The Polysilicon Price Floor Won't Hold. That's the Trade.

The Polysilicon Price Floor Won't Hold. That's the Trade. Hook Every serious crypto analyst knows the marginal cost of a Bitcoin is the price of electricity. Most of them treat that as an abstract idea - grid rates, curtailment, hydro season in Sichuan. They miss that the electricity market itself is a manufactured set of price controls on a physical feedstock, and once in a generation, a government attempts to re-price that feedstock through political decree. We are watching that attempt happen right now, and the crypto market has not priced it. Crypto Briefing, a blockchain outlet, reported what is effectively an energy policy story: the incoming US administration is planning a price floor and tariffs on imported polysilicon, the ultra-pure silicon that is the base ingredient for both solar panels and semiconductor chips. The article was thin - three information points, at most. A normal trader reads this and moves on. I read it as a structural event for the mining industry, because polysilicon is not a solar story. Polysilicon is a cost-curve story for the entire digital-asset industry. China controls over 90% of the world's capacity to make solar-grade silicon. Global spot prices collapsed from roughly USD 40 per kilogram in 2022 to below 5 per kilogram in 2024 - below the cash cost of most producers in every market. Now Washington announces a floor. A politician's price floor with no budget and no buyer of last resort is what a trader would call a support line that has never been tested. And we all know what untested supports do when the tape runs through them. They become resistance. Context The Material: What Polysilicon Actually Is Polysilicon is a bridge material. It is purified silicon that is melted, crystallized, and sliced into wafers. The same feedstock family that produces solar wafers also produces the semiconductor wafers that sit inside TSMC's fabs, inside Nvidia's GPUs, and inside the ASIC controllers of every Antminer and every Whatsminer that secures the Bitcoin network. The word "silicon" in Silicon Valley was not chosen randomly. The chemistry matters because the production process is a furnace game. The dominant path is the modified Siemens reactor process, in which trichlorosilane gas is deposited onto heated silicon rods under high temperatures. More than 90% of global polysilicon is made this way, and the process is brutally energy-intensive: 40 to 60 kilowatt-hours per kilogram of silicon. There is also a newer fluidized-bed reactor process (FBR) that produces granular silicon, which uses roughly 30% less energy and has reached meaningful production scale in China through GCL Technology. The purity gap between granular and Siemens rods is a contested subject in N-type solar cell manufacturing, which is precisely the kind of hair-splitting that matters at the margins of a commodity market. Here is the cost map. Chinese tier-1 producers, including Tongwei, GCL, and Daqo, run fully integrated, at scale, with low energy costs in Inner Mongolia or Xinjiang; their all-in cash cost sits below USD 5 per kilogram at current exchange rates. Western producers - witness Wacker in Germany, Hemlock in the United States - historically built their capacity for semiconductor-grade silicon, at higher energy and environmental compliance cost, and they carry a production cost benchmark 20-50% higher than Chinese solar-grade silicon. The United States has perhaps 30,000 to 50,000 tons per year of effective domestic polysilicon capacity, and the bulk of it is not oriented toward feeding cheap solar modules. Domestic demand for solar and semiconductor silicon runs at 100,000 to 150,000 tons a year. The arithmetic is not complicated. The US consumes what it cannot make, by a factor of roughly three to one, and the gap is made up by imports from China, Korea, Germany, and now increasingly Southeast Asia. The Policy Gadget: A Floor With a Toll Booth The political instrument being proposed is the US market's version of an order book floor. Set a minimum import price at, say, USD 8 to 10 per kilogram - a level well above the USD 4-5 world price - and stack tariffs on top of it. The policy effect would be immediate and mechanical. American downstream module assemblers would pay 15-25% more per watt of module cost. American installers would pass it through to project owners. American ratepayers would absorb a mandatory tax on sunlight. And local miners who signed power-purchase agreements with new solar farms would watch the counterparty's capital cost curve rise. But the more consequential layer is the one hidden in the phrase "chip supply chains." The nominal target is solar-grade polysilicon. The strategic target is semiconductor-grade polysilicon. That is the base material for ASIC flow, and the US does not dominate it. The provision that would make a US domestic silicon champion competitive is not consumer-panel protection; it is control over the first link in the semiconductor chain. This