> ## Content Index > Fetch the complete content index at: https://www.siliconsnark.com/llms.txt > Use this file to discover other available public pages before exploring further. # Deep Dive: Crypto Mining Rigs Still Make Money. Mostly for People Who Own a Power Plant. - URL: https://www.siliconsnark.com/deep-dcrypto-mining-rigs-still-make-money-mostly-for-people-who-own-a-power-plant/ - Published: 2026-08-21T00:25:54.000Z - Updated: 2026-08-21T00:25:54.000Z - Description: Crypto mining rigs still run in 2026, but profit depends on cheap power, efficient ASICs, and honest math. Yes, meme-coin mining is real. - Author: CircuitSmith - Tags: Crypto, Deep Dive, Bitcoin, Dogecoin, Mining, Fintech Somewhere in 2026, a machine the size of an uncooperative carry-on is inhaling 3,500 watts, screaming like a leaf blower trapped in a server rack, and earning its owner roughly the price of a convenience-store sandwich before depreciation. This machine is considered efficient. Crypto mining did not die. It industrialized, consolidated, merged with the energy business, began flirting aggressively with artificial intelligence, and left behind a smaller consumer hobby populated by tinkerers, speculators, heat-reuse enthusiasts, and people who enjoy turning a residential electrical panel into a capital-allocation committee. Yes, people still use mining rigs. A lot of them. MARA said it owned roughly [495,000 rigs as of March 31, 2026](https://www.sec.gov/Archives/edgar/data/1507605/000150760526000016/mara-20260331.htm?ref=siliconsnark.com). Bitdeer reported 262,000 under management. The Bitcoin network was still operating around the neighborhood of 900 exahashes per second this summer, an amount of specialized computation so large that writing it out mostly produces zeroes and emotional fatigue. Yes, some miners make money. The best operators can still produce a positive cash margin because they buy electricity at rates homeowners encounter only in historical fiction, negotiate grid credits, operate the newest application-specific integrated circuits, and spread labor, cooling, repair, networking, and land costs across warehouses full of machines. Also yes, meme-coin mining is a thing. Dogecoin is one of the most important proof-of-work mining economies outside Bitcoin. It is commonly mined alongside Litecoin using the same Scrypt ASIC and the same computational work. A newer Pepecoin even advertises triple merged mining with Litecoin and Dogecoin, because the frog economy has learned synergy. But no, you cannot natively mine the existing SiliconSnark coin. [$SNARK was launched on Solana through Pump.fun](https://www.siliconsnark.com/siliconsnark-meme-coin-hits-800k-proves-the-internet-loves-tech-sarcasm/). Solana uses proof of stake, not competitive proof-of-work mining. A “SiliconSnark $SNARK mining rig” would therefore have to mine some other coin and pay users in $SNARK, or launch an entirely new blockchain, or sit on a desk making reassuring fan noises while a small LCD displays the word *WAGMI*. That last option may be the most honest consumer crypto product ever made. ## The Mining Rig Did Not Die. It Lost Its Bedroom Privileges. The popular memory of crypto mining is stuck somewhere between 2017 and 2021: open-air frames packed with graphics cards, colored PCIe risers, box fans, and one friend explaining that the heat was “basically free” because the coins would pay for everything. This was the era when a gaming GPU could be both a graphics product and a tiny industrial asset, irritating gamers and delighting anyone who had discovered the phrase passive income before discovering utility bills. Ethereum ended that particular house party. Its September 2022 Merge switched the network from proof of work to proof of stake. As [Ethereum’s own documentation says](https://ethereum.org/roadmap/merge/?ref=siliconsnark.com), mining is no longer a valid way to produce blocks on Ethereum. The enormous market that had absorbed general-purpose GPU hashing disappeared from under the rigs. GPUs can still mine proof-of-work coins such as Ethereum Classic, Ravencoin, Ergo, Flux, and others. Some operators switch between them automatically. Some “spec mine” a young coin at little or no present profit in the hope that it later appreciates. And some already own the cards, have very cheap power, need the heat, or simply like participating in a network. But the broad consumer proposition—buy six new GPUs, point them at a major liquid coin, recover the capital, then enjoy a stream of money—never found a replacement of Ethereum’s scale. Bitcoin, meanwhile, has long belonged to ASICs. An ASIC is a chip built to perform one narrow job extremely well. A Bitcoin ASIC does SHA-256 hashing. It does not render games, train a model, edit a podcast, or become a tasteful media server when mining stops paying. Its fallback career is generally “used Bitcoin ASIC.” This is a labor market with limited mobility. The industrial machines kept improving anyway. Bitmain’s air-cooled Antminer S23 is rated around 318 terahashes per second at about 3,498 watts, or 11 joules per terahash. The hydro-cooled S23 Hyd pushes roughly 580 terahashes at 5,510 watts, or 9.5 joules per terahash, according to the company’s [product manual](https://file12.bitmain.com/shop-product-s3/firmware/807d3b27-f625-470f-a940-247f83b36854/2025/06/20/14/S23%20Hyd.