Alert. A line buried in a Q2 filing just moved the entire risk surface for one of the most unusual public mining companies in America. Cango Inc. — the Shanghai-born auto-financing firm that pivoted into Bitcoin mining under ticker CANG — reported $47.4 million in mining revenue for the quarter.
Most readers will compare that to Marathon Digital’s $145 million and move on. They shouldn’t. Because the more interesting question is not whether Cango is bigger or smaller than Marathon. The question is: How does a former car-loan company produce $47.4 million in mining revenue while disclosing almost nothing about its hashrate, its power contracts, or its AI compute pipeline?
I ran the forensic math. It doesn’t square with self-mining alone. And that discrepancy is not an accounting error. It’s the entire thesis.
Let me show you the exact calculation.
The Hashrate That Would Be Required
We have to start with a baseline. During Q2 2024, the Bitcoin network was running at approximately 600 exahashes per second. The average BTC price was somewhere near $65,000 — call it $64,000 to $66,000 if you want a range. Daily issuance, after the April halving, was 144 blocks multiplied by 3.125 BTC per block. That gives roughly 450 BTC mined per day across the entire network.
Now divide by hashrate. One exahash per second captures about 0.75 BTC per day at 600 EH/s network-wide. At $65,000 per BTC, that’s $48,750 per exahash per day. Multiply by 91 days in a quarter, and one EH/s of self-mining power produces approximately $4.44 million in revenue.
Cango reported $47.4 million. Divide $47.4 million by $4.44 million per EH/s, and the implied hashrate is 10.7 EH/s.
Let that sink in. Ten point seven exahashes.
That would put Cango — a company that was selling auto loans and insurance products in China three years ago — in the same hashrate league as Riot Platforms, one of the most established US miners. Riot generated roughly $70-80 million in Q2 revenue with a fleet of more than 10 EH/s. Cango claims $47.4 million on the same implied compute.
Is that possible? Maybe. But here is what jumps out at me as a forensic analyst: the earlier dissection of this data suggested an estimate of only 500-800 PH/s. That is off by roughly 15x. The gap between those two numbers is not a rounding error. It is a signal that the reported revenue is either:
- Self-mining revenue plus hosting, colocation, equipment sales, or energy resale;
- Revenue from an acquisition completed mid-quarter, which would skew the average;
- Or actual hashrate that has been massively underreported to the market.
I have seen this exact pattern before. In 2021, I investigated NFT projects that reported enormous trading volumes. When I compared the reported volume to on-chain transfers, the discrepancy revealed wash trading. I am applying the same discipline here. A mining company that does not disclose its hashrate in the same report where it announces revenue is asking the market to take the revenue on faith.
Faith is not a portfolio strategy.
The Company Nobody Expected to See Here
Let me give credit where it is due. Cango was never supposed to be a miner. The company spent years as a Chinese auto-financing platform. It went public on the NYSE, built a consumer lending business, and then watched that business get crushed by regulatory shifts and market saturation. Management did what distressed companies do: they looked for a new narrative.
Bitcoin mining was the first pivot. In 2023 and early 2024, Cango began buying mining infrastructure. The pace was aggressive. By Q2 2024, they were generating enough BTC sales to print a $47.4 million quarterly revenue line. That is not a shell company playing at crypto. That is committed capital.
But here is the uncomfortable part. Cango’s executive team built careers in consumer finance, not in power markets, not in ASIC procurement, and not in data-center engineering. Mining is not a software business. It is an energy logistics business. A mining company lives or dies on its all-in cost per terahash, its uptime, its power curtailment agreements, and its ability to source machines before the next difficulty adjustment.
Running a mining business with a consumer finance mindset is like piloting a cargo ship with a race-car team. You might have the throttle. You don’t have the navigation.
And now Cango says it is pushing deeper into the hottest narrative in the sector: energy plus AI compute. This is the exact same language being used by Core Scientific, Hut 8, TeraWulf, IREN, and Marathon. Every miner wants to be an AI data-center provider. The market rewards that pivot with premium multiples. But the pivot comes with a completely different set of risks.
GPU clusters require different power density than ASIC containers. They require liquid cooling. They require multi-year client contracts with penalty clauses. They require security standards. They require a sales team that can talk to enterprises, not just crypto exchanges. And they require enormous capital expenditure before the first AI dollar lands.
Cango has not disclosed a single AI customer. No GPU count. No megawatt commitment. No revenue contribution. No mention of who is buying the compute. That silence matters more than the $47.4 million number.
The Revenue Is Real. The Profit Is Unknown.
