The New Map of Hashrate Growth: Where Bitcoin Mines in 2026

For most of Bitcoin’s history, the phrase hashrate growth was almost a tautology. The amount of computing power pointed at the network went up, quarter after quarter, year after year, through bull markets and bear markets, through the 2017 mania and the 2022 collapse. The line on the chart bent in one direction. In 2026, for the first time, it stopped bending.

The important part is not that it stopped, but why. Every previous decline in Bitcoin’s mining power had a familiar cause: a price crash that bankrupted marginal operators, or a government that outlawed the machines. This one is different. In 2026 the same electricity that used to spin ASICs is being bid away by artificial intelligence, and miners are handing it over willingly, because a rack of GPUs training a model pays more than a rack of ASICs hashing SHA-256. The network’s aggregate computing power is roughly flat to slightly lower. Underneath that flat headline, the map is being redrawn.

That is the real 2026 story: hashrate growth did not stop, it relocated. Some countries are adding capacity as fast as they can pour concrete; others are bleeding it. Some machines are twice as productive as the ones they replaced. This is a guide to where Bitcoin’s mining power is actually growing in 2026, where it is fading, and what the redrawn map means for the network’s security, its economics, and the price bet buried inside every megawatt. As of late August 2026, Bitcoin traded near $78,747 with a market capitalization around $1.58 trillion, its computing power hovered between roughly 900 exahashes and one zettahash per second, and its mining difficulty sat at 125.81 trillion, down from a record set the previous autumn.

What hashrate growth actually measures

Hashrate is the number of SHA-256 hashes the network computes every second as miners race to find a valid block. Each hash is a single guess at a lottery ticket; the more guesses per second, the more security and the more energy the network consumes. Because no one can poll every machine on Earth, hashrate is never measured directly. It is estimated from two things anyone can read off the blockchain: the current difficulty and how fast blocks are actually arriving. The rough formula is difficulty multiplied by 2 to the 32nd power, divided by the average block time in seconds. When blocks come faster than the ten-minute target, the implied hashrate is higher; when they lag, it is lower. That is why single-day readings are noisy, and why serious analysts quote a seven-day or thirty-day moving average rather than a spot figure.

The units have climbed through the metric prefixes at a pace no other computing network has matched. In 2009 a laptop CPU hashed in the millions per second; by 2026 the network as a whole works in zettahashes, or sextillions of hashes per second. The ladder below shows how far the number has traveled.

The distinction that matters for 2026 is between two kinds of growth hiding inside that single number. One is more machines drawing more power: build another warehouse, energize another hundred megawatts, and hashrate rises. The other is more hashes per watt: swap last generation’s rigs for this generation’s, and hashrate rises even if the power bill does not budge. For fifteen years both engines ran at once and no one bothered to separate them. In 2026 they have pulled apart, and telling them apart is the only way to read the map correctly.

The 2026 anomaly: the first voluntary decline

Bitcoin’s difficulty fell year over year in 2026 for only the second time in the network’s history, sliding about 14% from its high for the year as collapsing mining revenue forced operators to power down or repurpose their sites, CoinDesk reported. The global hashrate told the same story: Hashrate Index’s Q2 2026 heatmap put the network at roughly 1,004 EH/s, down 5.8% from the previous quarter’s 1,066. Numbers like that would once have signaled panic. In 2026 they signaled a choice.

To see why this decline is different in kind, it helps to line up the network’s earlier contractions against the one happening now. The first three were involuntary. The fourth is not.

In 2018 and 2022, price forced miners off the grid. In 2021, the Chinese state did. In 2026, for the first time, miners are leaving because something else pays more for the same electricity, and the megawatts are being repurposed rather than abandoned. James Butterfill, head of research at CoinShares, called it “one of the most challenging periods” for miners since the last halving, noting that the cash cost to mine a single Bitcoin had climbed to about $79,995 in the fourth quarter of 2025 as the price fell from $124,500 in October toward $86,000 by late December. When revenue per terahash sinks below the cost of the terahash, the rational move is not to keep hashing at a loss; it is to find a better use for the power. That is the hinge the whole 2026 map turns on.

The new map: where Bitcoin’s hashrate lives in 2026

A flat aggregate hides a churning distribution. Hashrate Index’s Q2 2026 heatmap, still the most coherent single snapshot available, ranks the mining world like this.

