Bitcoin Block Space Economics: How Inscriptions, Runes, and L2s Compete for Scarce Bytes

Every ten minutes, Bitcoin produces a block capped at 4 million weight units. That is all the throughput the network offers: roughly 2,800 to 4,000 transactions per block, depending on transaction composition. Bitcoin block space is a finite commodity, and since the arrival of Ordinals inscriptions in January 2023, the set of buyers competing for that commodity has expanded dramatically. Payments, consolidation transactions, inscription mints, Runes token transfers, Layer 2 anchor transactions, and arbitrary data embeds now all bid against each other in a single fee market.

Understanding how this market works is essential for anyone building on Bitcoin. The fee environment determines whether a payment, a channel open, or a token mint is economically viable at any given moment. This article breaks down the demand sources, the pricing mechanics, and the economic consequences of competition for scarce block space.

Block Space as a Scarce Commodity

Bitcoin's block weight limit of 4 million weight units (4 MWU) is a hard consensus rule. Unlike Ethereum's adjustable gas limit or Solana's parallelized execution, Bitcoin offers a fixed supply of block space per unit of time. Miners cannot produce more of it. The only variable is the price: transaction fees, denominated in satoshis per virtual byte (sat/vB), which rise and fall based on demand.

This makes block space function like a commodity market. When demand exceeds supply, the mempool swells with unconfirmed transactions, and fee rates climb as users outbid each other for inclusion in the next block. When demand is low, fees drop to the floor (typically 1 sat/vB). The price of block space is pure market clearing.

What 4 million weight units actually means

Before SegWit, Bitcoin had a 1 MB block size limit. SegWit replaced this with a weight-based system where non-witness data (inputs, outputs, transaction headers) costs 4 weight units per byte, while witness data (signatures, scripts) costs only 1 weight unit per byte. This 4:1 discount was designed to correct an economic imbalance: spending a UTXO required more data than creating one, which discouraged consolidation and bloated the UTXO set.

In practice, a block full of standard SegWit payment transactions holds roughly 1.7 to 2.1 MB of raw data. A block stuffed with witness-heavy data (like Taproot inscriptions) can approach the theoretical 4 MB maximum. Before inscriptions, average block sizes hovered around 1.2 to 1.4 MB. Since the inscription wave, average block sizes have grown to approximately 1.7 to 2.5 MB: a 40 to 50 percent increase driven almost entirely by witness data.

The SegWit Discount and Its Unintended Beneficiaries

The SegWit witness discount was introduced to incentivize spending UTXOs and to enable the Lightning Network by fixing transaction malleability. It was not designed for embedding arbitrary data. But the 75% discount on witness bytes created an economic incentive that the Ordinals protocol exploited: storing images, text, and other media in the witness section of Taproot transactions at one-quarter the effective cost of equivalent non-witness data.

The discount math: A 100 KB image stored in witness data costs 100,000 weight units. The same data in non-witness space would cost 400,000 weight units. At identical fee rates, inscriptions pay 75% less per byte than traditional transaction data for on-chain inclusion.

This asymmetry means inscription users and payment users are not competing on equal footing. Inscriptions consume large amounts of block weight at a steep discount, while payment transactions pay full price for their non-witness data (addresses, amounts, change outputs). A single large inscription can fill half a block's weight capacity while paying fees as if it were a fraction of that size.

The Demand Sources Competing for Block Space

Before 2023, demand for block space came from a relatively stable set of use cases: payments, exchange withdrawals, UTXO consolidation, and multisig operations. The introduction of Ordinals, BRC-20 tokens, and Runes added entirely new demand categories that behave differently from traditional payments.

Traditional payments and transfers

Standard Bitcoin payments remain the baseline demand source. These include peer-to-peer transfers, exchange deposits and withdrawals, merchant payments, and batched transactions from large custodians. Payment transactions are typically fee-sensitive: users have a maximum willingness to pay, and when fees spike above that threshold, they delay or switch to alternative rails.

Ordinals inscriptions

Launched in January 2023 by Casey Rodarmor, the Ordinals protocol allows arbitrary data to be inscribed into Taproot witness space. By mid- 2026, over 117 million inscriptions have been created on Bitcoin, with the majority being text-based BRC-20 token operations rather than images. Only about 3 million of total inscriptions are image-based: the rest are token mints and transfers.

The December 2023 inscription wave demonstrated the protocol's impact. Average transaction fees surged above $37 as users created over 1.2 million inscriptions in a single weekend. The mempool backlog swelled past 300,000 unconfirmed transactions, and fee rates that had averaged $1 to $2 in September climbed to levels not seen since April 2021.

