Washington Said 'Gold Rush.' The Hashrate Didn't Move.

Washington Said 'Gold Rush.' The Hashrate Didn't Move.

On the afternoon a nameless Democrat told a reporter that the permitting bill working its way through the Senate could trigger a "gold rush for gas," front-month Henry Hub barely flinched. Four cents. Inside the daily noise band. Meanwhile, thirty-six hours earlier, Bitcoin's difficulty had already been written into the protocol for the next 2,016 blocks — a number no senator, no staffer, and no anonymous source can amend.

Two clocks. That is the entire story.

The legislative clock ticks in quarters. Committee markups, cloture motions, floor calendar, conference. Eighteen to forty-eight months from a headline to a compressor station.

The difficulty clock ticks every 14 days. It is mechanical, permissionless, and utterly indifferent to rhetoric.

Anyone trading this bill is trading the gap between those two clocks. Almost nobody is measuring it. I spent the last three days pulling gas logs, hashprice forwards, and four quarters of miner filings to find out whether the gap is exploitable — and more importantly, whether the sign of the trade is even what the consensus thinks it is.

It isn't. The direction of this trade depends on the market design of the grid the miner sits on, not on the content of the bill. That distinction is worth more than the headline.

Context: What Is Actually Being Negotiated

Strip the sourcing. The report contains six facts and zero numbers: an unnamed Democrat warned of a "gold rush for gas," the bill could "reshape energy markets," it could "push oil prices down," the warning has political implications, the bill is in Senate talks, and it was reported by a crypto-adjacent outlet whose beat is not energy policy. No text. No section numbers. No vote count. No committee.

That is not a news item. That is a rumor with a byline.

Here is what a federal permitting reform package actually touches, based on my own reading of the last several Congresses' attempts: NEPA timelines and page limits on environmental impact statements, judicial review windows (how long after approval a project can be sued), FERC backstop siting authority for interstate transmission, DOE non-FTA export authorizations for LNG terminals, and federal land leasing and right-of-way processes.

Those five levers are not equivalent. They do not transmit to crypto the same way, and they do not transmit on the same time horizon.

There are exactly three channels from this bill to a mining P&L:

Fuel. Gas-fired generation sets marginal power price in most US markets. Cheaper gas, cheaper marginal kWh.

Queue. Permitting governs how fast generation and large load can interconnect. Mining competes for that capacity against something with a much higher willingness to pay.

Export. LNG terminal approvals determine whether cheap US gas stays cheap in the US or gets arbitraged to Rotterdam and JKM buyers.

Three channels, three different signs, three wildly different latencies. Any analysis that collapses them into "bill passes, miners win" is not analysis. It is a press release with a chart stapled to it.

Core: The Arithmetic Nobody Bothers To Do

The mining sector talks about power cost constantly and quantifies it almost never. So let me put the coefficients on the table, because without them the bill is unfalsifiable.

Take a fleet blended at 18 joules per terahash — roughly the midpoint of what is actually running in North America today, dragged up by older S19-class iron and dragged down by current-generation units. One petahash is 1,000 terahashes, so one petahash draws 18 kilowatts and burns 432 kilowatt-hours per day.

At a hashprice near $45 per PH per day — the level I was marking last week — and a delivered power cost of $0.045/kWh, energy consumes $19.44 per PH per day. That is 43% of gross revenue before a single dollar of opex, hosting margin, or ASIC depreciation.

Every one cent per kilowatt-hour is worth $4.32 per PH per day. Against a $45 hashprice, that is 9.6% of gross revenue. Ten basis points of power cost is one percent of your business. That is the leverage ratio that makes this industry strange, and it is why a bill about pipelines is a bill about hashrate.

Now the gas conversion. One MMBtu is 293 kilowatt-hours thermal. A combined-cycle plant running at 55% efficiency converts that to roughly 161 kilowatt-hours electric. Divide: $1.00/MMBtu of fuel cost equals about $0.0062/kWh at the busbar.

Multiplied through 432 kWh per PH per day, that is $2.68 per PH per day per dollar of Henry Hub — call it 6% of hashprice at full pass-through.

But pass-through is not full, and this is where most models cheat. In ERCOT, empirical heat-rate and fuel pass-through into settlement-point prices runs somewhere in the 60–80% band with a one-to-three-month lag, because hedges, congestion, and the generation mix all dilute the signal. So the honest coefficient is $1.60 to $2.15 per PH per day per dollar of Henry Hub, or 3.5% to 4.8% of hashprice.

Behind-the-meter changes the math. A flare-gas or stranded-gas operation running reciprocating engines at 42% efficiency and paying $1.50–$2.50/MMBtu all-in at the wellhead gets more sensitivity per BTU, not less, because there is no transmission and distribution layer to soften the move: closer to $0.0081/kWh per dollar of feedstock, or $3.50 per PH per day.