%20Product%20Manual%5Fv1.0.6.pdf?ref=siliconsnark.com). “Hydro” is not a decorative gamer word. The machine expects facility plumbing that carries heat away with liquid. Your spare-bedroom oscillating fan has not been promoted to this role. That is the current mining market in miniature. The frontier rig is not really a personal computer. It is a replaceable production unit in an energy-and-cooling system. The silicon matters, but the site is the business. ## The Number That Explains the Entire Misery Is Hashprice Mining profitability is often presented as a colorful calculator with green daily earnings. The more useful starting point is *hashprice*: the revenue a miner can expect per unit of Bitcoin hashrate over time, usually expressed as dollars per petahash per second per day. Hashprice compresses several moving parts into one grim little number: - Bitcoin’s dollar price; - the block subsidy, currently 3.125 BTC after the 2024 halving; - transaction fees; - network difficulty and competing hashrate; - the luck and fee structure of the miner’s pool. If Bitcoin rises while competition stays flat, dollar hashprice tends to improve. If thousands of new machines join, each existing machine owns a smaller expected slice of the rewards. If transaction fees are sleepy, miners cannot count on them to rescue the subsidy. Every four years, the subsidy halves, which is Bitcoin’s elegant way of scheduling an industry-wide margin crisis in advance. In early August, Luxor’s [Hashrate Index put spot hashprice near $32.10 per petahash per day](https://hashrateindex.com/blog/hashrate-index-roundup-august-3-2026/?ref=siliconsnark.com), with the 30-day average around $31.54\. CoinShares described a broader 2026 squeeze in which hashprice fell through the high-$30s and toward the high-$20s, near or below breakeven for many fleets. This is the commodity revenue line. No amount of RGB lighting can negotiate with it. Take the air-cooled S23 and use $31.20 as a round late-summer hashprice snapshot. Its 0.318 petahashes would generate about $9.92 a day before pool fees. The machine consumes about 83.95 kilowatt-hours every 24 hours. | Electricity rate | Daily power cost | Daily margin before pool fees, cooling, labor, repairs, and hardware | | ---------------- | ---------------- | -------------------------------------------------------------------- | | $0.04/kWh | $3.36 | $6.56 | | $0.06/kWh | $5.04 | $4.88 | | $0.10/kWh | $8.40 | $1.52 | | $0.15/kWh | $12.59 | \-$2.67 | At an estimated market price around $7,600, even the $0.06-power scenario implies a simple hardware payback of more than four years if hashprice, uptime, and performance magically freeze. They will not. Difficulty changes. Bitcoin moves. Fans fail. Pool fees exist. Heat must leave the building. The 2028 halving approaches with the calendar discipline of a dentist appointment. The hydro machine looks stronger per watt but asks more of the site. At 580 terahashes and the same hashprice, it grosses about $18.10 daily while consuming 132.24 kilowatt-hours. Power costs $7.93 a day at six cents, leaving roughly $10.17 before everything else. At ten cents, that shrinks to $4.88\. If the unit costs roughly $12,000 to $15,000, “payback period” becomes a long-term relationship with several variables that have blocked your number. This is why a live profitability screenshot is not an investment case. It is weather. Useful weather, certainly. But financing a machine from one day’s weather is how you end up owning a highly optimized metal complaint. ## Who Makes Money? People Whose Electricity Rate Looks Like a Typo. Cambridge’s 2025 survey of 49 mining firms, representing nearly half the Bitcoin network’s hashrate at the time, found that electricity accounted for more than 80% of miners’ cash operating expenses. The [median reported electricity-only cost was $45 per megawatt-hour](https://www.jbs.cam.ac.uk/faculty-research/centres/alternative-finance/publications/cambridge-digital-mining-industry-report/?ref=siliconsnark.com), equivalent to 4.5 cents per kilowatt-hour, while the median all-in power cost was $55.50 per megawatt-hour. That is the neighborhood where serious mining lives. The average residential retail rate in many developed markets is not in that neighborhood. It is across town, behind a gate, wondering why the garage smells warm. The profitable operators generally combine several advantages: - **Cheap energy:** wholesale contracts, owned generation, stranded or curtailed supply, behind-the-meter arrangements, or locations where producers value a flexible buyer; - **Efficient machines:** low joules per