Now let’s talk about the actual economics. $47.4 million in revenue tells you almost nothing about whether Cango is profitable. Mining revenue is a top-line commodity number. The cost side is proprietary and it is everything.
If Cango has locked in cheap power — say, below $0.04 per kilowatt-hour — then $47.4 million in revenue could translate into healthy cash flow. If they are paying market rates for hosting at third-party facilities, they could be barely breaking even, or losing money on every bitcoin they mine.
The original analysis flagged this as “N/A - insufficient information.” That is a diplomatic way of saying the market is flying blind.
Let me give you a framework. At a $65,000 average BTC price and 3.125 BTC per block, a mining operation needs an all-in cost below roughly $0.06 per kWh in most efficient ASIC configurations to generate meaningful margins. Many US miners run at costs between $0.04 and $0.08. China-based or hydropower-based operations can get lower. Cango’s ability to source cheap energy is the single most important variable in their P&L.
The source material speculated that Cango might have an electricity cost below $0.05 per kWh. If true, that is a genuine structural advantage. If not, the entire AI pivot is just a marketing deck.
Here is the first actionable signal for anyone watching this name: demand the cost curve. In the next conference call, ask about average power price, fleet efficiency in joules per terahash, and curtailment policy. If management dodges, treat the stock as a speculative vehicle, not an operating business.
The AI Compute Pivot: Real Option or Theater?
This brings us to strategy. Cango says it is building an “energy and AI computing platform.” The phrase is carefully chosen. It avoids committing to a particular business model. That can mean anything.
Let me explain what this strategy actually requires.
AI compute platforms are not just mining rigs with a different sticker. They need:
- GPU servers, usually NVIDIA H100s or H200s, with long lead times and escalating prices.
- Data centers with high-density power infrastructure, typically 30kW to 100kW per rack, versus 5kW to 10kW for traditional colocation.
- Liquid cooling or advanced air-cooling retrofits.
- Redundant fiber connectivity and low-latency networking.
- A sales pipeline into AI labs, model inference companies, or enterprise customers.
- Long-duration power purchase agreements, because AI hosting contracts demand uptime certainty.
A Bitcoin mining facility has three of these: power, physical security, and 24/7 operational staff. It does not have the GPU supply chain, the enterprise sales motion, or the cooling density.
This is not impossible. Core Scientific proved the playbook. They repurposed existing mining sites, signed a massive hosting deal with CoreWeave, and transformed their market narrative. TeraWulf did the same on a smaller scale. Hut 8 is mid-transition. IREN is building purpose-built GPU capacity.
But every company that executes this transition successfully does it the same way: they announce a named customer, a committed megawatt allocation, and a timeline. Cango has announced none of that.
The market is pricing Cango for an AI pivot that has not been delivered. That creates downside risk. It also creates an asymmetric opportunity if management actually starts closing contracts.
Where Does Cango Fit in the Pecking Order?
Let’s place Cango in the competitive landscape.
Marathon Digital reported roughly $145 million in Q2 2024 revenue. Riot generated $70-80 million. CleanSpark did somewhere near $60-70 million. Cango, at $47.4 million, is not a micro-cap joke. It is a second-tier miner with first-tier revenue. Yet its market cap is a fraction of Marathon’s or Riot’s.
Why? Because the market distrusts the transition. It does not understand the person leading it. It does not see the hashrate. And it worries about opacity around China-linked capital.
That is the discount. And that is the arbitrage.
If Cango were a Texas-based miner with the exact same revenue, it would trade at a much higher multiple. The same business, the same cash flows, the same narrative — but no “China discount” and no governance skepticism. The gap between where Cango trades and where a Western miner with the same fundamentals would trade is an arbitrage window.
Arbitrage window closing in 10 minutes.
Not literally. But this is the most important lesson: in public markets, perception is a component of valuation. Cango’s perception problem is real. It is also correctable. Every announcement of an AI client or an audited hashrate reconciliation will compress that discount. The question is whether you have the risk tolerance to wait.
The Contrarian Blind Spot: Power Is the Asset
Here is the angle almost everyone misses.
The bear case is easy: Cango is a former auto lender with no mining history, a vague AI story, and a stock that can get crushed by BTC volatility. Fine. I can build that short thesis in my sleep.
But the bull case is not about mining revenue. It is about what Cango might own underneath the revenue.
Imagine a company that acquired real estate, power substations, interconnection agreements, and long-term electricity contracts during a distressed period. Then imagine that same company starts to generate $47.4 million in quarterly revenue. Now the valuation should be anchored not to the revenue multiple, but to the replacement cost of the physical assets.