is no longer a trade dispute. It is an industrial sovereignty program, and its internal logic mirrors the one that created the 1974 US-Japan semiconductor agreement, which was a minimum export price in all but name. That model failed in the end, because political floors do not clear markets. They create arbitrage. This is also where the policy collides with the current market cycle. Global polysilicon capacity is grotesquely oversupplied. China has well over 2 million tons of nameplate capacity online or under construction, while global solar and semiconductor demand is below 1.5 million tons a year. Operating rates in the Chinese silicon industry dipped below 60%. A US price floor does not fix any of this. It simply carves out a high-priced island in an ocean of cheap supply. And an island with a high price is exactly what arbitrageurs are looking for. Core Floors as Code: Auditing the Mechanism I came to trading through auditing, not narratives. In 2020, while still at UT Austin, I spent twelve hours manually walking through the Uniswap V2 factory contract, line by line, and spotted an integer overflow in the liquidity-token minting path that automated scanners had missed. Reporting it earned me USD 2,000 and a permanent habit: I read the settlement rules before I read the press release. The floor-price policy deserves the same audit. A financial security has a floor when there is an entity that is either economically required or politically compelled to absorb sell orders at that level. The US proposal contains no such entity. There is no Government Silicon Corporation, no buyer of last resort, no line item in the federal budget for taking delivery of 100,000 tons of silicon at USD 10 per kilogram. The mechanism on the table is a collection fee at the border, plus a minimum invoice price. Code doesn't care whether your floor is political or financial - it executes the settlement rules you wrote. The settlement rule here is: if you invoice below the floor, you receive a penalty instead of money. That is not a floor. That is a tariff, wearing a floor's clothing. The distinction matters because market participants will behave differently. A real floor is backed by a buyer of last resort. Consider the US housing market after 2008: the GSEs and the Federal Reserve effectively became the buyer of last resort, and that is what ultimately put a base under asset prices. Consider the Japanese yen's interventions - actual purchases of the currency by the Ministry of Finance. Those are floors. The polysilicon proposal is a toll booth. And toll booths do not stop flows; they tax them. I audit the logic, not the hope. The logic here has a missing branch: who absorbs the surplus when supply exceeds the floor? No one. That is exactly the flaw I look for in a smart-contract audit, except this audit has a geopolitical signature. The second structural layer is enforcement. Any minimum import price must be verified at the border, and verification is a data problem. Customs officials must compute the true cost-and-freight price of silicon that has moved across multiple intermediaries, sometimes through special-purpose vehicles in Singapore or the UAE, with volume rebates and hedging contracts obscuring the actual transaction price. You do not need to be an MEV searcher to see that this is a classic latency attack. Every import declaration becomes an attestation game - the same attestation game that DEX oracles faced when someone pushed an illiquid price feed. The Border Patrol does not have a Bloomberg terminal. It has a form. And the form will be gamed. Speed is the only shield in a flash loan; in customs policy, the shield is a disputed invoice, and the arbiter is an international trade tribunal that will take years. Enforcement latency creates a window in which importers can front-run the floor, and in that window, the effective US price is less than the announced floor. Political floors are always announced before they are enforced, and the gap between announcement and enforcement is the trade. I ran a similar playbook in 2023, when I allocated USD 25,000 into early EigenLayer restaking positions. I monitored the smart contract interactions manually to understand the slashing conditions, and I exited half the position once the incentives became unclear. The lesson I keep repeating: if you do not verify the mechanism, the mechanism will eventually verify you. A US import floor is a slashing condition written by lawyers, not by code, and its slashing event will arrive the first time a Chinese exporter invoices through a middleman in a free-trade partner country. The Energy Curve: What a Floor Does to Mining Let me make the crypto argument explicit, with numbers. Bitcoin mining is the world's most price-sensitive industrial buyer of electricity. Miners have zero tolerance for average electric cost above a hard threshold, because their product - hashrate - is a pure commodity that trades at a global price. When a miner signs a power-purchase agreement at USD 