Two facts jump out. First, the top of the table is sticky: the United States, Russia, and China together hold roughly two-thirds of the network, and their order has barely moved across consecutive quarterly snapshots. Second, the movement is all underneath. Kyrgyzstan grew its hashrate an estimated 167% quarter over quarter off a tiny base, while Iran lost about 7 EH/s to regional conflict, a shock Hashrate Index framed as localized disruption rather than systemic risk because neighboring capacity stayed stable. That is the pattern for 2026 in miniature: hashrate is not being created or destroyed so much as reshuffled, and reading the map means watching the cells that change color, not the ones that stay the same.

The United States: growth by policy and grid

The clearest growth story on the map is also the largest cell. The United States mines more than a third of all Bitcoin, concentrated on the deregulated grids of Texas and Wyoming where miners sign flexible-load contracts and power down within seconds when the grid needs the electricity back. Two policy shifts turned a favorable jurisdiction into a magnet. In March 2025 the White House established a Strategic Bitcoin Reserve, capitalizing it with forfeited coins and signaling that the federal government viewed the asset as strategic rather than suspect. Days later, the SEC’s Division of Corporation Finance stated that proof-of-work mining, whether solo or pooled, does not implicate US securities laws, removing a regulatory cloud that had hung over the industry for years.

The result is capital flowing toward American sites even as those sites face the same AI competition as everyone else. The US gained roughly two percentage points of global share in a single quarter despite the pivot, evidence that, so far, mining growth and AI growth are additive on US soil rather than mutually exclusive. Politically connected ventures have amplified the trend; American Bitcoin Corp, backed by Hut 8 and members of the Trump family, is one of several outfits treating hashrate as a national-champion project. The frictions that remain are mundane rather than existential. Access to banking is a recurring one, and the fight over whether crypto firms can keep a checking account, covered in our look at crypto debanking in 2026, touches miners as much as exchanges. On the whole, though, the US cell is growing because the rules got clearer and the grid stayed flexible.

Russia and China: legal but isolated, banned but back

The second and third cells are studies in paradox. Russia holds roughly 17% of global hashrate and legalized mining nationally in a law effective November 2024, then promptly layered regional bans on top of it. Ten regions, including Buryatia and Zabaikalsky Krai, moved to a permanent year-round prohibition starting in January 2026, a six-year restriction that state media attributed to grid strain, as CoinDesk reported. Russian hashrate is legal at the federal level, banned in specific regions, and isolated internationally: BitRiver, the country’s largest industrial miner, was the target of the first ever US sanction against a crypto mining company, designated by the Treasury in April 2022. Growth here is real but brittle, hostage to both domestic energy politics and external sanctions.

China is the sharper irony. Beijing banned mining in mid-2021, yet the country has quietly climbed back to third place, with Hashrate Index putting it at 12% and a Reuters investigation estimating closer to 14% as of late 2025; some trackers guess higher still. Illegal mining is hard to measure, so any China figure is a range, not a point. The deeper irony is industrial. Bitmain, MicroBT, and Canaan, all Chinese, still manufacture the overwhelming majority of the world’s ASICs, so China profits from the global hashrate boom as an exporter even while banning the activity domestically. Every new machine energized in Texas or Paraguay likely shipped from a Chinese factory. A trade escalation on ASIC imports would be one of the few genuine supply-side risks to hashrate growth, and it is not obviously priced in.

The new frontiers: Paraguay, the Gulf, and Ethiopia

If the top three cells are stable, the growth at the edge is where the map is most alive. Paraguay, now the fourth-largest mining country at 4.3%, runs on surplus hydropower from the Itaipu dam. HIVE Digital is expanding its Yguazu data center from 300 to 400 megawatts, targeting around 35 EH/s from a single Paraguayan build, the kind of concentrated bet that can move a country’s share by itself. The trade-off is familiar to frontier mining: power costs have roughly doubled and steep deposits are squeezing smaller operators as the national grid operator warns about strain.

