maturity, arguing that what remains is organic wallet and protocol adoption rather than mania. That is true, but it is also the point: organic adoption, so far, is quiet, and quiet does not pay much in fees.

Inscriptions did not die. Ordinals collections still traded tens of millions of dollars in secondary volume per month into 2026, and inscription activity kept adding a steady trickle of transactions. But a trickle is not a windfall. The block space that once commanded more than $100 per transaction now cleared for a cent or two, and the mempool that used to back up for days emptied out.

Bitcoin’s fee market, from boom to bust

The arc of the past five years is easiest to see in a single table. Each phase was defined less by Bitcoin the money than by whatever was using Bitcoin the block space at the time.

The empty mempool of 2026

By the second quarter of 2026, the numbers had become almost hard to believe for anyone who lived through the 2024 frenzy. In the week of April 16 to 23, 2026, transaction fees made up 0.59% of total Bitcoin miner revenue, less than a cent of every dollar miners earned. The network collected 18.31 BTC in fees across the entire week, worth roughly $1.43 million; the median fee rate opened at 1.00 satoshi per virtual byte and averaged 1.34 for the week, even as blocks ran 93.7% full (btc.network). Blocks were nearly full and fees were nearly nothing at the same time, because the transactions filling them were paying the floor.

That combination is the tell. When the report called the persistence of sub-1% fee contributions “one of Bitcoin’s most important open questions,” it was pointing at the same paradox from the miner’s side: plenty of activity, almost no willingness to pay for priority. A block that is full of one-satoshi transactions is a block that carries traffic but not revenue.

The price backdrop did not help the fee narrative, though it did help miners in dollar terms. Bitcoin traded around $85,686 on September 23, 2026, after a roughly 25% rally in August, still down about 10% on the year (Fortune). A higher coin price lifts the dollar value of both the subsidy and any fees, which is why a rally can paper over a weak fee market for a while; the market’s willingness to rally even after the Fed hiked to 4% kept mining revenue respectable in dollars. But price is not the same as fee demand, and it is fee demand that has to carry the load when the subsidy is gone.

Taproot’s second act pays even less

Here is where the paradox sharpens. The inscription era at least paid Bitcoin handsomely while it lasted, because inscriptions live on the base layer and compete for its block space. Taproot’s second act, the wave of infrastructure now being built on those same primitives, is designed to do the opposite. Its entire purpose is to move activity off the base layer, and off-chain activity does not pay on-chain fees.

Consider the Taproot-native app stack of 2026. Taproot Assets, the protocol from Lightning Labs that puts stablecoins such as USDt onto Bitcoin, issues and transfers those assets largely over Lightning, so many dollar-denominated payments can flow while the base chain sees only the occasional channel open or close. Ark packs many users’ balances into shared Taproot outputs and settles their payments as virtual UTXOs, deliberately minimizing how often anything touches layer one. Citrea, a zero-knowledge rollup that reached mainnet in early 2026, executes transactions off-chain and posts compact proofs to Bitcoin only periodically. Statechain and channel-factory designs follow the same logic. Even the Lightning Network’s own Taproot upgrade, which swaps hash-based contracts for point time-locked ones, is about making off-chain payments more private and efficient, not about generating base-layer traffic.

Every one of these is a Taproot success story. Every one of them also carries large amounts of economic value while paying the base chain only a thin, occasional settlement fee. That is not a flaw in the designs; it is the explicit goal of scaling. A payment network that forced every coffee onto the main chain would be a failure. But it does mean that the more successful Taproot’s second act becomes, the less its activity shows up in the mempool, and the less it contributes to the fee market that miners will eventually depend on. This is close to the opposite of the dynamic on Ethereum, where account-level programmability keeps much of the activity, and much of the fee revenue, on or near the settlement layer.

The security budget question

To see why any of this matters beyond a slow week for miners, follow the subsidy schedule. Bitcoin pays miners in two ways: newly issued coins (the block subsidy) and transaction fees. The subsidy halves roughly every four years and is currently 3.125 BTC per block. It will fall to about 1.5625 BTC at the next halving, expected in 2028, and it keeps halving until it approaches zero in the decades that follow. Satoshi Nakamoto’s design assumed that fees would rise to replace the subsidy as it shrank.