terahash delay the moment a rig’s revenue falls below its electricity cost; - **High uptime:** a machine earns nothing during repairs, network failures, delayed deployment, or summer shutdowns; - **Power-market participation:** some large miners earn curtailment credits by shutting down when a grid needs capacity back; - **Scale:** warehouses spread technicians, monitoring, security, spares, land, transformers, and network connections across more hashrate; - **Capital discipline:** buying machines cheaply matters almost as much as running them cheaply; - **A treasury decision:** selling mined coins immediately produces a different business from holding them as a leveraged bet on future prices. Riot Platforms offers a near-perfect lesson in how the accounting changes the answer. In the first quarter of 2026 it reported $111.9 million of Bitcoin mining revenue. Its direct self-mining costs, net of $21 million in power curtailment credits and excluding miner depreciation, were about $65.7 million. That is a healthy-looking cash production spread. Then the rigs remembered they were assets with useful lives. Including $76.1 million of miner depreciation, Riot calculated total mining cost at $141.8 million. Its disclosed [cost per Bitcoin was $44,629 before miner depreciation and $96,283 after it](https://www.sec.gov/Archives/edgar/data/1167419/000110465926053120/riot-20260331x10q.htm?ref=siliconsnark.com), versus a production value of $75,964 per Bitcoin. Cash answer: profitable. Full-cost answer: not on that measure. Marketing answer: please turn to the adjusted EBITDA slide. This does not mean the cash margin is fake. Depreciation is noncash in the current quarter, and accounting lives do not perfectly match economic lives. But the equipment really does become less competitive. A miner who excludes hardware cost forever is not measuring a business. He is describing an inheritance. IREN produced another version of the split. For the quarter ending March 31, it showed $111.2 million in Bitcoin mining revenue and $35.3 million in direct Bitcoin mining cost of revenue, excluding depreciation. The consolidated company nevertheless reported a $247.8 million net loss after depreciation, impairments, corporate expenses, financing, and other items. Again: the machines can throw off operating cash while the enterprise around them consumes capital with the appetite of a hyperscaler at brunch. Individual hobbyists can make money too. Someone with sunk-cost equipment, genuinely surplus solar, four-cent power, useful winter heat, repair skills, and no need to price their own time may be perfectly happy. A home miner can also value privacy, network participation, education, or lottery-like solo mining more than conventional return. Those are legitimate reasons. They are not passive income. They are hobbies with revenue. ## The Great AI Pivot Is Mining’s Most Honest Earnings Call The clearest evidence that Bitcoin mining economics are difficult is that major miners keep explaining how excited they are to do something else with the power. CoinShares counted more than $70 billion in announced AI and high-performance-computing contracts across the public mining sector by early 2026\. Its report argued that some companies were becoming data-center operators that happen to mine Bitcoin, with AI offering higher and more predictable returns than hashprice near cyclical lows. MARA now describes itself as a digital-infrastructure company that converts energy into “high-value compute workloads.” Its Bitcoin operation remains foundational, but its filings discuss AI, HPC, critical IT, owned power, and allocating capacity according to economics. That is not mere buzzword migration. Miners already control the scarce ingredients AI developers want: energized land, substations, grid interconnections, power expertise, cooling experience, and an institutional tolerance for capex figures that cause civilians to sit down. The conversion is not automatic. Bitcoin ASICs cannot run AI models. Mining facilities tolerate hardware and network architectures that an AI cloud customer may reject. AI racks can demand different cooling, redundancy, fiber, service levels, and engineering. The data-center shell and power rights are adaptable; the SHA-256 boxes are not. Still, the message from capital is blunt. When the owner of a mine discovers that the grid connection may be worth more serving GPUs than ASICs, the mining rig has become the opening act for a data-center real-estate business. SiliconSnark has seen this movie on the other side of the compute boom. In our guide to [tokenmaxxing](https://www.siliconsnark.com/definitive-guide-to-the-rise-and-fall-of-tokenmaxxing/), AI usage became the metric that justified larger infrastructure bills. Mining companies are now standing at the border between two electricity-hungry economies: one turns power into probabilistic money; the other turns it into probabilistic PowerPoints. Public markets have believed dumber