The Gulf is a different model of growth, deliberate rather than opportunistic. The United Arab Emirates and Oman together hold about 6% of global hashrate, anchored by Abu Dhabi-listed Phoenix Group and its more than 550 megawatts across several countries. Gulf mining is sovereign industrial policy, funded by state-linked capital and built for durability, which makes it less prone to the boom-and-bust of cheap-power frontier plays. Ethiopia, at 2.5%, is the riskier frontier. The Grand Ethiopian Renaissance Dam, inaugurated in September 2025 with a maximum output above 5,000 megawatts, drew more than twenty mostly Chinese-backed miners on power-purchase agreements. But the state utility wants to phase mining out in favor of exporting that electricity, and proposed rate hikes could make a large share of Ethiopian operations unprofitable within a year or two. Frontier growth arrives fast and can leave just as fast.

The cautionary tales: Kazakhstan, Iran, and Bhutan

Every growth map needs its ghosts, the places that grew fast and then gave it all back. Kazakhstan is the textbook case. After the 2021 China ban, it absorbed a huge slice of the mining diaspora and briefly reached an oft-cited 18% of global hashrate, only for grid strain to force a crackdown: a mandatory state electricity marketplace, per-transaction power caps, and real-time draw monitoring. Kazakhstan now sits at about 1.8%, a reminder that power which is cheap for a structural reason, in this case a stressed grid, is cheap right up until the moment it is rationed.

Iran is a fresher example of the same fragility, its roughly 7 EH/s loss in a single quarter traced to regional conflict rather than policy. Bhutan is the strangest ghost of all: a sovereign miner that ran hydropowered rigs through a Nasdaq-listed partner and then, according to CoinDesk, sold roughly 70% of its Bitcoin over eighteen months and appears to have slowed or halted mining altogether. The exact residual holding is disputed across sources, but the direction is not. The lesson these three teach is that hashrate at the frontier is a loan, not a deposit: it can be recalled by a grid operator, a war, or a treasury decision, and when it leaves it does not vanish from the network so much as reappear somewhere sturdier.

The real growth engine: efficiency, not machine count

Step back from geography and a quieter engine explains why the network keeps setting hashrate records even when the machine count barely moves. The single most important number in mining is not price or share; it is joules per terahash, the energy a rig burns to compute a trillion hashes. That number has fallen by roughly an order of magnitude in a decade.

That fall is the latest chapter of a hardware arms race that has run since the network’s earliest days. Bitcoin was mined on ordinary laptop processors in 2009, moved to graphics cards around 2010 once miners realized a GPU could hash roughly a hundred times faster, briefly passed through field-programmable gate arrays, and then, from 2013, settled on application-specific integrated circuits built to do nothing but compute SHA-256. Each transition was a step-change in hashes per watt, and each one obsoleted the machines before it almost overnight. The pattern has never really stopped; it has only slowed to a cadence set by chip fabrication, which is why a single new model launch can still reshape mining economics across three continents.

The S23 Hydro, which CoinDesk flagged as a profound shift when it was unveiled, is the first series miner to cross below 10 joules per terahash. That is roughly a tenfold efficiency gain over the S9 in ten years, and it is the hidden reason the map can grow without the power grid noticing. A site that swaps a hall of S19s for a hall of S23 Hydros can more than triple its hashrate on the same electrical connection. Efficiency is a form of hashrate growth that never shows up as a new warehouse. It also reshapes geography, because it steadily lowers the payoff for chasing the very cheapest, least reliable power. When a machine sips energy, the deciding question shifts from where can I find the cheapest kilowatt-hour to where can I find stable power that can also host AI, which quietly favors the United States, Canada, and the Gulf over the Ethiopias and Kazakhstans of the world.

AI and HPC: the new competitor for every megawatt

The force redrawing the map is the same one holding aggregate hashrate flat: artificial intelligence wants the miners’ electricity, and it will pay a premium for it. CoinShares estimates that publicly listed miners have signed more than $70 billion in AI and high-performance-computing contracts, and projects that up to 70% of listed-miner revenue could come from AI by the end of 2026, up from roughly a third today. A megawatt that earns a mediocre return hashing SHA-256 can earn a far better one renting GPUs to a model developer, so the rational operator with a power contract and a cooled building does both, and tilts toward whichever pays more this quarter.