0.03 per kilowatt-hour, they are implicitly pricing in a future cost of capital, a network hashprice, and a specific equipment efficiency curve. Anything that changes the levelized cost of new solar energy in the US changes that calculation. Solar module cost is a direct function of polysilicon price. In a standard PV system, the module represents 30-40% of total installed system cost. Polysilicon itself accounts for roughly 15-20% of module cost. Double the silicon price and you lift module prices by something like 15-25%. A price floor set at USD 8-10 per kilogram, when the world price is 4-5, has roughly the same mechanical effect on module pricing as a global silicon price shock - the kind of shock that shook the market in 2022, when polysilicon nearly hit USD 40 per kilogram and slowed utility-scale solar deployment in several countries. Now apply that to mining. New solar plants are developed because the projected IRR works at a given module price. When module prices rise, what actually falls is the number of megawatts that get contracted. It is not a one-for-one reduction, but the direction is unambiguous. At the margin, in the US, there will be fewer new solar megawatts behind the meter that can be paired with mining fleets. That reduces the future supply of the cheapest energy available to the American mining industry, and it raises the cost of the average new PPA. For an industry that lives and dies by the spread between hashprice and the electric bill, this is a squeeze on margins. Here is the counterintuitive part, the piece that the tariff's authors did not model. Chinese producers will not raise their export prices to the US. They cannot, in fact, because they are fighting a brutal domestic overcapacity war. Here is what they will do instead: they will route around the wall. Silica is heavy, energy is heavy, and the actual scarce input in silicon-making is not a secret - it is cheap electricity and cheap capital. The Middle East has both. Saudi Arabia and the UAE are actively building polysilicon capacity, they are the fastest-growing jurisdiction for Bitcoin mining in the world, and they are the most likely destination for Chinese capital fleeing a tariff wall. The US plan to preserve a domestic silicon industry may have exactly the opposite geographic result: it accelerates the creation of a new, ex-China silicon manufacturing hub in the Gulf, and it places that hub in the same time zones and power grids where the next generation of mining fleets is being built. Arbitrage is just patience wearing a speed suit. The direct arbitrage is silicon: buy low-cost silicon in China or the Gulf, sell it at the protected US price, and let the floor do the yield work. The indirect arbitrage is energy: let Washington push US solar costs upwards, then watch mining capital flow to regions where PV-plus-mining co-location still makes sense. Those regions are the Gulf and parts of Central Asia. Every trade that this policy tries to block, it will re-route. The ASIC Endgame: Silicon Sovereignty and Hashpower Control Now the actual hidden driver. The original trade piece is a crypto outlet covering energy policy, which is true, but the policy itself has a semiconductor tail that crypto should not ignore. Solar-grade polysilicon is the same base ingredient as chip-grade silicon, and the US government's stated rationale for the policy - the phrase "chip supply chains" appears in the lead - is a direct reference to the fact that China's polysilicon dominance is also semiconductor input dominance. The tariffs are a test run. If the US can establish a competitive domestic polysilicon industry for solar, the same furnace infrastructure can be upgraded for chip-grade silicon over the course of a decade. The tariffs are the first step of an industrial policy that aims to give the US a sovereign silicon base. Crypto sits directly in that policy scope. Bitcoin's ASIC manufacturing is virtually monopolized by Bitmain and MicroBT - two Chinese firms that fab their chips in Taiwan and South Korea. There is no American ASIC production. There is no American source of chip-grade polysilicon at scale. If the US pursues silicon sovereignty, then at some point the crypto ASIC supply chain becomes a national-security topic, and that is not a bullish or bearish one-way trade; it is a volatility event. As a trader, I position for volatility events, not direction. Let me be precise. The semiconductor-grade polysilicon market is dramatically smaller than the solar-grade market - probably on the order of 50,000 to 70,000 tons per year globally - but it is much more profitable, and it is where a strategic player would spend money quietly. Hemlock Semiconductor operates in Michigan and has semiconductor-grade capacity. Wacker operates polysilicon plants in Germany. But the oligopoly that actually controls the bottom of the silicon cost curve is Chinese. If Washington makes it politically painful to import solar-grade silicon from China, the immediate effect is that US demand