Runes token protocol

Runes, launched by Casey Rodarmor on April 20, 2024 (the same block as the fourth Bitcoin halving), uses OP_RETURN outputs for token operations rather than witness data. On launch day, Runes transactions consumed over 80% of all Bitcoin block space, and the average transaction fee spiked from $3.35 to $91.89: a 2,645% increase. Some blocks during this period contained fee revenue exceeding 10 BTC, far above the 3.125 BTC block subsidy.

The frenzy was short-lived. By May 2024, Runes accounted for roughly 12% of transactions. By July, the share fell below 9%. As of early 2026, Runes account for less than 2% of daily Bitcoin transactions and generate under $250,000 per day in fees. The trajectory illustrates how speculative demand for block space can spike violently and then subside, leaving behind a permanent but modest baseline of activity.

L2 settlement and anchor transactions

Layer 2 protocols require on-chain transactions for settlement, channel management, and fraud proofs. Lightning Network channel opens and closes are standard multisig transactions that add only 12 to 25 extra bytes compared to regular payments. But the aggregate impact grows with network size: thousands of channels opening and closing during fee spikes create additional congestion.

Other L2 architectures have different footprints. The Liquid Network uses federated peg-in and peg-out transactions. Ark batches many off-chain transfers into periodic on-chain settlement rounds. Spark, built on statechains, minimizes its L1 footprint by only touching the blockchain for initial deposits and unilateral exits: transfers between users happen entirely off-chain through FROST key rotation.

OP_RETURN data embedding

The OP_RETURN opcode allows embedding small amounts of arbitrary data in provably unspendable outputs. Until 2025, Bitcoin Core enforced a default relay policy limiting OP_RETURN data to 80 bytes per output. In June 2025, Bitcoin Core merged a change removing this limit in version 30, scheduled for release in October 2025. The change does not alter consensus rules but relaxes the default relay policy, allowing OP_RETURN outputs up to the full block weight limit.

Proponents argue this reduces UTXO set bloat by encouraging data embedding in provably unspendable outputs rather than in fake UTXOs. Critics worry it will attract more non-payment data to the blockchain. Regardless of perspective, the change reflects an ongoing tension between preserving block space for payments and accepting that data embedding is already happening through less efficient mechanisms.

Historical Fee Spikes and What Caused Them

Bitcoin's fee market has experienced several dramatic episodes that reveal how block space scarcity manifests under different demand conditions.

A consistent pattern emerges: fee spikes driven by new protocol activity (inscriptions, Runes) are sharp but brief. The December 2017 spike lasted weeks because the underlying demand (ICO-era speculation) was sustained. Post-inscription spikes resolve in days because the speculative activity that drives them burns out quickly. By contrast, the 2025-2026 baseline of $0.60 to $2.00 average fees reflects a period of relatively low demand where the mempool frequently clears at 1 sat/vB.

Who gets priced out: During the April 2024 Runes launch, a simple payment transaction cost over $90. At that rate, any Bitcoin transfer under roughly $500 becomes economically irrational when fees consume more than 15-20% of the value. Small-balance holders, Lightning channel operators needing to force-close, and users in emerging markets are the first to be excluded from L1 during fee spikes.

The Mempool as a Fee Auction

The mempool is where unconfirmed transactions wait for block inclusion. Each full node maintains its own mempool (Bitcoin Core defaults to a 300 MB cap), and miners select transactions that maximize their fee revenue. This creates a continuous auction where higher-fee transactions displace lower-fee ones.

As of mid-2026, the mempool typically holds around 40,000 to 50,000 unconfirmed transactions totaling approximately 150 to 200 MB. During quiet periods, all pending transactions clear within one or two blocks. During spikes, the backlog can grow to hundreds of thousands of transactions and take hours or days to clear.

Cluster mempool: a structural upgrade

In November 2025, Bitcoin Core merged the cluster mempool redesign, developed by Suhas Daftuar and Pieter Wuille. The new architecture groups related unconfirmed transactions into clusters (capped at 64 transactions and 101 kvB each) and subdivides them into chunks sorted by fee rate. This enables more accurate fee estimation, better eviction of low-fee transactions, and more efficient block template construction.

Cluster mempool does not change the supply of block space, but it improves how efficiently that space is allocated. Miners using the upgraded software can extract more fee revenue from the same 4 MWU limit, and users get better predictions of what fee rate will achieve confirmation within a target number of blocks.

Economic Implications of Block Space Competition

The expansion of demand sources for block space has several structural consequences for the Bitcoin ecosystem.

Fee floor stabilization

Inscription and Runes activity has established a higher baseline for block utilization. Even with Runes below 2% of transactions, average block sizes remain elevated compared to pre-2023 levels. Daily transaction counts regularly exceed 800,000, with average transactions per block near all-time highs. This sustained utilization creates a more stable fee floor, which benefits miners as the block subsidy continues to decrease with each halving.

