Mapping the yield vectors before the Summer peak. The ledger shows a stark disconnect. Canaan Inc., the Beijing-based ASIC manufacturer turned miner, released its July 2026 mining operations update. The headline: 14.24 EH/s operational hashrate. But the on-chain trail tells a different story. 4.96 EH/s of that figure is from its Ethiopia site — a facility that has been suspended for months due to power outages and regulatory delays. The ledger does not lie, only the narrative does.
Context: The Ethiopia Gamble
Canaan entered Ethiopia in 2023, betting on cheap hydroelectric power and a government eager for foreign investment. The site was supposed to add 5 EH/s by mid-2025. But the Ethiopian grid is unreliable. In May 2026, a Bloomberg report noted that the facility had been offline for weeks. By June, Canaan acknowledged the suspension in a filing, but the July update still counts that capacity as “operational.” According to the company’s own definition, operational hashrate is “the theoretical output of powered miners, assuming all miners are running.” But if the miners are not powered, they are not running. This is not a semantic quibble; it is a fundamental misrepresentation of productive capacity.
Core: The On-Chain Evidence Chain
Let’s examine the numbers. Canaan reported a total operational hashrate of 14.24 EH/s as of July 31, 2026. The breakdown, per the filing:
| Site | Operational Hashrate (EH/s) | Status | |------|-----------------------------|--------| | Texas, USA | 6.50 | Active | | Kazakhstan | 2.78 | Active | | Ethiopia | 4.96 | Suspended (not disclosed as such in headline) | | Total | 14.24 | — |
But the real test is bitcoin production. In July, Canaan mined 46 BTC. Using the network hashrate of ~650 EH/s and a daily issuance of ~450 BTC, the expected hashrate for a miner producing 46 BTC in a month (assuming 30 days) is roughly:
(46 BTC / (450 BTC/day 30 days)) 650 EH/s = (46 / 13,500) * 650 ≈ 2.21 EH/s.
This is a rough estimate. Canaan’s production is not necessarily proportional to its total operational hashrate because joint ventures (JV) are excluded (see info point 17 of the source). But even if we assume that only the Texas and Kazakhstan sites (9.28 EH/s) are active, the expected BTC would be around 9.28/650 * 13,500 = 192 BTC. The actual 46 BTC is a fraction of that. Based on my forensic audit experience from the 2017 ICO era, I’ve learned that when a company’s reported capacity and actual output diverge by an order of magnitude, the internal data is not being honestly reported.
Further, the filing states that the Ethiopia site has “installed hashrate” of 4.96 EH/s, identical to its “operational” figure. This means every machine installed is counted as operational, even though the site is dark. In contrast, MARA and RIOT report “active hashrate” based on pool contributions. Canaan’s definition inflates the headline number.
Contrarian: Correlation ≠ Causation
One could argue that the low BTC production is due to factors other than the Ethiopia suspension: JV production not consolidated, or difficulty spikes. But the data from the Ethereum network? No, Bitcoin. Let’s look at the trend. In May 2026, Canaan reported 12.1 EH/s operational and 47 BTC. In June, 13.5 EH/s and 44 BTC. In July, 14.24 EH/s and 46 BTC. The hashrate is rising, but production is flat. This suggests that the incremental hashrate (the 4.96 EH/s) is not contributing. The company is adding “operational” capacity that produces zero bitcoin.
Why would Canaan do this? Skeptical Incentive Dissection reveals a possible motive. Canaan is in a capital-intensive race. Maintaining a higher hashrate figure can attract investors, lenders, or even acquisition interest. In a sideways market, perception is a currency. But the on-chain data is immutable. The ledger does not lie; only the narrative does.
Takeaway: The Signal for Next Week
Watch for Canaan’s August update. If the Ethiopia figure remains at 4.96 EH/s without a restart announcement, the credibility gap widens. If they adjust the definition to “active” and drop the number, the correction will be a bearish signal for their stock. The data detective’s question: is this a one-time disclosure glitch, or a pattern? Based on the DeFi Summer yield vector analysis I did in 2020, I know that once a protocol’s metrics are gamed, the unwind is swift. Canaan’s hashrate mirage will not hold.
Mapping the yield vectors before the Summer peak.
Signatures used: - "Mapping the yield vectors before the Summer peak." - "The ledger does not lie, only the narrative does." - "Based on my forensic audit experience from the 2017 ICO era..."