This is why the 2026 decline is voluntary. The megawatts are not lost; they are re-tenanted. And it creates a genuine paradox at the heart of the growth story: AI is simultaneously the biggest competitor for mining power and, through those contracts, the biggest source of the cash flow that keeps miners solvent enough to keep hashing at all. The trust problem that AI compute raises, namely whether you can rely on a GPU you do not physically control, is the subject of our piece on decentralized inference in 2026; for miners the question is simpler and more mercenary, which tenant pays more per megawatt-hour this month. As long as the answer is AI, hashrate growth will keep bending toward wherever the same buildings can host both workloads.

Difficulty: the thermostat that rebalances the map

None of this reshuffling breaks the network, because Bitcoin has a built-in self-correction. Every 2,016 blocks, roughly every two weeks, the protocol adjusts difficulty so that blocks keep arriving about every ten minutes regardless of how much hashrate is online. When miners leave, whether to AI, a grid crackdown, or bankruptcy, difficulty falls and the survivors earn a larger slice of the same reward, which stabilizes their economics and stops the bleeding. That is why difficulty in late August 2026 sat at 125.81 trillion, roughly a fifth below its November 2025 record near 156 trillion, and was estimated to fall again at the September retarget even as the price recovered. Difficulty lags hashrate, and hashrate lags price, so the thermostat is always fighting the last war.

JPMorgan’s Nikolaos Panigirtzoglou put the mechanism plainly in a mid-2026 note, telling readers that “when bitcoin trades below its production cost, higher-cost miners power down, the hashrate declines, and difficulty adjusts lower.” His team pegged the all-in cost of producing a Bitcoin near $78,000, estimated that around 20% of the global fleet was unprofitable, and measured a difficulty-to-price beta of about 0.62 over the prior six months, meaning difficulty moves a little more than half as much as price, in the same direction and with a delay. The thermostat is what lets geography churn violently while the ten-minute heartbeat stays steady. Countries can rise and fall, machines can migrate, and the network simply retunes itself around whatever hashrate shows up.

What growth buys: security and the 51% question

Hashrate growth is not an end in itself; it is the network’s security budget, the wall of energy an attacker would have to out-muscle to rewrite recent history. So a flat or falling hashrate raises a fair question: is Bitcoin less safe in 2026? The honest answer is less than the headline implies. Campbell Harvey of Duke University modeled the cost of a one-week majority-hashrate attack and put it around $6 billion, split between roughly $4.6 billion of hardware, $1.34 billion of data-center construction, and $0.13 billion of electricity, about 0.26% of the network’s value at the time. Matt Prusak, president of American Bitcoin Corp, was blunt in response: “My attitude is that economic feasibility kills the 51% thesis. I live in the real world, and I am not concerned.”

Theory is not the only evidence. Bitcoin itself has never suffered a successful 51% attack in more than fifteen years, yet smaller proof-of-work coins with thin hashrate have, and repeatedly. Ethereum Classic and Bitcoin Gold were each rewritten by attackers who simply rented enough hashpower to overwhelm chains whose security budgets were small enough to buy. That is precisely why aggregate hashrate still matters even when it is geographically lopsided: the wall only deters if it is expensive, and Bitcoin’s remains the most expensive wall in the industry by a wide margin. A flat year does not change that as long as the network still commands hundreds of exahashes an attacker would have to match block for block.

Harvey has since sharpened the argument in a way every mining watcher should note. By pairing the hashrate acquisition with a large short position in offshore Bitcoin derivatives, an attacker could profit from the price crash the attack itself causes rather than needing to recoup costs through mining. That, he told reporters, lifts the effective cost to about $8 billion, or “about 50 basis points of the value of bitcoin,” while making the economics far more dangerous, because “the difference today is the derivatives markets.” The reassuring part is that the security budget scales with Bitcoin’s value: a more valuable network is a more expensive one to attack. The contrast with proof-of-stake is instructive here, where security comes from bonded capital and slashing rather than burned energy, a design we walk through in our guide to Ethereum solo staking after Fusaka. Bitcoin buys its security with electricity; the map of where that electricity is spent is, in the end, a map of where the network’s security is manufactured.

The 2028 clock: fees, the halving, and the security budget

There is a countdown running underneath the map. Around April 2028 the block subsidy halves again, from 3.125 to 1.5625 BTC, cutting in half the newly issued coins that fund most of mining today. Each halving raises the bar: to keep the same dollar-denominated security budget, either the price must roughly double or transaction fees must make up the difference. For most of Bitcoin’s history fees have been a rounding error, but the network has had periods, driven by Ordinals inscriptions and Runes token activity, when fees briefly rivaled the subsidy. Readers curious about that fee-generating machinery can see how it works in our field guide to etching and minting a Bitcoin Rune.

