Bitcoin mining: how it works and why home mining no longer pays

Key points

Mining is not solving equations: it is trying random nonces until the hash of the block header falls below a numerical target.
Difficulty readjusts every 2,016 blocks to hold the average interval at ten minutes, whatever the network hashrate happens to be.
As of August 2026 the network runs at ~893 EH/s with difficulty ~125.81 T; the next halving is estimated for 17 April 2028, about 85,600 blocks away.
A 200 TH/s ASIC would produce roughly $8 per day today while consuming 84 kWh: at typical residential electricity prices that is a systematic loss.
With fees at 1–2 sat/vB, transaction fees contribute under 1% of the block reward; the long-term security budget remains an open problem.

Market data: Market data on this page was collected on 28 August 2026 and changes constantly. Check the linked sources before relying on it.

The same video resurfaces every few months: someone sets up two machines in a garage and explains that they have built themselves a passive income. The numbers on screen almost never include the electricity bill, the difficulty readjustment, or what happens once the hardware is three years old.

Mining is the mechanism that lets bitcoin function without a central authority, and understanding it properly also explains why the network burns so much energy and why nobody can rewrite its history. As an economic activity, though, it is heavy industry: razor-thin margins, energy as the dominant cost, and hardware that depreciates at frightening speed.

Here are the two things separately: the mechanism, described precisely, and the economics, with a full calculation and every assumption stated out loud.

What a miner actually does

A miner does two things, and the first is not the one usually described.

First, it builds a candidate block. It collects transactions from the mempool — the set of broadcast but unconfirmed transactions — sorts them, usually by what they pay per byte, and packs them into a block until the available capacity is full. At the front it adds a special transaction, the coinbase, which pays the reward to itself.

Second, it searches for a nonce. It takes the block header, an 80-byte summary containing the version, the previous block hash, the Merkle root of the transactions, the timestamp, the difficulty target and a variable field called the nonce. It applies SHA-256 twice. If the resulting number is lower than the target, it has a valid block. If not, it changes the nonce and repeats.

There is no equation. There is no mathematical problem to solve. It is pure brute force: try and discard, hundreds of quintillions of times per second, until a small enough number comes out. The only way to find it sooner is to try faster than everyone else, which is why hashrate is the measure of everything.

The elegance of the design, already present in the original 2008 whitepaper, is its asymmetry: finding the nonce costs an enormous amount of work, and verifying it is correct costs a single hash operation. Any node on the network checks your block in microseconds.

The difficulty adjustment, every 2,016 blocks

If the network's hashrate multiplied tenfold, blocks would be found ten times faster. The protocol prevents that with an automatic mechanism.

Every 2,016 blocks — roughly two weeks at the intended pace — each node calculates how long those 2,016 blocks actually took and compares it against the 20,160 minutes they should have taken at ten minutes apiece. If they came faster, difficulty rises; if slower, it falls. The adjustment is capped at a factor of four in either direction.

Difficulty and hashrate are directly linked. As of August 2026, with difficulty at ~125.81 T, the implied hashrate is on the order of 900 EH/s, very close to the ~893 EH/s reported by mempool.space. The exact readjustment formula is documented in the Bitcoin wiki. The two numbers say the same thing from different angles.

This has a consequence beginners rarely see coming: your machine's profitability declines on its own. Not because it degrades, but because every time new hardware joins the network, difficulty rises and your slice of the pie shrinks. Your machine does exactly what it did on day one and earns less.

Block reward and the halving schedule

Every block pays its miner a reward made of two parts: the subsidy, newly created bitcoin issued by the protocol, and the fees from the transactions included. The subsidy halves every 210,000 blocks.

Halving	Date	Block	Reward
1st	28 Nov 2012	210,000	50 → 25 BTC
2nd	9 Jul 2016	420,000	25 → 12.5 BTC
3rd	11 May 2020	630,000	12.5 → 6.25 BTC
4th	20 Apr 2024	840,000	6.25 → 3.125 BTC
5th (estimated)	~17 Apr 2028	1,050,000	3.125 → 1.5625 BTC

On 28 August 2026 the block height was 964,393, leaving roughly 85,600 blocks until the next halving. At 144 blocks a day, that is a little under twenty months. The exact date cannot be fixed because it depends on the network's real pace, which is why it is always given as an estimate.

From CPU to ASIC: how home mining died

In 2009 people mined with their computer's processor. Anyone with a laptop took part in consensus. That phase was short, and what followed is a textbook lesson in hardware optimisation:

CPU. A general-purpose processor, capable of everything and fast at nothing in particular.
GPU (2010). Graphics cards perform many simple operations in parallel, which is exactly what SHA-256 asks for. A jump of two orders of magnitude.
FPGA (2011). Reprogrammable circuits configured specifically for the algorithm. Better energy efficiency than GPUs.
ASIC (2013 onwards). Integrated circuits manufactured to do one thing: double SHA-256. They cannot run anything else. In exchange, they are several orders of magnitude more efficient than any alternative.

