The most important number in Bitcoin right now is not the price. It is the share of miner revenue that arrives through transaction fees โ and that number has been bleeding out for eighteen months while the price chart tells a consolidation story that almost nobody wants to interrogate.
Here is the signal. Over the trailing ninety days, fee revenue has averaged somewhere between 1% and 2% of total block rewards across most weekly windows. During the April 2024 halving block โ block 840,000 โ that same metric briefly printed above 70% for a single block, which cleared roughly 37 BTC in fees alone, the largest single-block fee take in the network's history. The distance between those two numbers is the entire story of Bitcoin's next four years, and the market is not pricing it.
Meanwhile, in a different ledger entirely, Hong Kong's Securities and Futures Commission has been steadily issuing virtual asset trading platform licenses. The consensus read is that Asia's financial centre is finally embracing crypto. That read is wrong. I'll get to why. First, the fee market.
The Decay Function Nobody Models
Bitcoin's security model is a subsidy scheme with a decay function bolted on. Every 210,000 blocks the block subsidy halves. Fifty BTC became twenty-five in 2012, twelve and a half in 2016, six and a quarter in 2020, and 3.125 BTC in April 2024. Around 2028 it becomes 1.5625. Around 2140 it becomes zero.
The design assumption โ the one encoded in the whitepaper and repeated by every maximalist without examination โ is that transaction fees will gradually replace the subsidy as the network's security budget. Miners will keep hashing because fees will pay them. That is a hand-wave dressed as a mechanism. It has never been tested at scale, because until 2023 there was never enough fee demand to test it.
Then Taproot activated in November 2021, and with it came a witness discount. Data committed to the witness portion of a transaction is priced at roughly one-quarter the cost of data in legacy block space. That discount was designed to make signature aggregation cheaper. It also made it economically rational to write arbitrary data โ images, text, JSON โ into Bitcoin's blockspace at a seventy-five percent discount.
In January 2023, Casey Rodarmor exploited exactly that. Ordinals. Inscriptions. Within months, BRC-20 tokens โ most of them worthless โ were consuming the majority of block space during peak windows and pushing median fee rates from single-digit satoshis per vbyte into the hundreds.
I audited token sales during the 2017 ICO mania, and I can tell you what a fee spike looks like from the inside. It looks like a subsidy. It is a transfer of value from impatient users to the miners who sequence their blocks. In 2023 and 2024, Ordinals were the largest voluntary transfer of value to Bitcoin miners that has ever occurred outside the block subsidy itself.
And then it stopped. Not entirely โ inscriptions still mint, Runes still trade, fee spikes still happen during congestion events. But the baseline decayed. The inscription wave crested, the marginal speculator left, and what remained was a fee floor structurally lower than the one miners had been promised.
That is the context. Here is what the data actually says.
The Fee Share Is a Forward Curve, Not a Lagging Indicator
I've been building liquidity models since 2020, when I spent three months mapping Uniswap v2 and Compound depth against Ethereum gas spikes. The lesson from that exercise transferred directly. In any fee-based system, you cannot read the health of the system from aggregate volume. You have to read it from the shape of the marginal transaction. Bitcoin's aggregate on-chain volume looks fine. Its marginal transaction economics look increasingly like a network being paid for by inertia.
Decompose miner revenue. A miner's gross revenue per block equals the subsidy โ 3.125 BTC โ plus total fees. At a $100,000 BTC price, which I'm using for arithmetic and not as a forecast, the subsidy alone is $312,500 per block, or roughly $2.25 billion per year across approximately 52,560 blocks. Fees at 2% of that add about $45 million annually. Fees at 40% โ the level the inscription wave sustained during its peak months โ add roughly $900 million.
That is not a rounding error. That is the difference between a mining sector that can fund hashrate expansion and one that has to liquidate treasury to stay solvent.
Hashprice โ revenue per unit of hash โ is the number that actually clears the market, and it has been grinding toward all-time lows.
Hashrate kept climbing through the consolidation, which is the part that confuses people. If fees are collapsing, why is hashrate at record highs? Because hashrate is a lagging capital commitment. Miners buy ASICs eighteen months ahead of delivery, sign power contracts five years out, and amortise hardware over three. The hashrate you see today was underwritten by the fee environment of 2023 and early 2024, when Ordinals made every block a lottery ticket.