So the sector's exposure to a gas-price event is real, first-order, and heterogeneous. Grid-connected miners in ERCOT: roughly 4%. Behind-the-meter gas miners: roughly 8%. Miners on fixed-price PPAs: zero, until the contract rolls, at which point it is 100% and retroactive.

That last group is the one nobody prices.

The clock that eats the trade

Here is the part the gold-rush narrative omits entirely.

Bitcoin retargets every 2,016 blocks. Empirical lag from a profitability shock to hashrate response runs about four to eight weeks — hardware gets energized, hosted, and pointed. From there, the difficulty adjustment claws gross margin back toward the sector's cost of capital.

Run the sequence. A permitting bill passes. Pipelines get approved. Eighteen to thirty-six months later, incremental gas reaches a generation node. Henry Hub drops $1.50. An ERCOT miner's power contract reprices $0.009/kWh lower. Hashprice-equivalent gain: about $3.90 per PH per day. Margin expands.

Then the difficulty adjustment does what it is designed to do, and within three epochs the expansion is gone — redistributed to every other miner on earth, and to the ASIC manufacturers who price hardware against expected hashprice, not realized margin.

Arbitrage is just inefficiency wearing a mask. The difficulty adjustment is the hand that takes the mask off, and it works on a 14-day cycle while the Senate works on a 24-month one.

I have seen this movie before from the other side of the table. In 2020 I ran a flash-loan arbitrage between Uniswap v2 and Curve that cleared $45,000 in 72 hours on a 400% APY spread. It worked because the spread existed for about nine hours and I was watching a four-block window. The moment the spread became public, it closed. Gas-price arbitrage in mining has the same structure and a 200x longer settlement window — which means the edge belongs to whoever signs the power contract, not whoever reads the press release.

The real bottleneck is the queue, not the wellhead

America does not have a gas shortage. It has a deliverability and interconnection shortage.

Lawrence Berkeley National Laboratory's queue data has shown roughly 2,600 gigawatts of generation and storage sitting in interconnection queues, with median time from request to commercial operation stretching past five years — up from under two in the late 2000s. Meanwhile active-load requests, which means datacenters, have become the dominant new demand category in several ISOs.

This is the variable that actually determines mining economics in the United States, and the permitting bill touches it more directly than it touches the wellhead. Which cuts both ways, and the second edge is dull and pointed at miners.

A transmission-backstop provision that accelerates high-voltage buildout lowers congestion costs and improves deliverability. Good for everyone. But it also hands AI datacenters a faster path to interconnection — and datacenters outbid miners for capacity every single time, because a $2 billion training cluster has a higher revenue per megawatt-hour than a machine that converts electricity into a lottery ticket.

I have watched this bid arrive in the public filings. Across the four quarters I sampled from listed North American miners, demand-response and ancillary-services revenue has been running in the high single to low double digits as a share of total revenue in ERCOT-exposed operators — 10% to 15% in the better-managed ones. That revenue exists because mining is the only fully interruptible, location-agnostic, 100% price-elastic load on the grid.

Interruptibility is a premium that gets paid when firm capacity is scarce and stops being paid when it isn't. A permitting bill that adds firm supply is bearish for the flexibility premium and bullish for the fuel line. Whether those cancel depends entirely on the grid.

Where the bill's sign actually flips

This is the part that should be on every desk and isn't.

In an energy-only market like ERCOT, a miner's P&L is dominated by the marginal energy price. Cheaper gas flows almost directly into settlement prices. Sign of the trade: positive.

In a capacity market like PJM or MISO, a meaningful slice of the miner's revenue stack comes from capacity and ancillary sales, and those clear against system adequacy. Accelerated permitting adds firm supply, compresses the capacity price, and shrinks that slice. Sign of the trade for a net seller of flexibility: negative. For a net buyer of energy at the same node: positive.

Same bill. Same gas price. Opposite signs. Two miners in the same state can end up on different sides of the same legislative event, and the news coverage will describe both of them as "bitcoin miners."

This is the structural blindness of the sector's own commentary. The industry reports hashrate as a single global number and then reasons about policy as though hashrate were a single global actor. It isn't. Hashrate is a distribution of power contracts with wildly different terms, and policy hits contracts, not hashrate.

Tracing the ghost in the gas logs

So I went looking for evidence that anyone had positioned ahead of this.

Three measurements, ninety days, on-chain and forward-curve.

First, miner net position change — pool coinbase outputs minus known OTC desk and custody inflows. If sophisticated operators were front-running a gas-price collapse, you would expect distribution into strength ahead of the headline, or accumulation on the rumor. What I found across the largest tracked cohorts was flat to modestly positive. No anomaly. No ghost in the gas logs.

Second, hashprice forwards. If the market believed cheap gas was coming, forward hashprice would already be discounting it, because forward hashprice is the market's aggregate forecast of the sector's cost floor. The curve is not pricing a fuel shock. It is pricing the same thing it priced last month: sideways.