things. Sometimes in the same quarter. ## Yes, Meme-Coin Mining Is Real. No, Your Frog JPEG Is Not Doing the Hashing. “Meme coin” describes a culture and a marketing category, not a consensus mechanism. A meme coin can live on a proof-of-work blockchain, a proof-of-stake blockchain, or as a token issued on another network. Whether it can be mined depends on the underlying protocol, not how many dogs appear in the logo. Dogecoin is the canonical mineable meme coin. It uses Scrypt proof of work and, since 2014, supports auxiliary proof of work with Litecoin. In ordinary language, a compatible pool can submit the same work toward both networks and distribute both rewards. The miner is not splitting its hashrate between Litecoin and Dogecoin. It is using one stream of work to participate in both. [Dogecoin’s own mining guide](https://dogecoin.com/dogepedia/how-tos/mining-dogecoin/?ref=siliconsnark.com) is unusually candid: profitable mining generally requires a Scrypt ASIC, a pool, relatively cheap electricity, and a location capable of handling expensive, loud, hot hardware. This is not a Shiba Inu tapping at a laptop. It is industrial commodity production with better branding. The Bitmain Antminer L9 is a leading example: roughly 16 to 17 gigahashes per second at about 3,360 watts. Its economics weakened sharply by August. Contemporary calculators put an L9 around breakeven or slightly negative at ten-cent electricity, before capital recovery. At five cents, the power bill falls by about $4.03 per day, creating a modest operating margin—but still leaving the machine purchase, pool fees, ventilation, repairs, and future difficulty waiting in the hallway. Pepecoin extends the joke into protocol design. Its official site says Litecoin, Dogecoin, and Pepecoin can be mined simultaneously. It also offers an app that lets users contribute CPU or GPU work and receive Pepecoin payouts. That distinction matters: some consumer services advertise “mining” a coin when the software is actually mining whatever is economical and converting the proceeds into the selected payout asset. That is still a real service. It is not native mining of every coin in the dropdown. Many famous meme tokens are not mineable at all. Solana tokens rely on Solana’s proof-of-stake validator system. Ethereum tokens rely on Ethereum’s validators. The token issuer can distribute supply through sales, airdrops, liquidity incentives, staking-like programs, games, creator rewards, or interpretive dance. None of that turns a GPU into a block-producing miner for the token. This is particularly important for SiliconSnark. The original $SNARK adventure briefly reached an $800,000 market cap and left its creator with holdings worth less than $300, an outcome so perfectly aligned with independent media economics that it should be framed. The follow-up was even more instructive: [Pump.fun creator rewards eventually produced six dollars](https://www.siliconsnark.com/pump-fun-creator-rewards-what-you-actually-earn-for-a-viral-meme-coin/). Our later [deep dive into the meme-coin promotion machine](https://www.siliconsnark.com/shill-or-be-shilled-a-deep-dive-into-the-meme-coin-promotion-industrial-complex/) found a much more reliable business surrounding the tokens: people selling attention, raids, trending, listings, access, and confidence. Crypto repeatedly invents a volatile asset and then discovers that the steadier revenue belongs to the toll booths. SiliconSnark has covered the respectable version in [our stablecoin infrastructure deep dive](https://www.siliconsnark.com/deep-dive-stablecoins-turned-dollars-into-software-now-everyone-wants-the-toll-booth/) and in the way [Coinbase and PPRO turned tokens into a checkout option](https://www.siliconsnark.com/coinbase-and-ppro-turn-stablecoins-into-just-another-checkout-button/). Meme coins use similar rails, except someone has replaced treasury management with a raccoon holding a laser sword. ## Are the Rig Sellers the Only People Making Money? No. But this suspicion has survived every crypto cycle because it contains a durable economic truth: selling a shovel converts somebody else’s speculative upside into your current revenue. A hardware vendor gets paid when a rig ships. A reseller gets a spread. A hosting company collects recurring fees. A pool takes a percentage of mining rewards. Firmware vendors can charge licenses or dev fees. Repair shops get paid when hashboards develop opinions. Power producers sell electricity. Influencers earn affiliate commissions. The miner at the end absorbs coin price, difficulty, uptime, pool, curtailment, regulatory, and hardware-obsolescence risk. That looks like a shovel economy because it is one. But shovel makers do not enjoy guaranteed margins. Canaan, one of the oldest public ASIC companies, received about 78% of its 2025 revenue from products and about 21% from mining. It posted $529.7 million in annual revenue but only $41.2 million in gross profit. In the first quarter of 2026, amid