The marginal miner โ older-generation rigs, spot power, no hedge book โ sets the floor. And the marginal miner's unit economics right now are dictated almost entirely by the subsidy, because the fee component has become noise. When hashprice falls below the cash cost of the least efficient rig still plugged in, that rig unplugs. That is the feedback loop. It is mechanical, slow, and currently in motion. A security budget is a thermodynamic problem wearing a financial costume.
The Counterfactual the Industry Refuses to Run
Strip inscription and Runes fees out of the last three years, and Bitcoin's mining sector would have absorbed the 2024 halving with a revenue base roughly 15% to 25% lower than what it actually experienced. In an industry with high fixed costs and thin margins, a 20% revenue shortfall is the difference between consolidation and capitulation.
I ran a version of that model in late 2023, in a note that drew a lot of pushback. The bull case at the time was that inscriptions were a fad and would disappear. They faded. They did not disappear. And the residue is the reason the network's security budget survived the last halving without a hashrate collapse.
This is where I depart from the standard maximalist script. The script says Ordinals are spam, a nuisance, an attack on monetary purity, and the network would be better without them. I understand the aesthetic. I do not accept the economics. Without the inscription wave, Bitcoin's security model would already be in visible trouble โ not in 2140, not in 2036, but in the current cycle. The 2024 halving cut the subsidy in half at the moment when the fee market was the only growing revenue line. Ordinals were not a distraction from that reality. They were the bridge across it.
The Witness Discount Arbitrage
There is a technical detail underneath this that almost nobody discusses, and it explains why the inscription economy grew as large as it did. The witness discount means inscription data is effectively subsidised by every other user of block space. A transaction writing 400 kilobytes of image data into the witness pays roughly a quarter of what the same data would cost in legacy space. The subsidy is not paid by miners. It is paid by the users whose transactions get crowded out during fee spikes.
That is a structural subsidy for data, financed by monetary transaction demand. It is the cleanest example I know of a protocol-level design choice creating a second-order economic distortion that took eighteen months to become visible. It also explains the shape of the fee curve: spikes during inscription waves were not a signal of monetary demand. They were a signal of data demand crowding out monetary demand.
Which means the fee floor that remains after the wave is the honest number. It is what Bitcoin's blockspace is worth when nobody is arbitraging the witness discount.
What the Residue Actually Looks Like
The wave left sediment. Inscription activity did not die โ it normalised. What changed is who is doing it and why. The speculative mint-and-flip cohort is gone. What remains is a smaller, more persistent base of collectors, plus the Runes ecosystem that launched at the halving and has since settled into a low-volume trading regime.
The important metric is not inscription count. It is inscription share of block space during non-congested windows. During the peak, that share crossed 50% routinely. In the current consolidation it sits far lower, and the fee premium attached to it has compressed. That compression is the fee floor.
Here is why that matters for positioning. A fee floor is a valuation input. If you believe the floor is permanent, you are underwriting a security budget that depends on a structural subsidy from a small group of data consumers. If you believe the floor is cyclical, you are underwriting a security budget that depends on the next inscription wave arriving before the next halving. Both are bets. Neither is a certainty. The market currently prices neither.
When I mapped NFT trading volume in 2021 against money supply indicators, the conclusion was that digital collectibles were liquidity siphons from the broader ecosystem. Ordinals are the same mechanism running on a different ledger. The difference is that on Bitcoin, the siphon feeds the security budget directly. That is not a bug. That is the only reason the last halving did not break the mining sector.
The Second Ledger: Hong Kong's Routing Decision
Now the part of the map that actually gets drawn.
Hong Kong's licensing regime is not an innovation policy. It is a routing decision. The SFC's virtual asset trading platform regime came into force in June 2023, and the licensing cadence since has been deliberately slow โ a handful of approvals, a heavy compliance burden, strict custody requirements, and a retail access framework far more restrictive than what offshore venues offer. On its face this looks like caution. Read against Singapore's posture, and it looks like something else.
Singapore's Monetary Authority spent the same period tightening. Retail access to spot crypto was curtailed, advertising restrictions were imposed, and the Digital Payment Token regime made it expensive and slow to serve the retail segment. Hong Kong moved the other way on stablecoins โ a dedicated ordinance, an issuer licensing regime, and a clear signal that it intends to be the venue where regulated, bank-adjacent digital asset settlement happens in Asia.
