On the evening of October 27, bitcoin block 973,413 was mined, paying out 3.14 BTC. At the price at the time, that was just over $250,000, or roughly 1.8 million yuan.

Public data from the CKPool mining pool shows the block was found in solo mode — no pooling with the crowd for small, steady payouts, just one machine taking its chances alone and, having won the bet, keeping the entire reward.

The winning machine runs at only about 6.5 TH/s. Based on that, rig seller Altair Tech reckons it was "most likely" a Canaan Avalon Nano 3S: router-sized, 140 watts, quieter on a desk than a laptop fan, and available online for two to three hundred dollars.

For comparison, a current industrial-grade S21 XP delivers 270 TH/s. The little box has less than a fortieth of that.

Small Boxes Keep Winning, and Not for the First Time This Year

On a normal day, bitcoin hands out 144 block rewards, and the vast majority go to pools like Foundry, AntPool and F2Pool, which run warehouses full of tens of thousands of machines. A few-hundred-dollar box beating those operations is about as likely as winning first prize on a two-yuan lottery ticket.

This week, though, the long shots came in clusters. The day before the $250,000 win, on October 26, block 973,262 was mined solo by a FutureBit Apollo Solo, a device from a small US brand, for 3.154 BTC — the fourth block in that brand's history. Mining circles say this too was a retail solo win, though the page only shows "Braiins Pool," so it can't be fully verified. Treat it as unconfirmed.

Lucky hits like these have happened before this year. On May 30, another Nano 3S took $232,000 — a $300 secondhand rig that found a $230,000 block. On July 10, a $150 open-source box, a Bitaxe Gamma, walked off with $200,000.

According to solo mining data, retail miners going it alone found 24 blocks worldwide over the past 12 months, up 40% from the year before. That works out to a $200,000-plus "lottery ticket" being cashed roughly every two weeks.

The Odds Are Worse Than Getting Struck by Lightning

The math is simple: your machine's hashrate divided by the combined hashrate of every rig on earth equals your chance of taking the next reward. Bitcoin draws roughly every 10 minutes, so 144 chances a day.

Take the Nano 3S that just won. A single unit does 6 TH/s. Assume a conservative network total of 800 EH/s, or 800 million TH/s. Six divided by 800 million is 0.0000000075. In other words, each 10-minute draw carries odds of 0.75 in 100 million; multiply by 144 draws a day and the daily chance comes to 1.08 in a million.

And 800 EH/s is a generous denominator. Working backward from the current mining difficulty of 140.58T, the real network hashrate sits around 1,000 EH/s, with some data sources showing intraday readings as high as 1,080. A bigger denominator trims the odds by another fifth: roughly 0.86 in a million per day, or about one in 3,000 per year.

An infographic circulating in mining circles pegs the October 27 machine's annual odds of finding a block at about 1 in 1,500. Rerun with 6.5 TH/s, the real figure is closer to 1 in 2,900, and Altair Tech's own calculator on its website puts it at once every 3,200 years.

Why is the figure on the graphic twice as optimistic? A user on Stacker News dug through CKPool's stats page and found that the winning address actually had five small machines attached to it, with News.bitcoin reporting a combined 16.2 TH/s. Run the numbers on 16.2 TH/s and the annual odds of finding a block come to roughly 1 in 1,200 to 1 in 1,500 — considerably better than a single unit. But only one of those machines actually hit.

Leave one Nano 3S plugged in and running nonstop, and on average its turn comes around once every three thousand years or so.

It Loses Money on Power, But in the Nordics It's a Space Heater

Skip the once-in-three-thousand-years jackpot and plug the little machine into a big pool for steady, grinding income, and the economics are straight underwater.

At current pool payouts of about $36 per PH per day, a 6.5 TH/s machine earns $0.23 a day. But running 140 watts for a day consumes 3.36 kWh, and at a US household rate of $0.10 per kWh, that's $0.34 in electricity. Mine honestly for a day and you're out about ten cents.