User stratification

Rising block space demand creates economic stratification among Bitcoin users. During fee spikes, the network effectively becomes tiered:

• High-value transfers ($10,000+) remain viable at any fee level
• Medium-value transfers ($100-$10,000) become expensive but feasible
• Small transfers (under $100) are priced out of L1 entirely
• Dust UTXOs become economically unspendable regardless of fee conditions

This is not a temporary condition. As mining economics increasingly depend on fees rather than subsidies, sustained high fees become desirable from a network security perspective. The 2028 halving will reduce the block subsidy to 1.5625 BTC, further increasing the economic importance of fee revenue.

The data-versus-payments debate

The community remains divided on whether non-payment uses of block space are legitimate. One perspective holds that block space should serve financial transactions: payments, settlements, and value transfer. The opposing view argues that miners are incentivized to include whatever transactions pay the highest fees, regardless of purpose, and that attempting to distinguish “legitimate” from “illegitimate” transactions undermines Bitcoin's censorship resistance.

In practice, the protocol makes no distinction. A miner including a 400 KB JPEG inscription at 5 sat/vB earns more from that single transaction than from dozens of standard payments at the same rate. The SegWit discount amplifies this dynamic: witness-heavy transactions deliver more revenue per weight unit to miners while consuming a larger share of the raw data capacity.

Why Block Space Scarcity Strengthens the Case for L2

Every fee spike reinforces the same lesson: routine payments cannot sustainably compete for L1 block space against well-funded inscription minters, token speculators, and high-value settlement transactions. The long-term trajectory of Bitcoin is toward a settlement layer where L1 transactions are reserved for high-value or security-critical operations, while everyday payments move to Layer 2 solutions.

Different L2 approaches to block space efficiency

Not all Layer 2 protocols consume block space equally. The on-chain footprint varies significantly by architecture:

• Lightning Network requires one on-chain transaction to open a channel and one to close it, with all intermediate payments handled off-chain. During high-fee periods, opening and closing channels becomes expensive, and force-closes triggered by unresponsive peers compound the problem
• Liquid Network uses federated peg-in transactions (confirmed on-chain) and peg-out transactions signed by federation members, with all internal Liquid transactions on a separate sidechain
• Ark batches multiple transfers into periodic on-chain settlement rounds, amortizing the L1 cost across many users
• Spark minimizes L1 interaction by using statechain key rotation for transfers. Users only need an on-chain transaction to deposit into Spark and to unilaterally exit. Transfers between Spark users are pure off-chain key operations that consume zero block space

The Spark advantage during fee spikes

Spark's statechain architecture is particularly well-suited for environments where block space is expensive. Because transfers involve FROST threshold signature key rotation rather than on-chain transactions, the marginal cost of a Spark transfer is effectively zero regardless of L1 fee conditions. A user sending Bitcoin or USDB on Spark pays the same near-zero fee whether L1 fees are at 1 sat/vB or 1,000 sat/vB.

This decoupling from L1 fees means Spark users are not affected by inscription waves, Runes mania, or any other source of block space competition. The only time L1 fees matter to a Spark user is during deposit (moving Bitcoin on-chain into Spark) and during unilateral exit (moving Bitcoin from Spark back to L1). For users who primarily transact within the Spark ecosystem, these are infrequent events.

The Road Ahead: Block Space Demand in a Post-Subsidy World

Looking forward, several trends will shape the economics of Bitcoin block space:

• Declining block subsidies mean fees must grow to maintain miner incentives and network security. The current 3.125 BTC subsidy drops to 1.5625 BTC in 2028 and 0.78125 BTC in 2032
• New demand sources will continue to emerge. The OP_RETURN limit removal in Bitcoin Core 30 may enable new data-embedding protocols. Covenant proposals like CTV and OP_CAT could introduce on-chain computation patterns that consume block space differently
• L2 maturation will shift routine activity off-chain, reducing payment demand on L1 while increasing the share of block space consumed by settlement, anchor, and dispute resolution transactions
• The fee market will become increasingly competitive, favoring users and protocols that can batch, defer, or avoid L1 transactions entirely

Bitcoin's fixed block space supply is both its greatest constraint and its most important feature. The scarcity that makes L1 transactions expensive during demand spikes is the same scarcity that gives Bitcoin its security guarantees and censorship resistance. The protocols that thrive in this environment will be those designed from the ground up to minimize their on-chain footprint.

For developers building on Bitcoin, the Spark SDK provides a way to offer users instant, fee-insensitive payments without competing for block space. For users looking to transact today, General Bread is a Spark-powered wallet that demonstrates what payments look like when they are decoupled from L1 congestion. And for a deeper look at how fee dynamics affect the mempool, see our mempool congestion economics explainer.

This article is for educational purposes only. It does not constitute financial or investment advice. Bitcoin and Layer 2 protocols involve technical and financial risk. Always do your own research and understand the tradeoffs before using any protocol.