11:00 UTC, July 20, 2026. A Chinese lighting company called Yangdian Technology (301012.SZ) drops an announcement: an 8.6 billion yuan (≈$1.2B) "computing power service" contract with an anonymous client. The term: 60 months. The subsidiary executing the deal: Sichuan Hanyang Intelligent Technology—registered in the province that once hosted 60% of the world's Bitcoin hashrate. The contract represents 67.22% of Yangdian's total 2025 revenue.
No client name. No technical spec. No collateral. This is not a contract. This is a blind bet.
The first scar is already there: the anonymity. Every transaction leaves a scar; I find the wound. For a listed company to pin two-thirds of its future revenue on an unknown counterparty in a legally grey sector is not a sign of confidence—it is a distress signal. Based on my experience auditing over 150 ICO contracts in 2017, I learned to spot when the narrative hides the risk. 80% of those whitepapers failed because their tokenomics were built on sand. This contract is built on the same sand.
Context: The Anatomy of a Regulatory Arbitrage
Yangdian Technology was never a computing power company. Its core business is smart lighting and energy efficiency—legitimate, low-margin, unexciting. By 2025, its revenue had stagnated at around 1.3 billion yuan. The management needed a story. And in 2026, the hottest story in Chinese A-shares is "computing power services"—a euphemism for cryptocurrency mining and AI compute, but mostly mining.
China's notorious "924 Notice" (September 24, 2021) explicitly banned cryptocurrency mining. But regulators cannot ban "computing power services." The contract carefully avoids the words "mine," "Bitcoin," "hashrate." Instead, it promises "to provide stable, high-performance computing resources." This is the oldest trick in the regulatory playbook: rename the action, keep the substance.
In May 2022, the algorithm ate its own tail. The Terra collapse was not caused by code failure—it was caused by humans ignoring the structural risks embedded in a model that promised high yields with no backing. This contract is a similar model: high revenue upside, no transparency, no shock absorbers.
Core: On-Chain Forensic Deconstruction of the Contract
Let me apply my standard forensic pipeline—trace the money, model the flows, find the hidden assumptions.
Step 1: Decompose the Cash Flow
- Total contract value: 8.6 billion yuan
- Term: 60 months (5 years)
- Implied monthly service fee: 143.3 million yuan (≈$20M)
- Annual service fee: 1.72 billion yuan
- Yangdian's total 2025 revenue: 2.56 billion yuan → this contract alone covers 67%.
But that is revenue, not profit. To generate 143 million per month in computing power service fees, Yangdian must first deploy capital. How much?
Step 2: Estimate the Hardware Needed
Assume they are mining Bitcoin (the most likely, given Sichuan's cheap hydro and existing miner relationships). Current ASIC miner price: ~$3,000 per unit for a S21 Pro at 200 TH/s. Monthly electricity cost per unit: ~$150 at Sichuan rates ($0.05/kWh). To generate $20M in monthly service fees (which likely includes a profit margin for Yangdian—say 30% gross margin), the hardware cost alone would be roughly:
- Required gross revenue for hardware recovery: $20M / 0.3 = $66.7M total hardware pool? No—need to separate service fees from hardware cost. If Yangdian owns the machines, its cost is depreciation + power + ops. A common industry benchmark: gross margin for mining hosting is 20-30%. So cost of goods sold (COGS) ≈ $14-16M per month. The bulk of that is electricity and depreciation.
- At $0.05/kWh, a 1 PH/s Bitcoin miner draws ~3,000W and costs ~$1,314/month in electricity. A S21 Pro (200 TH/s) draws 3,000W = 0.21 PH/s. To power 1 EH/s, you need ~4,762 units at $1,314/month = $6.26M/month power. 1 EH/s is roughly 0.2% of current global hashrate (~500 EH/s).
- If Yangdian deploys 2 EH/s (a plausible number for a single mining farm), its monthly electric bill alone would be ~$12.5M. Add depreciation ($2M), ops ($1M) → total cost ≈ $15.5M/month. To hit $20M in service fees, they'd need to charge ~$10 TH/s/month. Current market rate for hashpower is ~$0.07-0.10 per TH/s/day, or $2.1-3.0 per TH/s/month. So 2,000,000 TH/s (2 EH/s) would generate $4.2-6M in revenue—far below $20M.
Conclusion: The numbers don't add up. Either the service fees are massively inflated (unlikely for a real client), or the computing power is not Bitcoin mining but high-end AI compute (e.g., NVIDIA H100 GPUs). At $3 per GPU-hour, to generate $20M/month you'd need ~7,000 H100s running 24/7, which costs ~$200M in hardware—a 10-year payback. Not plausible.
Step 3: The Anonymous Client Trap
If the numbers are this off, why sign? Because the client may be a shell—a conduit to raise capital on the stock market. The real "service" is not computing power, but a narrative to pump Yangdian's stock.
The 2017 code was honest; the humans were not. In 2017, I built an audit pipeline for ICOs. The token contracts were often transparent—the whitepapers were lies. Here, the contract text is transparent (8.6B, 60 months). The lie is what it represents.
Contrarian: The Market Cheers, But the Data Screams Risk
Retail investors and momentum funds see this as a "Dragon and Tiger" story—a mundane company morphing into a tech powerhouse. Some analysts are already drawing parallels to CoreWeave. But correlation is not causation. The fact that a lighting company signs a big contract does not mean it can execute.
Here is the counter-intuitive truth: This contract makes Yangdian more fragile, not stronger.
- 67% revenue concentration on a single anonymous client is a classic single-point-of-failure. If client A defaults or the regulatory ax falls, Yangdian loses two-thirds of its business overnight.
- The contract does not disclose penalties, force majeure, or dispute resolution. In China, contracts tied to illegal activities (mining) can be voided by courts without penalty to either side.
- Yangdian's balance sheet: as of Q1 2026, cash and equivalents were 210 million yuan. To deploy the hardware needed to fulfill this contract, they would need to raise debt or dilute equity. The 8.6B is not incoming cash—it is future receivables. The upfront capex is real and massive.
This is not a growth story. It is a leveraged derivative on regulatory tolerance.
Takeaway: The Next Signal to Watch
I've traced enough on-chain scars to know that when a deal looks too neat—anonymous client, huge revenue, legally perfect language—it leaves a trail. The trail here is intangible, but predictable.
Over the next two weeks, watch for two things:
- Disclosure of Client A's identity. If they keep hiding it, assume it's an insider or a special-purpose vehicle. That alone should flash red.
- Regulatory reaction from Sichuan Energy Administration or the National Development and Reform Commission. A quiet inquiry, a meeting notice, or a public statement will be the first real wound. If nothing happens for a month, the bet is still alive—but risk has not disappeared, only deferred.