shifts to semiconductor-grade-adjacent domestic capacity, and the prices for that material rise. Do not be surprised if the result is an emerging price premium in every silicon-adjacent commodity that crypto touches - solar-rich mining PPAs, hardware exports, and the unit economics of mining as a whole. There is a strong parallel with layer-2 tech in crypto. Most so-called Bitcoin layer-2s are actually Ethereum projects wearing a Bitcoin jacket. This policy is similar: a trade policy for solar panels that is actually an industrial strategy for semiconductors, wearing a trade-dispute jacket. The framing is not the mechanism. If you buy the frame, you will miss the flow of real value. Case Study: The 2022 Silicon Shock as a Stress Test Look at the 2021-2022 silicon price spike. In December 2021, polysilicon spot prices hit roughly RMB 300,000 per ton, the equivalent of more than USD 40 per kilogram, up from USD 10-12 a year earlier. The cause was entirely demand-driven: Chinese silicon companies had under-built during COVID, module demand was exploding, and new electricity costs in Sichuan and Yunnan elevated production economics. The strategic effect was under-appreciated by crypto analysts. Solar deployment in countries with price-taker module markets slowed substantially. In the US, module prices jumped, EPC costs rose, and several utility-scale projects were delayed. Now imagine that shock, but not because of a demand-supply imbalance. Imagine it because a government slapped a floor price on top of a USD 4 global market. The market would do what it did in 2022, except faster, because there would be no underlying physical shortage to resolve. All of the signal would be policy, not physics. When the policy evolves, the price collapses to the physical floor. But that does not stop the spread from being harvested in the meantime. This is exactly the pattern I saw in the May 2022 Terra and Luna collapse. When UST lost its peg, the "floor" was a minting mechanism that assumed a buyer of last resort would step in. There was not one. The moment the spread between the peg and the market exceeded the cost of the mechanism, arbitrageurs did not save it; they broke it. A political floor on silicon is not a stablecoin floor, but the structural lesson is the same: any floor that depends on an external belief in an actor's willingness to price without limits is a floor that will be tested the moment belief wavers. In May 2022 I did not panic-sell because I had already marked my position to a scenario where the mechanism had a flaw. I lost 40% on the ride, but I survived because I had allocated away from the mechanism. The lesson here is the same: do not allocate at the floor level. Allocate at the spread level. The MENA Reroute: Where the Flow Actually Goes I have spent a good part of my career in the yield-farming end of crypto. In 2021, I deployed a Python script that extracted USD 14,500 in three weeks of pure arbitrage between SushiSwap and Uniswap by finding price discrepancies in small pools with low slippage tolerance. The script was boring. It scanned, it waited, it submitted transactions, it collected. The alpha was not in the code; it was in the structure that let the code wait. It was patience wearing a speed suit. That script is the metaphor for what happens next in silicon. The high-alpha position is not in American silicon producers. It is in the geographic chokepoints where silicon, energy, and hashrate will be co-located in the next cycle. That is the Gulf, particularly the UAE, which already hosts one of the largest known contiguous mining deployments in the world, and Saudi Arabia, which is building a national renewable-energy and industrial complex. In these countries, the cost of power is low, sovereign capital is abundant, and trade policy in Washington is irrelevant. Every tariff wall raises the value of being outside the wall. It does not destroy the flow; it routes it around the wall. This is not a fringe view. The same logic is driving Chinese polysilicon manufacturers to explore factories in the Gulf. They know the US market will pay a premium for non-Chinese silicon, and they know the Gulf offers cheap natural gas, abundant solar, and capital that does not ask awkward questions. The US floor will inadvertently create a Made-in-the-Gulf silicon export industry, and the same Gulf will also host the mining fleets that benefit from the stranded energy around those factories. Washington is not building a wall around America. It is building a toll road to the Gulf. Contrarian Who Actually Wins: The Winner Matrix The standard media interpretation will be: this hurts China, helps America. That is wrong on both legs. The first victim of a US polysilicon floor is not China. It is the American solar developer whose module costs just rose 20%, and the American mining fleet whose PPA just got more expensive. The actual winners are hiding in plain sight. First Solar, the American thin-film manufacturer, uses cadmium telluride and does not consume