Once ASICs arrived, competing with other hardware stopped making mathematical sense. And the game became industrial: whoever secures cheaper electricity and buys more machines wins. Today mining concentrates in facilities with wholesale-priced energy access, something no residential contract can match.

Pools: how the reward is split

An individual miner with a minuscule share of the network faces brutal variance: decades can pass without finding a block. Pools solve that by aggregating hashrate.

The pool distributes work to participants and asks them to submit shares: partial solutions that meet a target far easier than the real one. Shares are worthless in themselves, but they demonstrate verifiably how much work you contributed. When someone in the pool finds a real block, the reward is split according to shares and the pool deducts its fee.

There are two main schemes. PPS (pay per share) pays you a fixed amount per share, with the pool absorbing all the variance in exchange for a higher fee. PPLNS (pay per last N shares) pays out only when the pool finds a block, in proportion to your recent shares: lower fees, lumpier income, and a penalty if you hop in and out constantly.

Worked example: what an ASIC actually produces

This is the calculation almost nobody shows in full. I will start by separating verified data from assumptions.

Verified data as of 28 August 2026: network hashrate ~893 EH/s, difficulty ~125.81 T, block reward 3.125 BTC, BTC price ~$79,560.

Assumptions I cannot verify, which you should replace with your own: a unit rated at 200 TH/s drawing 3,500 W, and an electricity price of $0.15 per kWh. That electricity figure is an assumption, not a verified market rate — retail tariffs vary enormously between countries, between suppliers, and between time-of-day bands, and I have no verified residential kWh price for August 2026. I also have no official specification sheet for a specific ASIC model on that date. Substitute the number on your own bill and your own machine's rating, and the method still holds.

Step 1 — Your share of the network. 893 EH/s is 893,000,000 TH/s. Your share is 200 / 893,000,000 = 0.0000002240, a bit over two ten-millionths.

Step 2 — The network's daily issuance. 144 blocks per day × 3.125 BTC = 450 BTC per day.

Step 3 — Your expected production. 450 × 0.0000002240 = 0.0001008 BTC per day. With BTC at $79,560, that is $8.02 per day, roughly $240 a month.

Step 4 — Your electricity bill. 3,500 W running 24 hours is 84 kWh per day. At the assumed $0.15/kWh, that is $12.60 per day, roughly $378 a month.

Step 5 — The result. You produce about eight dollars and spend about twelve and a half. You lose money every single day, and that is before counting hardware depreciation, cooling and noise.

Step 6 — Where the threshold sits. To cover the electricity alone you would need those 84 kWh to cost no more than $8.02, which is about $0.095 per kWh. Whether any tariff available to you reaches that level is the single question that decides the whole exercise, and for most residential customers in most markets the answer is no. That number, and nothing else, is why mining has migrated to industrial facilities with generation-priced energy contracts.

And this is the optimistic scenario. The calculation assumes constant difficulty, a constant price and 24-hour uptime. In practice difficulty almost always rises, hardware depreciates, and in April 2028 the halving will cut the subsidy in half at a stroke: on the same assumptions, that $8.02 per day becomes roughly $4.

A note on solo mining. With that share of the network, the mathematical expectation says you would find a block roughly every 4.47 million blocks — that is, once every 85 years. This is not rhetorical exaggeration: it is one divided by your share. Which is why pools are not an option but the only way small-scale mining produces anything other than zero.

Energy: the debate without either caricature

The network consumes a lot of electricity by design, not through inefficiency. The energy cost is the security: rewriting history would require redoing the accumulated work.

From there, the two usual positions are both incomplete.

Whoever says "bitcoin consumes as much as a country" has the magnitude right but measures consumption without asking where the energy comes from or what it displaces. Whoever says "mining is all renewable" is relying on figures self-reported by the industry with contested methodologies.

The arguments that do have substance:

Load flexibility. A miner can shut down in seconds, which makes it an interesting interruptible consumer for grid operators. This has been used in Texas during periods of system stress.
Stranded energy. Some generation cannot reach the market for lack of transmission: remote hydro, overnight wind surpluses. Mining with it carries a low opportunity cost.
Flared gas. Oil fields burn off surplus methane because transporting it does not pay. Using it to generate electricity on site and mine converts combustion with no use into combustion with one. It is not clean energy; at best it is making use of something that was going to burn anyway. And it creates an economic incentive tied to hydrocarbon extraction, which is precisely the objection its critics raise.

