I didn’t say the golden age is over. The numbers did. The gross margins collapsed. From 80-90% in 2017 to barely 20-30% in 2025. Sales volume stayed flat at 300-400 billion yuan across three cycles. The same revenue, but the profit per machine got surgically removed.
Yang Zuoxing, the founder of Whatsminer, stood on stage at a closed-door industry summit in late July 2026 and confirmed what every serious operator already felt: the easy money in Bitcoin mining is gone. “The golden age of Bitcoin mining is over,” he said. He didn’t sugarcoat it. He laid out the data, the trend lines, and the brutal math.
This is not a bear market whine. This is the structural decay of an industry that once printed money faster than any central bank. The hook here is not the quote—it’s the order flow behind the margin compression. Let me walk you through what I see when I strip away the narrative.
Context
Bitcoin mining has always been a capital-intensive game. ASIC manufacturers like Bitmain and Whatsminer dominated the supply chain. Miners bought machines, plugged them into cheap electricity, and sold the coins. For years, it worked. The 2017 boom saw obscene margins. The 2020 DeFi summer didn’t directly hit miners, but the 2021 bull run kept the party alive. Then came the 2024 halving. Block rewards dropped from 6.25 to 3.125 BTC. The revenue per machine halved overnight. But the competition for hash power didn’t ease. Modern ASICs like the Antminer S21 and Whatsminer M60 pushed efficiency, but the industry flooded with supply.
Simultaneously, a new competitor emerged: AI. Data centers hungry for GPUs began bidding up the same electricity contracts and capital markets that miners relied on. The result? A perfect margin squeeze.
Yang’s data confirms it: “Sales volume of mining machines remained at 30-40 billion yuan, but gross profit margin dropped from 80-90% to 50% and now further to 20-30%.” This is not a temporary dip. It’s a secular shift.
Core
The core insight is not the margin decline itself—it’s what drives it. I’ve audited mining operations for my copy trading community since 2022. I’ve poured over P&Ls from farms in Texas, Kazakhstan, and Norway. The pattern is consistent: the marginal cost of producing a Bitcoin has risen faster than its price. In 2017, a decent farm could produce Bitcoin at $3,000 when the price was $10,000. By 2025, the all-in cost for the average miner is pushing $45,000, while Bitcoin trades in a range. The gap shrinks daily.
But let’s go deeper than averages. The order flow of miner selling reveals a brutal truth. Miners are now forced sellers more often. In 2021, they could hodl. In 2025, they must cover electricity bills immediately. This structural selling pressure caps Bitcoin’s upside during accumulation phases. I’ve seen it in the tape: every 5% rally in 2025 was met with a wall of sell orders from institutional mining treasury desks.
Yang proposed three escape routes: natural gas flaring capture, AI integration (converting mining infrastructure to host AI chips), and solar-powered mining. Each is a bet on energy arbitrage, not on Bitcoin’s price. Hype is a liability; liquidity is the only truth. And the liquidity of mining profits is evaporating.
The financial engineering behind these new directions is shaky. Natural gas mining works only if the gas is nearly free—and it is, in the Permian Basin. But scaling it requires regulatory approval and infrastructure that takes years. AI integration sounds sexy. I looked into it. The cooling and power delivery systems for ASICs are completely different from those for GPUs. Retrofitting a mining farm to run H200s costs more than building a new AI data center. The numbers don’t pencil out yet. Solar mining suffers from intermittency: Bitcoin mining needs 24/7 baseload power unless you couple it with batteries, which destroys the economics.
Contrarian
The market narrative is that AI will save the mining industry. I disagree. AI integration is the most overhyped lifeline. The big cloud providers want dedicated AI data centers, not converted mining barns. The capital expenditure to convert is immense, and the operational expertise is different. Mining is about uptime and power cost. AI is about latency and network architecture. Cross-training a mining team takes years.
The real contrarian play? The death of the golden age is actually a healthy purge. High margins attract amateurs. Low margins force efficiency. The miners that survive will be the ones that operate with institutional discipline: hedging their production, negotiating fixed-power contracts, and maintaining zero debt. I’ve seen this in my own trading. When the weak hands exit, the remaining players consolidate power. The Bitcoin network’s hash rate might drop temporarily, but it will bottom with a stronger base.
Another blind spot: regulatory risk is asymmetrically negative. The article didn’t mention it, but the three new directions each have a regulatory sword hanging over them. Natural gas flaring is under scrutiny from the EPA. Solar farms require land permits that local communities often oppose. AI data centers are facing power rationing in regions like Europe. The compliance burden will crush small miners who try these paths without legal backing.
Takeaway
The golden age is over. Stop trying to predict the storm. Start building the ship. The actionable level here is not a price target for Bitcoin. It’s a signal for mining-related equities and ASIC manufacturers. Watch the gross margins of publicly listed miners like Marathon Digital and Riot Platforms. If they stay below 30% for two consecutive quarters, expect consolidation. For traders, the key metric is not hash price but the bid-ask spread on used ASICs. A widening spread signals distress. Trust the code, verify the chain, own the outcome.
The next move isn’t a bullish breakout. It’s a structural shift in who owns the hash. I’ll be watching the order books.
Signatures - I didn’t say the golden age is over. The numbers did. - Hype is a liability; liquidity is the only truth. - Trust the code, verify the chain, own the outcome.