Run the numbers that way, though, and you've misread the product entirely. To buyers, home solo miners and industrial rigs are two different species. Industrial machines exist to make money at scale, judged on efficiency and payback period. Home units aren't selling steady returns at all — the Nano series runs at just 33 to 40 decibels, sits in a bedroom like an ornament, and works out of the box. What it sells is the fun of buying a lottery ticket, plus a bit of heat.

The people actually paying for these boxes fall into two camps, neither of which cares about payback.

One is bitcoin diehards who don't want hashrate monopolized by industrial farms and are happy to spend a couple hundred dollars contributing to the network for a sense of participation. The other is households in cold, high-latitude parts of Europe and North America that are running the heating anyway. Feed 140 watts into an ordinary electric heater and you get 140 watts of heat; feed it into one of these miners and you get the same 140 watts of heat. The only difference is that while it's warming the room, the rig also runs a few trillion hashes and takes 144 free swings at a bitcoin block.

One Nordic user is a textbook case: his home needs heat all winter, so he bought several Nano 3S units and set them up in the living room and study as supplementary heaters, mining in a pool most of the time and occasionally switching to solo for fun. For him, the "ten cents a day in electricity" math simply doesn't apply — that power was going into heating regardless, none of the heat is wasted, and the mining, along with the once-in-three-thousand-years shot at $250,000, is a free bonus.

Big Farms Are Walking Away, So Rig Makers Are Courting Retail

Rig makers used to cater only to free-spending industrial farm operators. Why the sudden effort to charm ordinary consumers and put $300 "heating lottery machines" in living rooms?

The answer is simple: out in the professional mining world, the big farms are barely hanging on.

This October, costs pushed industrial miners into a corner. On October 26, bitcoin fell below $80,000 intraday, bottoming out near $79,300, even as mining difficulty rose 5.92% on October 16 to 140.58T, with another increase of around 5% expected around October 29. Daily revenue per PH of hashrate has slid from $41 a month ago to $36. On-chain transaction fees, the tips that go to miners, are shrinking too: Glassnode data shows on-chain transfer volume rose 26% last week to $5.5 billion, while miner fee income fell 8.8% to just $185,000.

Revenue is falling and costs are punishing. CoinShares calculates that listed miners' cash cost to produce one coin ran as high as $75,500 in the second quarter. With bitcoin at $80,000, that leaves a few thousand dollars of gross margin per coin, before machine depreciation.

With mining unprofitable, the big operators are switching off and moving on. In its October 22 weekly report, CryptoQuant noted that five US-listed miners — IREN, Core Scientific, Cipher, TeraWulf and Bitfarms — mined just 7,075 bitcoin between January 1 and October 19 this year, nearly half the 13,263 they mined in the same period last year. Because electricity earns roughly three times more running AI than mining ($1.5 million per megawatt per year versus $500,000), they have collectively signed more than $50 billion in AI leases, and are even pulling out some $1.5 billion worth of old rigs to dump on the secondhand market. Life isn't easy for these converts either: facing a $50 billion funding gap on AI buildouts against annualized revenue of only $1.1 billion, they were targeted by Wall Street short sellers on October 22, with share prices falling between 8% and 11%.

Even Binance Pool, a veteran operation that launched in April 2020 and has issued $500 million in loans, has announced it will shut down entirely on November 10.

With their biggest customers either powering down and pivoting or dumping used machines, upstream rig makers have had to find new business fast. Beyond investing in AI projects themselves — on October 27, Canaan and MicroBT joined a $16 million AI computing power round for Luxor, and Tether has lined up Bitmain, Canaan and 13 other companies to commit to building version 1.0 of the open-source MDK toolkit by year-end — selling small rigs into ordinary living rooms has become the most practical option available.

A Kilowatt-Hour Is a Cost on the Farm and a Lottery Ticket in the Bedroom

Look back at the week: listed giants commanding hundreds of megawatts ran the numbers with great precision, decided mining paid too little, and busied themselves switching off machines to chase the next thing. Meanwhile an ordinary person in a living room, warm air blowing from a $300 box, quietly walked off with a $250,000 bitcoin jackpot.

The same kilowatt-hour that industrial farms now spurn for failing to clear a margin turned into the most enviable lottery ticket going in somebody's bedroom and study.