weak Bitcoin prices and inventory pressure, [Canaan reported a $22.9 million gross loss](https://investor.canaan-creative.com/news-releases/news-release-details/canaan-inc-reports-unaudited-first-quarter-2026-financial?ref=siliconsnark.com). Selling the rigs was not a cheat code. It was manufacturing in a cyclical market, which is what crypto becomes whenever the hoodie comes off. Bitdeer is even more useful because it makes machines and mines with them. In Q1 2026 it reported $188.9 million in revenue and a $39 million gross loss. Its self-mining segment generated $146.9 million of revenue but a $34.4 million gross loss after electricity, depreciation, stock compensation, and other direct costs. Sales of SEALMINER machines and accessories generated only $3.7 million in revenue and about $0.2 million in gross profit. The company’s [own disaggregated results](https://bitdeer.gcs-web.com/news-releases/news-release-details/bitdeer-reports-unaudited-financial-results-first-quarter-2026?ref=siliconsnark.com) are an excellent antidote to the theory that inserting “ASIC manufacturer” in the business model causes money to spray from the vents. The stronger version of the shovel thesis is this: intermediaries often have clearer unit economics than miners. They may still execute badly, overproduce inventory, misprice credit, suffer tariffs, miss chip targets, or discover that their customers have stopped buying because hashprice fell through the floor. There is no risk-free seat. Some seats merely have better lumbar support. ## Home Mining Is Now Part Heater, Part Lottery Ticket, Part Tiny Rebellion Industrialization did not eliminate home mining. It changed what consumers are buying. At one end are smaller commercial ASICs marketed for garages, basements, sheds, or hosted facilities. They can earn pooled rewards but demand serious power and ventilation. A 3,500-watt machine running continuously uses about 2,520 kilowatt-hours a month. Before asking what it mines, ask whether the circuit, outlet, wire gauge, breaker, noise tolerance, insurer, landlord, spouse, and local fire code have joined the pool. At the other end are desktop and open-source lottery miners: low-hashrate devices such as Bitaxe-class boxes and compact products like Canaan’s Avalon Nano line. Canaan has sold the [Avalon Nano 3 for roughly $99 to $129](https://shop.canaan.io/products/avalon-nano-3-1?ref=siliconsnark.com), configurable around 65 to 140 watts. It can contribute to a pool, warm a desk, teach someone how mining works, or solo mine with odds that make a state lottery look like conservative fixed income. Calling these devices scams would be unfair if the seller explains the odds. They are educational hardware and participatory collectibles. A solo miner could find a block. So could a person buy one ticket and win Powerball. Probability allows many things that a retirement planner does not. Heat reuse makes the proposition more interesting. Nearly every watt entering a miner becomes heat. In a cold climate, a quiet miner can offset some electric-resistance heating that would have occurred anyway. The correct comparison is not “free heat”; it is whether the mining device produces useful heat at a net cost below the heater it replaces, after hardware expense and inconvenience. If the home uses a high-efficiency heat pump, the mining rig usually loses the thermodynamic beauty contest. If the alternative is resistance heat and the machine earns a little Bitcoin, the weirdness tax can be tolerable. This niche has room to grow because it sells something industrial mining cannot: intimacy with the network. The owner can see the hashes, control the keys, choose a pool, run a node, learn the protocol, and enjoy the physical absurdity of converting electricity into a globally tradable number beside a houseplant. Just do not call it passive. Fans collect dust with an active commitment to the craft. ## Should SiliconSnark Sell a Branded $SNARK Mining Rig? SiliconSnark should not sell a product marketed as a profitable $SNARK miner. The existing token is not mineable, the economics would be misleading, and the brand has already documented exactly how little money its meme coin generated for the person who created it. Contradicting that record with a glowing “earn passive income” box would be less satire than evidence. There is, however, a product hiding inside the joke. SiliconSnark could release a small, limited-run, clearly labeled Bitcoin lottery miner or pool miner—a branded open-source device in the Bitaxe or compact-home-miner tradition. It could feature CircuitSmith’s mustard-yellow face, pixel sunglasses, a live hashrate display, temperature, power draw, sats earned, current hashprice, and one unusually important number most mining dashboards bury: estimated profit after electricity at the owner’s entered rate. The product should mine Bitcoin or another protocol it truthfully supports. It could optionally award a fixed promotional amount of $SNARK from a