polysilicon; it is the largest US module maker, with roughly 20 GW of production capacity, and a polysilicon tariff is a direct tax on every competitor that ships silicon-based modules. First Solar effectively receives a subsidy by incumbency. This is not cheerleading; it is a direct consequence of taxing a substitute feedstock. Meanwhile, every non-Chinese polysilicon producer that can get certified for US imports - Wacker in Germany, REC in Norway, the upcoming Gulf entrants - gains a protected price umbrella. In a market where the global silicon price is below their cash cost, a US floor above their cash cost is a license to print money at low volume. The policy creates a high-margin sliver for non-Chinese producers without expanding total silicon supply. That is the classic trade-policy lottery, and it is why I do not trade the policy itself. I trade the spread it creates. The loser matrix is equally clear. American module assemblers lose, American installers lose, American ratepayers lose, and the American green-mining narrative loses. Chinese producers do not lose much at all; they simply divert shipments away from the US and let the floor define a premium for everyone else. The second victim is the "green Bitcoin" story in the United States - the claim that US mining fleets run on a growing base of renewable generation. When new solar gets more expensive, one of two things happens to that narrative: mining dries up, or mining shifts to gas-based power. In 2024, many US miners were already pivoting to flare-gas and thermal integration. Raising the cost of solar creates a comparative advantage for natural-gas-powered hashrate, which is the opposite of the climate message the industry has relied on for public standing. A policy that claims to protect American industry will, in effect, degrade the only part of the American mining industry that had a green halo. There is also a subtle smart-money detail: the price floor will be framed as a "minimum price" rather than an "anti-dumping duty." That difference is material. Anti-dumping duties are determined by a facts-based review and are challengeable at the WTO. A minimum import price is a managed-trade instrument that requires ongoing administrative machinery of price verification - and it creaks. Every managed price has a leak rate. In crypto, the equivalent is the difference between a hard-coded invariant and an admin key: one can be audited; the other can be changed by an internal team you do not monitor. Trust the stack, verify the exit. What the Floor Really Means for Hashprice Let us take the domino chain one step further, because the cleanest trade in this story is not in silicon at all. It is in global hashrate distribution. If the US floor raises module prices by 10-25%, US miners' marginal cost of energy rises. The most efficient US operators survive; the marginal ones, who rely on low-cost solar PPAs, will stall or shut down. As US hashrate retracts, global network difficulty adjusts downward, and the hashprice captured by every remaining miner - in Texas, in the Gulf, in Scandinavia - rises proportionally. This is the same mechanical logic as a token supply cut. A political policy that taxes US solar energy economics effectively functions as a supply-side cut on the highest-cost segment of global hashrate. That is bullish for the hashprice realized by non-US miners, and it is gently negative for US mining margins. The naive take is "tariff on China equals good for American miners." The correct take is "tariff on solar equals bad for American miners, good for everyone else." I hear the objection: "But the US is the largest mining market; if US miners cut back, network security decreases." The network does not care where hashpower is located. There is no geographic loyalty in the protocol. Code doesn't hate or love a country; it only measures difficulty. Any political intervention that pushes hashrate out of one region and into another is just an arbitrage game with a jurisdictional costume. Algorithms don't get tired. They keep scanning for the spread between the political price and the physical price, and if the physical price of energy is lower in the Gulf, that is where the next megawatt of hashrate answers. This is also the answer to the question nobody asks in the trade press: why would a mining operation care about a silicon price floor? Because the PPA in that operation is signed with a solar developer who is one module-price shock away from a renegotiation or a curtailment. A module price shock of 15-25% does not cancel the sun, but it absolutely changes the capital structure of the developer that promised you power. In crypto terms, this is counterparty risk that does not appear in a CEX order book. It is in the footnotes. I have been reading footnotes for five years, and the single most valuable skill is to find, not the thing that is priced, but the thing that is implied. The floor is implied. The counterparty risk in US solar PPAs is the implication. The Bitcoin Layer-2 