Note that Heatbit's launch material for Bitwater contains a price, a chip count, and no wattage. Eight hundred ninety-nine dollars. Twenty mining chips. The unit replaces the resistive element inside a household hot water tank and mines bitcoin with the same electricity that heats the water. That is the entire claim.
There is no hashrate in TH/s. No efficiency in J/TH. No chip model. No tank compatibility list. No UL or CE listing number. No noise figure. No pool configuration method. No warranty terms. No measured yield. Six qualitative data points, a single source, and the source is the seller.
I have spent enough years reading primary documents to know what that pattern means. In 2017 I hand-audited 45 ICO-era contracts and found three reentrancy bugs that would have cost roughly $2 million in user funds; the projects with clean code published it, and the projects with dirty code published a roadmap. Hardware behaves identically. Bitaxe-class home miners ship full spec sheets and third-party measurements because their numbers survive scrutiny. When a consumer device arrives with a price and a philosophy but no datasheet, the missing number is usually the unfavorable one.
That is the finding. Everything below is the arithmetic that makes the silence expensive.
What Bitwater actually is
Strip the marketing and the object is a retrofit component. You drain the tank, remove the resistive element, thread in a housing that contains 20 ASIC chips, and wire it to the same circuit. Heatbit's previous product, the One, was a space heater built on the same premise. Bitwater is the second iteration of one idea: treat the miner as a resistor that happens to emit a paycheck.
The physics is real, and I will defend it before I criticize it. An ASIC converts essentially all of its input power into heat, minus a rounding error that leaves as fan motion and electromagnetic radiation. At the level of the first law of thermodynamics, mining while heating water loses nothing. Anyone who calls the concept a scam on energy grounds is wrong, and I have watched that argument get recycled for eight years by people who never once opened a textbook.
The claim that survives is narrower: the electricity was going to be spent anyway, the heat was going to be produced anyway, so the bitcoin is marginal revenue on a load you already carry. That is coherent. It is also coherent against exactly one baseline, and that is where the vendor's framing begins to leak. The competing product is not another miner. It is a water heater β and, in most American homes, a plumber who charges $200 to $500 to install a retrofit element. The real entry cost is closer to $1,100 to $1,400, which is not the number on the page.
Four problems the press release does not address
The coefficient of performance is locked at one. Resistive heating, whether the resistor is nichrome or silicon, returns 1 kWh of heat per 1 kWh of electricity. An air-source heat pump water heater returns 3 to 4. This is not a footnote; it is the entire comparison. On the heating function alone, Bitwater is three to four times less efficient than the best appliance in its own category. The mining revenue is not a bonus layered on a good heater. It is compensation for installing a worse one.
Water and electricity now share an enclosure. A resistive element is a bare metal loop designed to sit in water. An ASIC must be electrically isolated, thermally bonded and temperature-controlled. The engineering answer is almost certainly a dry well with a thermal interface, and that introduces thermal resistance, scale buildup on the wet side, and a failure mode in which the barrier between mains voltage and sixty gallons of conductive fluid degrades quietly. Water damage is already among the leading causes of household insurance claims. Adding silicon to the tank changes who is liable and probably what your policy covers. No certification has been published.
The power budget does not match the appliance. A typical residential storage tank runs a 3 to 4.5 kW element. Twenty contemporary chips at around 2.5 TH/s and 18 J/TH draw roughly 900 W, about a quarter of the original heating power. Recovery time after a shower stretches, effective capacity shrinks, and most buyers will keep the old element as a backup β which quietly deletes the "one device, two jobs" pitch and returns you to a tank with an expensive accessory bolted to it.
The lifetimes do not line up. A tank is an 8-to-15-year appliance. An ASIC's economic life is 2 to 4 years before difficulty erodes it, and its physical life is bounded by thermal cycling. Somebody has to decide whether the chips are field-serviceable, or whether a dead board means a new element, or a new tank. The document does not say. In hardware, an unanswered serviceability question is a cost you have not been quoted.
The numbers, stated as assumptions
Here is my baseline, built from public chip specifications rather than vendor data. Treat it as an order-of-magnitude frame, not a measurement. Conservative: 1 TH/s at 30 J/TH across 20 chips gives about 20 TH/s at roughly 600 W. Neutral: 2.5 TH/s at 18 J/TH gives about 50 TH/s at about 900 W. Optimistic: 5 TH/s at 18 J/TH gives about 100 TH/s at 1,800 W, which would need aggressive cooling and would punish both noise and lifespan.
Take the neutral case, in a sideways tape where hashprice has compressed and every joule of inefficiency shows up directly in payback. At 144 blocks a day and a 3.125 BTC subsidy, the network issues about 450 BTC daily. At roughly $100,000 that is about $45 million of revenue spread across roughly 900 EH/s, or about $0.05 per TH per day. Fifty terahashes is therefore about $2.50 a day at full load.