disclosed marketing wallet for setup, ownership, or community participation. That would be an airdrop or loyalty reward, not mining. The interface should say so in language large enough to survive a screenshot. An even better version would behave like an anti-hype appliance: - show gross mining revenue and electricity cost side by side; - include pool fees and an optional hardware-amortization slider; - display solo-block odds without translating geological time into “potential upside”; - default to a quiet, low-power mode; - publish the firmware and document where pool traffic goes; - make wallets user-controlled; - never custody customer earnings; - never promise appreciation, yield, ROI, or a moon with transportation already arranged. Call it the **SiliconSnark Probability Heater**. Put “EXPECTED FINANCIAL OUTCOME: COMPLICATED” on the startup screen. Ship it with a certificate stating that the owner now controls several trillion guesses per second and still cannot guess what crypto will do on Tuesday. This would be merchandise with educational utility, not a miniature investment scheme. That distinction matters beyond brand hygiene. The SEC staff’s 2025 [meme-coin statement](https://www.sec.gov/newsroom/speeches-statements/staff-statement-meme-coins?ref=siliconsnark.com) said typical meme coins described as entertainment-oriented collectibles generally do not involve securities transactions, but emphasized that the view is not a rule, that economic realities control, and that fraud remains actionable. Add an earnings product, managerial promises, or a story about buyers profiting from SiliconSnark’s efforts, and the facts become more complicated. A novelty device should not cosplay as a yield-bearing ecosystem because someone discovered a gradient background. The brand also does not need to invent another chain. Launching “SNARK PoW” would fracture the identity of the existing Solana token, create security and exchange problems, require ongoing protocol work, and turn a funny hardware drop into a monetary-policy internship. We have already explored the social economics in [the week the meme coin took over SiliconSnark](https://www.siliconsnark.com/this-week-in-snark-the-meme-coin-heard-round-the-web/). There is no editorial need to add chain reorganizations to the content calendar. If the project happens, use an established, auditable board from a credible supplier; take preorders only after a working prototype and landed cost exist; commission electrical and thermal safety testing appropriate to every sales market; separate the product warranty from all coin outcomes; obtain counsel on consumer, tax, sanctions, money-transmission, and promotional questions; and make the joke target mining hype rather than the customer. That could be delightful. It could also sell dozens of units, which in mining-hardware terms qualifies SiliconSnark to announce a vertically integrated infrastructure platform and begin evaluating strategic AI workloads. ## How to Evaluate a Rig Without Becoming the Yield The safest way to shop for a mining rig is to assume every displayed profit number is the beginning of a calculation, not the end. A seller may use the most profitable coin available to the algorithm at that instant, a low electricity default, perfect uptime, no pool fee, no cooling overhead, no tax, and no hardware cost. This is not necessarily fraud. It is the financial equivalent of photographing a vacation rental with a wide-angle lens. Start with the algorithm. SHA-256 equipment mines Bitcoin-family proof-of-work chains. Scrypt equipment targets Litecoin, Dogecoin, and compatible merged-mined coins. An Etchash or KAWPOW machine belongs to a different market. Hashrate units are not comparable across algorithms: one terahash of SHA-256 is not “more mining” than one gigahash of Scrypt in any useful cross-protocol sense. It is like comparing the RPM of a blender to the cruising speed of a ferry because both numbers move when the machine is on. Then calculate from gross revenue down: 1. **Revenue:** use current network difficulty, block rewards, fees, coin prices, and realistic pool payouts; 2. **Power:** multiply measured kilowatts by 24 and by the actual marginal electricity rate, including delivery charges where applicable; 3. **Facility overhead:** include ventilation, pumps, fans, networking, transformers, and cooling power; 4. **Pool and service fees:** small percentages matter when the gross margin is already wearing ankle socks; 5. **Uptime:** model something below 100%, because maintenance and reality remain undefeated; 6. **Capital:** divide the landed rig price—including shipping, tariffs, wiring, power supplies, and setup—over a plausible economic life; 7. **Difficulty and halvings:** run scenarios in which revenue per unit of hash falls; 8. **Exit value:** assume a specialized machine may be worth very little precisely when you most want to sell it. The break-even electricity rate is a useful quick filter. Divide daily