The reason this is a routing story and not an adoption story is simple. The marginal institutional buyer of Bitcoin in Asia does not care about ideology. It cares about which jurisdiction lets it hold the asset with a clean audit trail, a bankable custody arrangement, and a regulator that will not retroactively change the rules. For five years the answer was Singapore. Hong Kong is now trying to make the answer Hong Kong.
That is the mechanism. It is a competitive land grab dressed as a licensing regime, and the target is the flow currently clearing through Singapore.
Fractures in the ledger reveal the truth of value. And the fracture here is not in the code. It is in the regulatory perimeter โ the place where the asset stops being an asset and becomes a compliance line item.
Why the Two Ledgers Are One Story
Connect them. Bitcoin's fee market and Asia's regulatory routing look like separate stories, and sell-side research treats them that way โ one is a protocol economics problem, the other a policy problem. They are the same problem. A security budget is funded by whoever is willing to pay for block space. A regulatory regime determines who is allowed to pay for it, through which venue, with which custody arrangement, at what tax treatment.
When Hong Kong licenses a platform, it does not just create a venue. It creates a class of buyers who can hold Bitcoin inside a regulated perimeter and therefore treat it as a portfolio asset rather than a speculative position. Those buyers are less fee-sensitive. They are also, critically, less likely to be the ones minting inscriptions. The regulatory perimeter and the fee market are pulling in opposite directions: institutionalisation deepens the holder base while stripping out the high-fee marginal transactions that fund the security budget.
Nobody is modelling that interaction. It is the most interesting open question in the asset class right now.
Positioning in Chop
Sideways markets are not the absence of information. They are the compression phase in which the next move gets funded. The signals worth reading right now are not price-based. They are structural: the smoothed fee share, the hashprice trend, the difficulty adjustment cadence, and the licensing cadence in Hong Kong. Those four lines will tell you more about the next twelve months than any chart pattern.
The Decoupling Thesis Is Backwards
Here is the consensus view, stated plainly. Ordinals were a speculative bubble that has now deflated, and Hong Kong's licensing push signals that Asia is finally embracing digital assets. Both halves of that sentence are wrong, and they are wrong in the same direction โ they mistake a structural shift for a sentiment shift.
Start with decoupling. The dominant narrative for eighteen months has been that crypto is decoupling from macro โ that Bitcoin trades on its own liquidity cycle now, that it has become an asset class with its own rate curve. I think that thesis is backwards in an important way. Crypto has not decoupled from macro. It has decoupled from retail macro. What it is now coupled to is the plumbing of regulated custody and settlement. When the marginal buyer is an institution holding through a licensed venue, the binding constraint is not the Fed funds rate. It is whether the venue exists, whether the custody is bankruptcy-remote, and whether the auditor will sign.
Since 2022 I have tracked the correlation between Treasury yields and DeFi total value locked, and it held tightly for two years โ a clean causal chain from rate hikes to stablecoin minting to TVL contraction. That correlation has loosened, and the reason is not that crypto escaped macro. It is that the marginal dollar stopped entering through DeFi and started entering through custodial venues with their own jurisdictional gatekeeping. The macro transmission channel changed. The sensitivity did not disappear.
That is why the Hong Kong licensing cadence matters more than any single macro print this quarter. It is the constraint that actually binds.
And the Ordinals point: the fade is real, but fade is not irrelevance. What the industry got wrong in 2023 was believing inscription demand was a permanent bid for block space. What it is getting wrong now is believing that because the bid is cyclical, it does not matter. A cyclical fee market is exactly what a security budget looks like when the subsidy decays. You do not get to choose whether your revenue is cyclical. You only get to choose whether you have enough of it.
There is a darker version of this argument, and I will state it because it is the honest one. If the fee market cannot sustain a security budget above roughly 5% to 10% of block rewards on a smoothed basis, the long-run equilibrium for Bitcoin is either a much smaller hashrate, a much higher price, or a change to the issuance schedule that nobody currently wants to discuss. Entropy is the only constant in liquid markets. The subsidy is entropy with a timer on it.
What I'm Watching
What I'm watching over the next ninety days is not the price. It is three numbers. The smoothed fee share of miner revenue โ whether it establishes a floor above 3% or continues decaying toward 1%. Hashprice โ whether the marginal miner capitulates, which shows up as a hashrate drawdown and a difficulty adjustment, the first real stress test of the post-halving regime. And the Hong Kong license count โ because every approval is a routing decision, and routing decisions compound.
The consolidation is not the story. The consolidation is the interval in which the story gets decided. Position accordingly โ or at least know what you are positioned in.