Power infrastructure is scarce. In the United States, getting a new substation or a new grid interconnection can take years. Bitcoin miners who bought those assets early are sitting on a call option for AI data-center demand. Cango, if it has secured even modest power rights, holds a similar call option. The market is treating that option as worthless right now.
There is another contrarian layer. The source material noted that Cango’s team is inexperienced in mining. That is usually a negative. But in this industry, the valuation is often tied to capital access, not operational skill. Cango has access to US public markets. It can raise equity. It can issue convertible debt. It can acquire private mining fleets without paying cash.
That financial engineering capability is exactly how a former consumer finance company buys its way into the mining sector. It may not be elegant. But it can work.
Do not mistake a boring balance sheet for a boring outcome.
What Actually Moves This Stock Next?
Now let’s talk about catalysts. I wouldn’t spend too much time trying to predict Bitcoin price. Bitcoin is a macro trade. Cango is a micro stock with idiosyncratic drivers.
The first catalyst is the next 10-Q. Look for a line item labeled “AI computing services” or “Data center services.” If that appears, the story changes overnight. But better yet, look for a footnote. A footnote that says “company has entered into a hosting agreement with a third party for GPU deployment” is worth more than a thousand headline articles.
The second catalyst is capital raising. If Cango announces an equity offering to fund GPU purchases, the stock may initially drop on dilution. That will be the moment when long-term investors can build a position. If the offering is oversubscribed and management names a strategic partner, the drop becomes a launchpad.
The third catalyst is hashrate disclosure. Cango needs to do a better job of telling the market how much mining capacity it controls. If the next filing includes a “hashrate as of date” figure, investors can finally validate whether $47.4 million makes sense. If that number comes in far below the implied 10.7 EH/s, the revenue must be coming from somewhere else. That is not automatically bad, but it raises questions about sustainability.
The fourth catalyst is Bitcoin leverage. Mining stocks are high-beta plays. If BTC enters a sustained uptrend, Cango will rally harder than Bitcoin. If BTC drops, the stock will get destroyed. In the current sideways market, this stock is a coiled position waiting for direction.
The Risk Matrix No One Wants to Read
Let me be explicit about the risks. If you are going to touch Cango, you need to know what can kill it.
First, Bitcoin price. This is the largest risk. Cango’s mining revenue is priced in BTC. If Bitcoin falls from current levels to, say, $40,000, the mining economics deteriorate sharply. At $40,000, many miners with average cost structures are underwater. Cango’s cash flow disappears and the AI pivot cannot save a company bleeding money on its core operation.
Second, difficulty. Bitcoin network difficulty has been climbing as new miners flood into the market. Every difficulty increase squeezes margins. Cango must constantly upgrade its machine fleet just to maintain the same BTC output. That is a capex treadmill.
Third, execution risk. Building an AI compute platform is not a side project. It is a new company inside an old company. Cango must hire engineers, buy GPUs, sign contracts, and manage a completely different customer base. All of this while the stock trades as a crypto proxy. The probability of delay is high.
Fourth, dilution. I expect Cango to raise capital in the next two to four quarters. Shareholders will be diluted. If management raises at a low price, the damage is permanent. This is the sneakiest risk in the entire thesis.
Fifth, governance and opacity. The market still doesn’t have full confidence in this management team. A single accounting restatement or a missed filing would destroy the stock. This company needs to operate with the transparency of a Western bank and the speed of a startup. That is a hard combination.
Liquidation pending. Don’t mistake quarterly revenue for sustainable cash flow.
The Takeaway: Watch the Footnotes, Not the Headlines
Here is where we land.
Cango’s Q2 mining revenue of $47.4 million is a real number. It proves the company can operate mining equipment and generate revenue. But it does not prove profitability. It does not prove a competitive hashrate. It does not prove AI compute capability.
The opportunity lies in the discrepancy. The revenue implies a much larger mining footprint than the management has ever disclosed. Either the company is hiding something, or the market is underpricing the assets. My forensic bias tells me to assume the first, but my financial instinct sees the second.
The next two quarters will determine which one is true. Wait for the AI revenue line. Wait for a customer name. Wait for a hashrate disclosure. If those appear, the current stock price will look cheap in hindsight. If they never appear, you will be glad you did not pay up for a story.
Alpha detected. Position established.
But position size accordingly. This is a second-tier miner with a first-tier narrative, a whole lot of tail risk, and a market that still cannot decide if Cango is a crypto miner, a power company, or a former car lender with a good line of financial engineering skills. The answer will be written in the next 10-Q.
That filing is the only chart that matters.