Full load is the fiction. A household consumes 8 to 11 kWh of hot water daily; the unit produces 21.6 kWh if it runs around the clock. The thermostat sets the duty cycle, and it lands near 40 to 50 percent, lower in summer when the tank loses less heat to ambient. Real gross revenue is roughly $1.00 to $1.25 a day. Against an $840 premium over an ordinary element, payback lands at 22 to 28 months β before difficulty growth and before the 2028 halving cuts the subsidy to 1.5625 BTC and halves the revenue of any unit still in service.
The sensitivity is lopsided. Double the bitcoin price and payback compresses to 11 to 14 months. Let network difficulty compound at 40 percent a year and payback extends 30 to 50 percent, which in the tail means it never closes at all. Electricity price, the variable everyone argues about, barely moves the result.
There is one more check worth running, and it is the fairest thing I can say about the price. At $899 for a neutral 50 TH/s, the unit costs about $18 per terahash. New-generation machines price near $20 to $25 per terahash. If the neutral case is real, the buyer is paying roughly market rate for hashrate and receiving the heating function nearly free. That is a defensible design. It stands or falls entirely on whether 50 TH/s is real β which is precisely the number the vendor declined to publish.
The argument the vendor gets right, and the one it gets wrong
Here is the insight most commentary misses. Because ASIC heating and resistive heating both run at a coefficient of performance of one, the electricity bill is identical either way. Mining adds no cost to the load. Mined bitcoin is therefore a rebate on a bill you were already paying, not a revenue stream that has to cover power. The standard objection β at thirty cents a kilowatt-hour this never pays β is simply wrong, and I have seen it repeated by people who should know better.
But the same identity destroys the marketing. The correct baseline is not a resistive element. It is a heat pump or a gas burner. Against a heat pump, every kilowatt-hour this device consumes buys a quarter of the heat it could have bought. The product's economics are almost completely insensitive to electricity price and extremely sensitive to duty cycle and hashprice β the exact opposite of what both its critics and its sellers assume. That also confines the addressable market to households already heating water with resistance electricity and already paying a high enough price per kilowatt-hour for the rebate to matter. Everyone else is buying a novelty.
One thing deserves a straight answer. There is no token here. No allocation, no points program, no governance rights, no airdrop. Revenue comes from real block subsidies, an external cash flow, not from later entrants. This is a hardware business settling in bitcoin, and it is not a Ponzi structure. I will say that plainly, because the absence of a token is the most honest feature in the entire package.
Where the retail read and the trade read diverge
Retail reads the headline and sees free bitcoin from hot water. The trade reads the same sentence and prices a resistive heater carrying an $840 premium and no datasheet.
But the second-order observation is where the real value sits, and it is not cryptographic. Moving from air heating to water heating is the smartest decision Heatbit has made. Space heating is seasonal; a heater-miner idles through summer and drags annual utilization toward 25 percent. Hot water is a 365-day load. Utilization roughly doubles. That is an engineering-and-marketing insight, not a physics one, and it is the reason this iteration deserves analysis at all rather than dismissal.
Then there is the part nobody puts in the brochure. The vendor retains the firmware. Whether the device can be remotely throttled, region-locked or bricked is unspecified, and that is the same structural problem I have written about in DAO governance: the upgrade key always sits with a small group of admins, and "code is law" describes the deployed artifact, not the authority standing behind it. The code does not lie, but it can be misunderstood. You do not own this device's behavior. You license it, at the seller's discretion, for as long as the seller exists. Pool configuration method matters for the same reason β if hashrate is mediated through vendor cloud infrastructure, you have added a counterparty to a heating appliance.
That is also a delivery question rather than an ideological one. Heatbit shipped a prior generation, and consumer mining hardware has a long documented record of slipped schedules and measured performance below claimed figures. Trust is earned in drops and lost in buckets, and a spec sheet is the cheapest drop available to any vendor.
What to watch, and the threshold I would use
Three disclosures would settle most of this: J/TH at the rated wall draw, a UL or CE listing number, and a field-serviceability statement. Until then, run the payback yourself at your own duty cycle, not the vendor's.
My decision rule is blunt. Below an 18-month premium recovery, the bitcoin is a genuine rebate and the heater is close to free. Between 18 and 24 months, you are buying a hobby with a small return. Above 24 months, you are buying an appliance, and you should price it like one β which means asking why it costs eight times an element that heats water just as well. On my numbers, clearing 18 months requires either more than 65 TH/s from the unit or a price near $500.
In the silence of the dip, the weak hands break. Vendors go quiet about wattage for the same reason traders go quiet about drawdown.