That replacement is what people mean by the security budget: the total value paid to miners, which is what makes attacking the chain expensive. If the dollar value of block rewards falls, so does the cost of assembling enough hashrate to threaten the network. Today the subsidy does almost all of the work, and fees, at well under 1% of revenue, do almost none. Miners have adapted to thin margins by chasing efficiency, as the race below 10 joules per terahash shows, and by treating hashrate expansion as a capital-allocation decision rather than a given, a shift we explored in our look at how hashrate growth became a boardroom call. Difficulty has kept climbing regardless, as miners switch machines back on whenever price and hashprice allow.

None of that changes the arithmetic underneath. For now, a strong subsidy and a decent coin price keep the security budget large. But each halving cuts the coin-denominated subsidy in half, and if fee demand does not grow to compensate, the network ends up relying on the Bitcoin price to double every four years, indefinitely, just to hold its security budget flat. That is the crux of the debate, and it is why an empty mempool in 2026 is not a trivia point but a warning light.

The math the halving forces

A simple illustration shows the pressure. The table below holds the Bitcoin price flat on purpose, not because that is realistic, but because it isolates what the subsidy schedule does on its own.

Hold the price flat and the halving mechanically cuts the dollar security budget roughly in half, from around $14 billion a year to around $7 billion. A flat price is an unrealistic assumption, and bulls will point out that Bitcoin has always risen enough across cycles to more than offset the cut. That is the historical record. It is not a law of nature, and it is not a plan; it is a hope that the price keeps outrunning the subsidy schedule forever. The fee market was supposed to be the plan.

Where the experts disagree

Whether an empty fee market is a crisis or a curiosity is one of the oldest live arguments in Bitcoin, and 2026 has given it fresh urgency.

The measured middle is best represented by macro analyst Lyn Alden, whose fee-based security modeling has become a standard reference. Alden frames the shift plainly: “Bitcoin will gradually shift from paying miners primarily through bitcoin block rewards to paying miners primarily through bitcoin transaction fees,” and she judges the transition to be, in her words, “middling in terms of risk potential” (lynalden.com). Her condition is adoption: Bitcoin needs a sizable and persistent fee market, which she argues becomes plausible only if the network settles enough high-value transactions to make block space genuinely scarce again, with smaller payments pushed to Lightning and other layers.

A more relaxed camp says the worry is misframed. Pierre Rochard, a longtime Bitcoin researcher and chief executive of The Bitcoin Bond Company, calls the security-budget argument a category error. “The block subsidy halvings do not threaten Bitcoin’s security because it was never the subsidy keeping the network secure in the first place,” he argued in May 2026, adding that “ultimately what gives Bitcoin transactions finality is the fees. It’s not the new issuance of Bitcoin” (TheStreet). In his view, demand for settlement will produce whatever fee market Bitcoin needs, when it needs it.

Both can be right about different time horizons, which is why the question refuses to resolve. As Blockworks framed it in surveying the debate, the honest answer to whether fees alone can keep Bitcoin secure is that nobody yet knows, because the network has never operated in a world where the subsidy is negligible (Blockworks). What 2026 adds is evidence, and the evidence so far is that Taproot’s most successful applications are not the ones refilling the mempool.

Why this is a Taproot story, not just a mining story

It would be easy to file all of this under mining economics and move on. That would miss what makes 2026 specific. The fee market’s weakness is not a random macro accident; it is downstream of choices baked into Taproot and the ecosystem it enabled.

Taproot shaped both sides of the ledger. On the demand side, it made the inscription era possible, and that era’s boom and bust is the single biggest reason the fee chart looks the way it does. On the side of alternatives, the same primitives (Schnorr aggregation, MAST, adaptor signatures) are exactly what make Ark, Taproot Assets, Citrea, and modern Lightning efficient enough to pull real economic activity off the base layer. The upgrade that could have anchored a durable on-chain fee market instead helped build the most compelling reasons to transact somewhere other than the base chain.

So the future of Bitcoin’s fee market is not only a question about price cycles or ASIC efficiency. It is a question about how Taproot’s app layer evolves: whether tokenization and settlement grow large enough that even occasional on-chain footprints add up to a real fee market, or whether Taproot succeeds so thoroughly at moving value off-chain that the base layer it depends on goes hungry. That is a design and adoption question, and it sits squarely in the domain Taproot opened.

What could refill the mempool

There are a handful of plausible ways the fee market comes back, and they are worth naming because the bull and bear cases both live here.

