The Antminer S9 is dead. Not metaphorically dead. Economically dead. At 98 J/T, it burns capital faster than it burns power. The code does not lie; it only waits to be read. And the code of the Bitcoin mining industry, as revealed by a recent study from Shannon University of Technology (TUS), is unequivocal: even with free energy—curtailed wind power—the current generation of miners cannot turn a profit. The study, published in Energy Economics, modeled a 20MW mining farm paired with an Irish wind farm. The conclusion was stark: at current hashrate and price, the project generates a negative NPV of €10.1 million over six years. Free energy. Negative returns. That is the structural contradiction at the heart of the mining industry.
Let me be clear: I have spent years auditing protocols and modeling risk. In 2020, during DeFi Summer, I modeled Compound Finance’s interest rate curves across 50,000 blocks. I discovered that liquidity traps form when volatility spikes. The same principle applies here. The trap is not the price of electricity. It is the differential between Bitcoin's price growth and the global hashrate growth. The TUS study exposes this with surgical precision. Their sensitivity table is the most valuable part of the paper. It shows that when both price and hashrate grow by 30%, the NPV collapses to -€10.1 million. Only when hashrate grows at half the rate of price (15% vs 30%) does the model generate a positive NPV of €7.7 million. That is a razor-thin margin. And the real world has already exceeded the study's baseline hashrate: 911 EH/s in August 2024 versus the 780 EH/s assumed in the model. The trap is already closing.
Context: The Model and Its Assumptions
The TUS study is not a whitepaper. It is an academic simulation. The authors used 2024 Irish electricity market data, Antminer S21 Hydro hardware (16 J/T), a six-year equipment life, and a 25% wind curtailment rate. The farm absorbs 83.1% of the curtailed energy at 20MW, rising to 93.4% at 30MW. The key output is the net present value of the project over six years, assuming the miner sells all Bitcoin immediately. The model’s assumption that the miner can achieve perfect foresight on electricity prices and curtailment patterns is a stretch. But the structure is sound. The method is replicable. The conclusion is brutal: under realistic price and hashrate scenarios, the investment is underwater.
The study also highlights the hardware dependency. The S21 Hydro at 16 J/T is the minimum viable. The S9 at 98 J/T is catastrophic in all scenarios. This is not a surprise to anyone who has tracked the industry. But the TUS study quantifies it: the efficiency gap between old and new hardware is a factor of six. That means the operating cost per TH/s for an S9 is six times that of an S21. In a market where the hashprice (revenue per TH/s per day) is $31.73, an S9 miner is losing money every second it runs. Integrity is not a feature; it is the foundation. And the foundation of the mining industry is hardware efficiency. Without it, there is no business.
Core: The On-Chain Evidence Chain
Let me lay out the evidence chain, step by step. First, the hashrate. The global hashrate has grown from 780 EH/s (the study’s baseline) to 911 EH/s in just a few months. That is a 17% increase. The model assumes no hashrate growth over six years. That is an absurdly optimistic assumption. In reality, hashrate has grown at a compound annual rate of 30-50% over the past decade. Even if we assume a conservative 20% annual growth, the miner’s share of the network drops by 50% in three years. The study’s NPV calculations are based on a static hashrate, which means the real-world returns are significantly worse.
Second, the price-hashrate sensitivity. The study’s sensitivity table is a diamond in the rough. It shows four scenarios: (1) BTC price +30%, hashrate +30% → NPV -€10.1M; (2) BTC price +30%, hashrate +15% → NPV +€7.7M; (3) BTC price -30%, hashrate +30% → NPV -€18.5M; (4) BTC price -30%, hashrate +15% → NPV -€2.5M. The only scenario with a positive NPV is when price growth doubles hashrate growth. That is a rare condition. In the 2024 market, price has been range-bound ($60k-$70k) while hashrate has surged. The differential is negative. The code does not lie: the miner is bleeding.
Third, the hardware replacement cycle. The S21 Hydro is a 2024 machine. The S9 is 2016. The industry has moved through three generations in eight years. The TUS study assumes a six-year equipment life. That is generous. By year three, the S21 will be obsolete. The miner will either need to replace the hardware or accept a lower share of the network. The capital expenditure is not a one-time cost; it is a recurring obligation. The study’s NPV assumes no further investment. That is a flaw.
Fourth, the curtailment assumption. The model uses a 25% curtailment rate. In reality, wind curtailment is highly variable. It depends on grid congestion, weather patterns, and storage capacity. In Ireland, curtailment rates can hit 50% in high-wind months and drop to 5% in low-wind months. The miner cannot rely on a consistent supply of free energy. The farm’s uptime is at the mercy of the grid operator. The study’s assumption that the miner can operate at full capacity for 83.1% of the curtailed hours is optimistic. Real-world curtailment is unpredictable.
Fifth, the transaction fee revenue. The model ignores transaction fees. In 2024, fees have accounted for 2-10% of miner revenue. During the Runes protocol frenzy, fees spiked to 40%. The model’s revenue is entirely based on the block subsidy. That is a significant omission. If fees become a larger share of revenue, the economics improve. But the model does not account for that. The takeaway is that the study’s conclusions are conservative. They are the worst-case scenario for a miner relying solely on the block subsidy.
Contrarian: The Myth of Free Energy Mining
The common narrative is that renewable energy mining is the future. “Free energy makes mining profitable.” The data says otherwise. The TUS study shows that even with zero-cost electricity, the capital expenditure on hardware and the relentless hashrate growth make it a losing proposition. The real driver is not energy cost but the relative growth rate of hashrate versus price. Most investors overlook this. They focus on the electricity cost and ignore the hashrate decay. That is a classic error.
Consider the zombie miner problem. There are still S9 units running. They are consuming energy and producing nothing but heat. They are not profitable. They are subsidized by early believers who refuse to shut them down. The TUS study confirms that these miners are dead. They just don’t know it yet. The industry is heading toward a purge. The efficient miners will survive. The rest will be forced to pivot or exit.
The pivot to AI is the other side of the coin. Riot Platforms signed a 191MW lease with an AI lab, valued at $9.1 billion to $16.1 billion. CoinShares reports that listed miners have accumulated over $70 billion in AI contracts. By the end of 2024, 70% of miner revenue could come from AI. This is not a diversification. It is a transformation. The mining industry is becoming a data center industry. The Bitcoin network loses its energy security buffer. The code does not lie: the mining industry is migrating away from Bitcoin.
Takeaway: The Next-Week Signal
The next-week signal is the hashrate growth rate. If it continues to outpace price, more miners will be forced to shut down or pivot. The TUS study is a warning. It is not a prediction. It is a structural analysis. The model shows that even under ideal conditions—free energy, new hardware, stable price—the returns are negative. The only way to make money is to bet on a price-hashrate differential that has historically been rare. The intelligent miner will hedge. The foolish miner will buy S9s.
I have seen this pattern before. In 2022, I traced the Terra/Luna collapse back to the death spiral in the code. The market believed the narrative. The data revealed the truth. The same is happening here. The narrative is that renewable energy mining is green and profitable. The data says it is neither. The code does not lie. It only waits to be read.
Integrity is not a feature; it is the foundation. The foundation of the mining industry is cracking. The only question is whether the miners will see the cracks before the structure collapses. The S9 is dead. The S21 is barely alive. The next generation will be AI servers. The Bitcoin mining industry is not dying. It is evolving. But the evolution is brutal. The data is clear. The choice is yours.