gross revenue by daily kilowatt-hours. For the S23 example above, $9.92 divided by 83.95 kWh gives about 11.8 cents per kWh before pool fees, cooling, maintenance, and capital. That does *not* mean anyone paying 11 cents has found a viable investment. It means the power company has not yet consumed every dollar of mining revenue. Congratulations on making it to line two. Next, stress the variables together. What happens if dollar hashprice falls 25%, uptime is 95%, the pool takes 2%, and one hashboard needs repair? What happens after the 2028 subsidy halving if transaction fees do not replace the lost issuance? What happens if the coin rises but difficulty follows it because every warehouse in the world plugs machines back in? A model that survives only when one variable moves favorably is not robust. It is a wish with conditional formatting. Hosted mining adds another layer. It can solve the home problems of noise, heat, voltage, and maintenance, and it may provide access to a better power rate. It also moves your machine into someone else’s building under someone else’s contract. Ask who owns the rig, who chooses the pool, who holds the payouts, how uptime is measured, whether curtailment is credited to you, who pays for repairs, what happens if the host fails, and how the machine can be retrieved. “Your miner is in a secure facility overseas” is not a control environment. It is a sentence. Cloud mining removes even the machine. The customer buys a contract tied to hashrate while the provider owns or sources the equipment. There are legitimate versions, but the structure makes verification hard and creates room for fees, unfavorable termination clauses, counterparty failure, and outright fiction. If a contract promises stable returns from an unstable commodity while withholding facility, pool, machine, and payout evidence, the most efficient miner in the arrangement may be the sales funnel. Finally, compare mining with simply buying the coin. A rig is a leveraged operating bet on the spread between mining revenue and costs. Buying Bitcoin is exposure to Bitcoin without fans, repairs, facility risk, delivery delays, or a depreciating appliance. Mining can outperform when power is cheap, hardware is purchased well, and network economics cooperate. Direct ownership can outperform when they do not. The rig must earn its complexity. This is the same maturation SiliconSnark has watched elsewhere in crypto. When [SoFi began minting its own stablecoin](https://www.siliconsnark.com/sofi-started-minting-a-stablecoin-because-the-super-app-wasnt-complicated-enough/), the interesting part was not the token alone but the regulated banking and settlement machinery around it. Mining deserves the same unsentimental treatment. The coin gets the thumbnail. The operational stack decides who gets paid. ## The Verdict: Mining Makes Money, but the Power Contract Is the Product Crypto mining in 2026 is neither dead nor broadly easy money. It is a mature, cutthroat commodity business wrapped around a volatile monetary asset. The network is enormous. The hardware is astonishing. The margins are frequently rude. People still use rigs because proof-of-work networks still pay them, because some operators possess structural advantages, because home miners value more than immediate return, and because human beings will always enjoy a machine that appears to manufacture money even when the spreadsheet describes it as an artisanal space heater. People do make money. They tend to own cheap power, efficient hardware, optimized facilities, repair capability, flexible curtailment, patient capital, or several of these at once. Others make a cash margin until depreciation enters the chat. Still others lose money in dollars while accumulating coins they expect to appreciate, which is mining combined with a second speculation and therefore considered diversification in certain Telegram groups. Meme-coin mining is real where the meme coin uses proof of work. Dogecoin matters. Merged mining makes its economics stronger and its security relationship with Litecoin genuinely interesting. But most launchpad tokens are not mined; they are issued and traded on somebody else’s proof-of-stake network. The meme is the asset. The mining rig cannot smell irony. And the sellers? Some make money. Some lose it. The steadier winners are often the businesses selling power, hosting, pools, financing, repairs, chips, and attention—the infrastructure around the wager. The plumbing is the point, even when the plumbing is liquid cooling for a box screaming SHA-256 into the void. So yes, SiliconSnark should consider a branded miner, but only as an honest novelty: small, open, safe, user-controlled, and almost comically transparent about the economics. It should not claim to mine $SNARK. It should not promise profit. It should make the invisible machinery legible and the delusion optional. That is a narrower business than selling digital gold machines. It is also one I might trust.