• Scale. Alden’s condition, enough high-value settlement that block space becomes scarce again, is the orthodox path. If Taproot Assets, Ark, and rollups like Citrea grow into serious rails, their channel opens, closes, rebalances, and proof postings could, in aggregate, produce steady base-layer demand even without another inscription mania. Off-chain success would eventually generate on-chain settlement; the question is how much, and when.
• New on-chain use cases. This is why the covenant debate matters beyond the engineering. Proposals to add richer spending conditions to Bitcoin script would enable vaults, better custody, and payment pools, some of which put more structured transactions back on the chain. None has activated, and Bitcoin’s inability to agree on any soft fork since Taproot is itself a running story. But if the toolset expands, so might the reasons to pay for block space.
• Regulated settlement. Stablecoins on Bitcoin are the clearest example, and they arrive tangled in policy. Taproot Assets issuers pushing dollar tokens onto Bitcoin have to reckon with a US framework still being assembled after the collapse of last year’s comprehensive bill, a saga we tracked as Washington’s crypto deadlines reset. The SEC regulates the assets and the intermediaries that issue and trade them; it does not, and cannot, set Bitcoin’s fees or its consensus rules.
• Or none of the above. The fee market may simply never fully arrive, with Bitcoin muddling through on price appreciation and a shrinking but still positive subsidy for another decade or two before the question truly bites. That is the path of least resistance, and it may be the one Bitcoin takes. It is also the one that makes analysts nervous, because it defers a structural question rather than answering it.

What it means for miners and holders

For miners, the near-term takeaway is the one they already live: in 2026, fees are a rounding error, and the business is a bet on coin price, subsidy, and machine efficiency. That keeps the pressure on hardware and power costs and makes hashrate an increasingly financial decision. For as long as the subsidy dominates, an empty mempool is a manageable inconvenience, not an emergency.

For holders, the signal is longer-dated and easy to misread. A quiet, cheap Bitcoin chain is genuinely good news for anyone who wants to move coins today, and it is a sign that scaling is working as designed. The catch is that the same quiet, if it persists across the next two or three halvings without a fee market forming underneath it, is exactly what the security-budget skeptics warn about. It is worth watching the fee share of miner revenue, not just the price, as a slow-moving health metric for the network.

None of this is something a regulator will fix or break. The SEC can shape which Taproot-issued assets reach US investors and on what terms, but Bitcoin’s fee market is set by supply and demand for block space, and its consensus rules are changed, if at all, by the messy social process that has not managed to ship a soft fork since Taproot itself. Five years on, the upgrade did its job. Whether the economy it enabled will one day pay Bitcoin’s bills is still, genuinely, an open question.

Frequently Asked Questions

Does Taproot generate transaction fees for Bitcoin miners?

Indirectly, and in 2026 not much. Taproot itself is just an output type; it does not create fees. The applications built on it do. Ordinals inscriptions, which are Taproot-native, drove record fees in 2023 and 2024, but that demand has since collapsed. The newer Taproot apps, such as Taproot Assets, Ark, and rollups, deliberately move activity off the base layer, so they generate very little on-chain fee revenue even when they carry large volumes.

Why are Bitcoin transaction fees so low in 2026?

Because demand for base-layer block space has fallen sharply. The inscription and Runes frenzy that bid up fees in 2023 and 2024 cooled, and the new wave of Taproot-based scaling keeps most activity off-chain. In the week of April 16 to 23, 2026, fees were about 0.59% of miner revenue and the median fee rate sat near 1 satoshi per virtual byte, even though blocks were more than 90% full with transactions paying the minimum.

What is Bitcoin’s security budget problem?

Bitcoin pays miners with a block subsidy plus fees. The subsidy halves about every four years and is heading toward zero, so fees are supposed to grow to replace it. If fee demand stays low, the total paid to miners depends almost entirely on the coin price rising enough to offset each halving. If that does not happen, the cost of attacking the network could fall over time. That long-run funding gap is the security budget problem.

Did Bitcoin fees ever exceed the block subsidy?

Yes, briefly. On April 20, 2024, the day of the fourth halving and the launch of the Runes protocol, the average transaction fee hit $127.97 and fees reached about 75% of per-block miner revenue, overtaking the subsidy for the first time. It was a short-lived spike rather than a permanent shift, and the fee share fell back to well under 1% within two years.

Will low fees make Bitcoin insecure?

Not today. In 2026 the block subsidy still dominates miner revenue and secures the network. The concern is structural and long-term: across future halvings, a persistently empty fee market would leave security increasingly dependent on price appreciation. Analysts disagree on how serious this is, with views ranging from a middling, manageable risk to the argument that demand will always produce whatever fee market Bitcoin needs.

By the HOGE Wire markets desk, covering Bitcoin infrastructure, mining, and the fee market.