None of those arguments makes mining neutral. They describe a more nuanced picture than the headlines on either side.

The open problem: the security budget

This is the point where intellectual honesty requires saying there is no answer.

The subsidy trends to zero around 2140. From some point onwards, network security will have to be paid for by fees alone. And today's data is uncomfortable: as of August 2026, the mempool is so empty that fees run at 1–2 sat/vB. It is the same phenomenon that has made network fees generally so cheap, except that here it is bad news for whoever lives on collecting them.

Do the arithmetic. A block's capacity is on the order of one million virtual bytes. At 1.5 sat/vB, a full block would collect around 1,500,000 sats, which is 0.015 BTC. Against a subsidy of 3.125 BTC, fees contribute roughly half a percent of what the miner earns. Issuance provides the other 99.5%.

The positions are well known. Some argue that demand for block space will grow enough. Others, that the price will rise so far that proportionally small fees will still be enormous in absolute terms. Others, that the model does not close and something will have to change. Nobody has an experiment that settles it, because the problem belongs to the coming decades. If you want the contrast with the other model, the comparison is in proof of work versus proof of stake.

Mining on other networks

Ethereum is no longer mined. On 15 September 2022, with The Merge, it replaced proof of work with proof of stake. There are no miners and no hashrate: blocks are proposed by validators who have deposited 32 ETH. It was the largest consensus mechanism change ever executed in production on a network with billions at stake, and it shipped without incident. If you want to generate yield on Ethereum, the mechanism is staking, not mining, and its rules — entry and exit queues included — are entirely different.

Proof-of-work networks still exist with algorithms designed to resist ASICs, on the idea that general-purpose hardware should stay competitive. Historically that resistance lasts until the market is large enough to justify developing a dedicated chip. And mining a small coin adds price and liquidity risk on top of an operation whose margins are already thin.

One regulatory note: the joint SEC and CFTC interpretation of 17 March 2026 explicitly clarified the treatment of protocol mining within its crypto-asset taxonomy. That is a step forward for legal certainty in the United States, though it does not move a single digit of the electricity bill arithmetic. Tax treatment of mining income and of the coins you receive varies by jurisdiction — check yours, and take professional advice rather than assuming what applies somewhere else applies to you.

Checklist before buying a machine

Calculate your share of the network using the current hashrate, not the figure in a video you watched six months ago.
Use the kWh price from your own bill, taxes and grid charges included, not the wholesale price.
Assume difficulty will rise over the machine's useful life. If the operation only works at today's difficulty, it does not work.
Factor the April 2028 halving into any projection running beyond twenty months.
Add cooling, adequate electrical installation, and the fact that an ASIC makes a noise that is hard to live alongside in a home.
Compare the result against simply buying the same amount of bitcoin. If mining comes out worse, you already have your answer.

Do not use profitability calculators that will not let you enter your own electricity price and a difficulty increase.
Do not buy cloud mining contracts without verifying who operates the machines and where. It is a sector with a long history of fraudulent schemes.
Do not project income using a future bitcoin price. It is the variable you do not control and the one that turns any spreadsheet into fiction.

Frequently asked questions

Can I mine bitcoin with my computer or graphics card?

The software will technically run; economically it is absurd. A high-end GPU manages tens of millions of hashes per second against the hundreds of trillions an ASIC delivers. Your share of the network would be so small that the mathematical expectation of finding a block is measured in hundreds of thousands of years, while the electricity bill arrives every month.

What happens to Bitcoin's security when the reward runs out?

It is the protocol's unresolved question. Around 2140 issuance reaches zero and miners will earn fees alone. Today, with the mempool at 1–2 sat/vB, a block's fees amount to less than one percent of what the subsidy contributes. Sustaining the same level of security would require far higher fees or a far higher price. Nobody has a tested answer.

Why is Ethereum no longer mined?

Because on 15 September 2022, in the upgrade known as The Merge, Ethereum replaced proof of work with proof of stake. From that day there have been no miners and no hashrate: blocks are proposed by validators who have deposited 32 ETH. The graphics cards that mined ETH were left without a use overnight.

Is joining a mining pool worth it?

If you are going to mine at all, there is no sensible alternative: solo variance is unbearable. A pool aggregates the hashrate of thousands of participants, distributes rewards according to the shares each one contributes, and deducts a fee. It converts an enormous, improbable prize into a small, regular income. What it does not change is the arithmetic of profitability, only its variance.

Sources and references

Important notice. InfoCrypto is an educational site. Nothing here is financial, tax or investment advice. Crypto assets are high risk: you can lose all the capital you invest. Always verify data against the original sources before making any decision, and consult a licensed professional if you need to.