Lesson Applied to Solar Let me return to a pet peeve. Probably 90% of the projects that call themselves "Bitcoin Layer-2" are Ethereum projects that wrapped themselves in the Bitcoin brand for funding. The tactic is not new. In the policy world, the same tactic is called "national security." A tariff justified by solar self-sufficiency is, in fact, a semiconductor-industrial-policy maneuver. When the frame and the mechanism diverge, the correct response is not to argue about the frame; it is to extract the mechanism. The mechanism in this case is not protection of American consumers, and not even necessarily protection of American solar workers. It is an attempt to re-create the entire silicon base of the American semiconductor industry inside a tariff-friendly cocoon. Crypto is going to learn the cost of that cocoon because mining ASICs are the most exposed part of the American tech stack: if chip-grade silicon becomes a national-security quota, then ASIC procurement becomes a diplomatic issue. Hashrate sovereignty is a real thing. It matters for who can mine, and who cannot. In crypto, the most durable moats are now regulatory licenses. Binance paid a USD 4.3 billion fine and emerged more entrenched because the license became the barrier to entry for new exchanges. The silicon floor is the same play: the moat is the compliance apparatus, and the winners are incumbents who can afford it. New entrants cannot. Whether in exchanges or in silicon, when the cost of compliance exceeds the cost of production, the compliant incumbent wins by default. That is the actual design pattern of this policy. I also want to flag a danger in the market narrative. Just as "guaranteed returns" are a red flag in DeFi, "guaranteed domestic production" is a red flag in trade policy. Yield farms that promise guaranteed returns are usually structures where the risk has been deferred, not eliminated. The 2022 Terra crash taught me that yield is often a deferred risk premium. The same phrase should be applied to a domestic silicon champion: if you guarantee a price above the world price, you are not reducing risk. You are selecting winners, and the winners are the people who can meter that subsidy. The subsidy will be paid by American solar consumers and American mining ratepayers. I have been through enough cycles to know that the first market reaction to such a policy is a price spike, followed by a slow bleeding as the arbitrage mechanism reroutes physical flows. It is exactly what happened with the Section 201 tariff on solar modules in 2018: module prices rose, imports shifted to Southeast Asia, and American module assembly capacity - the exact thing the tariff was intended to protect - lost market share to tariff-circumvention routes. In 2022, the US court system struck down parts of the roaming solar-tariff policy, but only after years of damage had been done. The lesson is not to be early. The lesson is to be positioned for the reroute, which always happens, and takes far longer than politicians promise. The ZK Proof of a Price Floor I want to give you one more mechanism, and this one comes from my old L2 research. ZK rollups are elegant until the proving bill arrives. Proving a batch of transactions is expensive, and the gas price that justifies the proving cost only exists in a bull market. When the market collapses, the ZK rollup operator is left holding a proof whose cost exceeds the value it secures. That is, in one sentence, the entire problem with a political price floor on a commodity. The support mechanism is expensive, it only works when the market is willing to pay for it, and when the market turns, the mechanism is the first thing to fail. Applied to polysilicon: the support mechanism is the administrative machinery of the US import price floor. It is expensive to operate. It requires a permanent staff of customs reviewers, an appeal process, and an assumptions-based cost model of global production. When the global price is low and the US burns money on administration, the floor is a negative-sum operation. When the global price is high, or when a supply disruption hits as it did in 2022, the floor becomes irrelevant, because the market has moved above it. A floor that is relevant only in cheap markets and does nothing in tight markets is a transfer from US consumers to the administrative class and to non-Chinese silicon incumbents. The proof cost exceeds the output in exactly the same way a ZK rollup bleeds in a bear market. ZK operators are bleeding money now unless gas returns to bull-market levels. The US government is about to discover the same thing: the floor will bleed unless the free-market world price returns to US floors, which is unlikely. In 2025, I audited an AI-driven trading bot that claimed 30% monthly returns. By reviewing its API keys and transaction logs, I found it was merely executing high-frequency, low-margin trades on decentralized exchanges, incurring excessive gas fees. I shorted the associated token after exposing the lack