The data is brutal. Between 2017 and 2025, the total sales revenue of Bitcoin ASIC miners stayed flat at 300–400 billion yuan. Gross margins collapsed from 80–90% to 20–30%. Ledgers do not lie, only analysts do. The numbers are clear: the golden age of mining is finished.
Yang Zuoxing, founder of Shenma (Whatsminer), delivered this diagnosis at a closed-door industry summit in late July 2026. His words carry weight—he built the second-largest ASIC manufacturer on the planet. Yang offered three escape routes: natural gas mining, AI-integrated mining rigs, and solar-powered operations. Each sounds promising on the surface. Beneath the surface, they reveal structural cracks that few are willing to audit.
Context: The Collapse of the Old Model
The ASIC market has matured. Gross margins have fallen by 60–70 percentage points over eight years. Two forces drive this: the 2024 Bitcoin halving halved block rewards, and competition among manufacturers turned into a race to the bottom. Shenma, Bitmain, Canaan, and MicroBT all fight for the same shrinking pie. Meanwhile, AI is stealing both electricity and capital. Data centers for large language models now consume energy equivalent to 5% of global Bitcoin mining hashrate, and that share is growing.
The traditional model—buy rigs, plug into cheap hydro, sell coins—no longer works. Marginal miners have already been squeezed out. Hashrate concentration is rising. The survivors need a new edge.
Core: Three Flawed Lifelines
Let’s dissect each escape route with the rigor of a code audit.
1. Natural Gas Mining
The idea is simple: locate rigs near oil wells that flare natural gas. Capture that waste energy and turn it into Bitcoin. This is not new—projects like Crusoe Energy already operate at scale. But Yang frames it as a mainstream direction, not a niche. The economics rely on gas being free or near-free. However, regulatory risk is high. The Biden administration’s proposed methane fees (2025) and EU’s Carbon Border Adjustment Mechanism threaten to make flaring expensive again. If carbon costs are passed through, the advantage evaporates. Natural gas mining is a geopolitical arbitrage, not a technological breakthrough.
2. AI-Integrated Mining Rigs
This is the sexiest narrative: repurpose ASIC facilities to service AI inference. Yang hinted that future rigs might host both SHA-256 chips and general-purpose GPUs, sharing power, cooling, and networking. The problem? Hardware specialization. ASICs are designed to do one thing efficiently: compute SHA-256 hashes. Adding GPU clusters requires fundamentally different architecture—PCIe lanes, memory bandwidth, software stacks. No existing mining rig can do both without crippling efficiency. The industry is talking about co-location, not integration. True convergence would take 3–5 years of R&D and billions in investment. The market is pricing this as a near-term catalyst. It is not.
3. Solar-Powered Mining
Solar has fallen in cost by 90% over the past decade. Pair panels with batteries, and you get 24/7 mining with zero marginal energy cost. But the geographical reality bites: only a handful of regions (Atacama Desert, Australian outback, parts of the Middle East) have enough solar irradiance and land. Even there, battery storage for nighttime operation doubles the capital cost. At current Bitcoin prices (~$65k), the levelized cost of solar mining hovers around $0.08/kWh—not far from industrial grid rates. The margin advantage is thin unless subsidies exist. Solar mining is a hobby for green enthusiasts, not a scalable industry.
Contrarian: What the Bull Market Misses
Retail investors and mining stocks jumped on Yang’s speech. Shenma’s valuation has popped 15% in two weeks. But enthusiasm masks the hard truth: none of these directions can reverse the structural decline in mining profitability. The hashrate growth has already slowed from 50% YoY in 2021 to 15% in 2025. Post-halving, only the most efficient rigs (S21, M60S) earn positive cash flow at current power prices. Every new direction requires years of infrastructure buildout. Volatility is the tax on uncertainty. The market is discounting a future that might never arrive.
The contrarian bet is bleak: mining will become a long-tail industry—persistent but small. Hashrate may plateau at 700 EH/s. ASIC manufacturers will pivot to AI hardware or die. Miners who survive will be those with locked-in power contracts below $0.04/kWh, not those chasing hot narratives. Risk is not a rumor, it is a variable. Measure it.
Takeaway: Watch the Real Signals
Ignore the conference hype. Track two numbers: used rig prices and network hashrate. If S19s fall below $5/TH (they are at $8/TH today), that’s a signal that marginal miners are fleeing. If hashrate drops 10% month-over-month, the golden age is truly buried. Until then, treat the three new directions as experimental—not investable. The market owes you nothing. Audit the code, not the hype.