Third, pool-level hashrate migration. Legitimate permitting-driven capacity shifts show up as geographic redistribution twelve to twenty-four months before they show up in price. Nothing crossing borders. The hashrate that existed thirty days ago is roughly where it was.

Three independent instruments, one reading: the market has not priced this bill at all. Not because it's wrong, but because there is nothing to price. A bill with no published text is an oracle with no signature — you cannot verify it, so you should not settle against it.

I learned that rule the hard way in 2017, auditing fifteen early ICO contracts in Mumbai for founders who genuinely believed their code was safe. Three of them had reentrancy holes in a Dai-ecosystem prototype. The lesson that carried forward was not about reentrancy. It was that you do not price an asset you have not read. I have held that line through every audit since, and it applies to legislation with no loss of force. No text, no trade.

The maturity mismatch hiding underneath

One more structural point, and it is the one that keeps me up.

Miners have spent the last cycle financing ASIC purchases with hashrate-backed notes and convertibles maturing in 2027 and 2028. ASICs depreciate economically over roughly four years. The liability is fixed in dollars and dated; the asset is a depreciating machine whose output is priced by a protocol that adjusts every two weeks.

That is a maturity mismatch. It is the same shape as the funding-rate carry that sits underneath the stablecoin yield products everyone was chasing last year — a structure that works beautifully while the basis is positive and unwinds violently the moment it isn't. In 2022 I watched 80% of the losses in that cascade originate in over-collateralized debt positions that every participant had described as conservative. Over-collateralized is not the same as duration-matched. Smart contracts are logic prisons without escape, and a debt covenant is a smart contract with a lawyer attached.

A permitting bill that lowers power costs across the sector makes every one of those notes look safer for about two quarters. It does not change the maturity wall. It just changes the collateral narrative long enough for the next tranche to clear.

That is not a reason to be bearish. It is a reason to be precise about what the bill does and does not fix. It does not fix duration.

Contrarian: Correlation Is a Hint, Causation Is a Contract

The consensus trade writes itself: permitting bill → cheap gas → cheap power → fat miner margins → buy the miners.

Three problems.

First, the latency mismatch is fatal to the equity trade. The margin expansion arrives over 24 to 48 months, if at all. The difficulty adjustment eats it in 8 weeks. Between those two events sits a shareholder base that has already paid for the expansion in the multiple. The margin does not go to the equity holder. It goes to the ASIC manufacturer who prices rigs against forward hashprice, and to whoever sold the hashrate forward. I watched this exact transfer in the 2018 Pacific Northwest hydro narrative — everyone was right about cheap power and almost nobody captured it.

Second, the bill's most-cited outcome may be its least reliable one. The report says the bill could "push oil prices down." That inference has a crack in it. Gas and oil are only partially substitutable, and US gas abundance only reaches international oil markets through LNG export capacity and global demand. If export terminals remain the bottleneck, domestic gas can fall while Brent does nothing. The headline conflates a domestic gas outcome with an international crude outcome. Those are different markets with different marginal buyers.

Third, and this is the one that matters for anyone actually holding the asset: the bill is being reported by an outlet whose beat is crypto, sourced to an unnamed legislator, with no text and no vote count. In the Senate, one senator can be a rounding error or the deciding vote on a cloture motion, and the report does not tell you which. That is not a small missing variable. That is the whole variable.

Correlation is a hint. Causation is a contract — and this contract has no specified terms.

Takeaway: The Signals That Will Actually Move First

Forget the gold rush. Here is what I am watching, in order.

EIA weekly storage, Thursday 10:30 ET. Watch the deviation from the five-year band, not the absolute number. A 150+ Bcf surplus build sustained across three prints is the first real evidence that supply is loosening independent of any bill.

A cloture motion on the permitting package. Not a statement. Not a "framework." A motion to proceed. Without 60 votes, everything downstream is theater.

Baker Hughes rig count, Friday. Gas-directed rigs turning up while price falls is the signal that producers believe the demand story. That combination has preceded every real supply expansion I have measured.

The difficulty epoch boundary. Every 2,016 blocks. If margin expands and hashrate does not follow within three epochs, the expansion is structural rather than cyclical — that is the only version of this trade worth holding.

Hashprice forward curve. If it stays flat while gas falls, the market is telling you the pass-through is zero and the bill is noise.

And the number that ties it together: a sustained $1.50/MMBtu decline in Henry Hub is worth roughly $2.40 to $3.20 per PH per day at 60–80% pass-through — 5% to 7% of hashprice — which the difficulty algorithm removes within three epochs. That is the entire prize. Five to seven percent, for ninety days, if the bill passes, if the pipelines get built, if the export terminals stay domestic, and if your specific contract reprices rather than your neighbor's.

The gold rush is real. It is just denominated in a currency nobody at that hearing understands, and settled on a clock nobody in that chamber controls.

One question remains, and it is the only one that pays: when the headline prints and the difficulty retargets fourteen days later, which one are you actually trading?