The 2028 clock interacts with geography in an unforgiving way. Each halving is a stress test that hits the highest-cost jurisdictions first, because a miner paying frontier electricity prices with an aging fleet is the first to go underwater when the reward drops. That is exactly the dynamic squeezing Ethiopia and the smaller Kazakh operators today, and it will intensify at every halving until fees carry a durable share of the load. The map of 2028 and beyond will therefore skew even harder toward places that combine cheap, stable power with the ability to host AI, because those sites can survive a halving on mining alone and thrive on the AI revenue in between. Growth, in other words, is becoming a test of who can still afford to mine when the subsidy shrinks.

What to watch next

If the 2026 decline is voluntary, then the resumption of growth is a price bet, and the terms are unusually explicit. CoinShares forecasts the network reaching 1.8 ZH by the end of 2026 and 2 ZH by the end of March 2027, but attaches a condition: Bitcoin has to recover toward $100,000. In its Q1 2026 mining report, the firm sketched the scenarios plainly. Sustained below roughly $80,000, the hashprice miners earn falls, higher-cost rigs switch off, and hashrate declines further. Near $100,000, the hashprice recovers toward the high thirties per petahash per day and the climb resumes. A run at the old $126,000 high would push it higher still. The 2 zettahash milestone is not a technical inevitability; it is a wager on the coin’s price.

That makes the watchlist for the rest of 2026 short and specific. Watch Bitcoin’s price against the roughly $78,000 production cost, the line above which the marginal miner keeps hashing and below which it does not. Watch the difficulty retargets for confirmation of who is switching on and off. Watch the flow of AI contract announcements, since each one re-tenants megawatts that would otherwise hash. And watch the cells at the edge of the map, the Gulf and Paraguay building, Iran and Kazakhstan fading. Because the price bet runs through monetary policy, the macro backdrop matters as much as the mining data; our read on the Federal Reserve’s posture in the September countdown is as good a leading indicator for hashrate as any difficulty chart. The map of 2026 is not a picture of decline. It is a picture of relocation, waiting on a price to tell it which way to grow next.

Frequently Asked Questions

What is Bitcoin’s hashrate right now?

In late August 2026 Bitcoin’s network hashrate hovered between roughly 900 exahashes and one zettahash per second, with the steadier seven-day average near 920 EH/s. Single-day spot readings are noisy because hashrate is estimated from block times rather than measured directly, so a range is more honest than a single figure. The network peaked near 1.1 zettahashes per second in late October 2025.

Why did Bitcoin’s hashrate fall in 2026?

For the first time, mining power declined by choice rather than by force. Instead of a price crash or a government ban, artificial intelligence and high-performance computing began paying more for the same electricity, so miners redirected power and whole sites toward AI tenants. CoinShares expects up to 70% of listed miners’ revenue to come from AI by the end of 2026.

Which country has the most Bitcoin mining?

The United States, with about 37.4% of global hashrate, roughly 375 EH/s, in Hashrate Index’s Q2 2026 heatmap, ahead of Russia at 16.9% and China at 12%. The top three together control roughly two-thirds of the network, a ranking that has held steady across consecutive quarters even as capacity migrates underneath it.

Does falling hashrate make Bitcoin less secure?

Less than the headline suggests. When miners leave, difficulty automatically adjusts down so the survivors still find blocks every ten minutes, and the cost to mount a 51% attack remains enormous. Duke’s Campbell Harvey estimates a majority-hashrate attack would cost between $6 billion and $8 billion, a figure that scales with, not against, the network’s value.

Will Bitcoin’s hashrate reach 2 zettahashes?

CoinShares forecasts 1.8 ZH by the end of 2026 and 2 ZH by the end of March 2027, but only if Bitcoin recovers toward $100,000. Growth has become a price bet: below roughly $80,000 the hashprice miners earn falls, higher-cost rigs switch off, and the climb stalls rather than resumes.

By Marcus Okafor, Bitcoin and mining correspondent at HOGE Wire.