of edge. The polysilicon floor has the same structure: an expensive machine that looks like industrial strategy and is actually a toll collector. If you cannot verify the mechanism, do not buy the narrative. The Signal List: What to Actually Monitor What should a crypto reader monitor? I have a small checklist, and I keep it tighter than most. One: the actual text of the executive order. It will specify whether the floor is a "minimum import price" or a "duty." The difference is the difference between a contract and a threat. If the floor is implemented as a duty, it will be challenged and litigated, which takes years - and during those years the spread is tradable. If it is a minimum import price with enforcement, the spread narrows faster. Two: the invoice price books for US imports. The first six months of enforcement will show wide dispersion in reported silicon prices. That dispersion is the US customs equivalent of an oracle-manipulation event. Whatever the floor is set to, the first six months of data will show that the true effective price is lower. That dispersion is the trade, and it is the signal I watch. Three: the US PPA term-sheet market. Watch for the rate at which new solar PPAs in Texas and California quote higher prices. If the module-cost pass-through shows up in the PPA curve within two quarters, that is the earliest confirmation that the floor is biting US mining energy costs. If PPA prices rise but mining buildouts continue, the market has found a way to route around the wall - likely through non-module procurement, CdTe modules from First Solar, or merchant power that was not footprinted by the policy. Four: the Gulf's hashrate share. The most direct confirmation of my thesis is the share of global hashrate hosted in the Middle East. If that share rises in the next 18 months while US hashrate growth slows, the tariff routed the flow exactly where the physics says it should go. The reason these four signals matter is that they allow the reader to trade the structure, not the narrative. I do not have a directional view on the price of silicon itself. I have a directional view on the spread between Washington's stated price and the world's physical price. Spreads are where I do my best work, and a floor is the most durable spread-creating instrument a government can create without meaning to. One more layer: the carbon angle. The US floor might later be paired with a carbon border adjustment, because polysilicon is carbon-intensive. Producing one kilogram of polysilicon consumes 40-60 kilowatt-hours of electricity, and Chinese producers running on coal-fired grids carry a carbon footprint of 30-50 kilograms of CO2 equivalent per kilogram, versus 10-20 for hydro or gas-based producers. If the US later imposes a carbon fee on imported silicon, the floor becomes a climate policy, and the international optics improve. European CBAM is already moving in that direction. That is the long-dated tail of this trade, and it is why Chinese producers will not simply wait out the tariff; they will relocate to low-carbon regions, further consolidating the Gulf's position. Takeaway Floors do not hold. They never have. The US housing market floor held only because the government became the buyer of last resort. The yen floor holds only because the Bank of Japan actually buys. The floor on polysilicon will not be backed by a buyer, and so it will not be a floor; it will be a toll booth, and the toll will be paid by American solar and American mining, while the flow of capital moves to the Gulf. If you want the trade, do not buy the American manufacturer. Buy the spread. Buy the region that can route around the wall. The checklist is simple. Watch the executive order text, watch the dispersion in import declaration prices, watch the Texas PPA curve, and watch the Middle East hashrate share. If the Middle East share rises, the tariff routed the flow as the physics says it should. If it does not, the tariff is smaller than it appears, and the trade dies early - in which case, you have lost only the opportunity cost of a patient position. Position sizing and exit strategy matter more than direction. I do not trade hopes. I trade verification. When I audited the Uniswap V2 factory, the report that told the market something existed and the code that actually executed something were two different documents. I read the code. Read the floor, then read the physical trade, and then price the gap between them. In that gap, there is a yield. The yield is just patience wearing a speed suit, and Washington is the counterparty that does not even know it is in the trade. The end is not a price target. It is a warning about mechanism design. The US floor is a proof system with no prover, a support line with no market maker, and a yield-generating instrument that was not labeled as such. Markets will find it out. They always do. The only question is whether you are positioned before the floor is tested, or after. I know which side of that question I want to be on.