Malaysia seized 75,000 mining rigs. The grid wars are real

In one hemisphere, Bitcoin miners are signing $6.6 billion leases with tech giants. In the other, police are crushing their machines with steamrollers. The same hardware, the same algorithm, the same electricity hunger, and two opposite fates decided by one variable: the price of power.

The Royal Malaysia Police have developed a signature move for concluding a cryptocurrency mining investigation: they line the confiscated machines up in a parking lot, invite the press, and drive a steamroller over them. The spectacle has run several times over the years, thousands of ASICs cracking under the drum, and it returned to relevance this month when Deputy Home Minister Shamsul Anuar gave parliament the campaign’s cumulative arithmetic. Between 2022 and May 2026, Malaysian authorities seized 75,578 mining rigs in 3,049 raids and arrested 629 people, in coordinated operations with the national utility Tenaga Nasional Berhad. The minister was answering a question about why prosecutions remain scarce, and his numbers, read carefully, answer a bigger one.

Seventy-five thousand machines and six hundred arrests is not the record of a crime being defeated. It is the record of an economy being policed, and the economy in question, the relentless global arbitrage between the price of electricity and the price of Bitcoin, is the same one currently making American miners rich enough to become the legal side of the same asset. The steamroller and the $6.6 billion>What Malaysia is actually fighting

Start with the precision of the Malaysian position, because it is more coherent than the headlines suggest.

Crypto is not illegal in Malaysia. Owning it, trading it on registered exchanges under the Securities Commission’s oversight, even mining it, all permitted. What is illegal is the way industrial-scale mining actually gets done there: bypassing electricity meters, tampering with connections, tapping distribution lines directly, and running unlicensed operations in shophouses, factories, and residential buildings wired to draw constant heavy load without paying for it. The crackdown, as the government frames it and as the raid targets confirm, is an electricity-theft enforcement campaign that happens to be about crypto, run jointly by police, local authorities, and TNB, the utility eating the losses.

The losses are the story’s real number. In late 2025, Malaysia’s energy ministry linked roughly $1.1 billion in power losses to about 14,000 illegal mining sites uncovered over five years, and TNB’s own data shows theft cases tied to mining rising roughly 300% between 2018 and 2024, from 610 detected cases to 2,397. The mechanics explain the growth: a mining rig runs around the clock at constant draw, which makes stolen power the single largest input cost eliminated, and a tampered meter hides the consumption until the utility notices the gap between what a neighborhood bills and what it burns. Enforcement has accordingly become a data problem. The ministry’s newer approach, a committee pulling in the finance ministry and the central bank alongside TNB, plus what the deputy minister described as intelligence-led targeting of high-risk areas before raids, treats illegal mining less like street crime and more like systematic fraud against national infrastructure, which is what it is.

The theatrical steamrolling, whatever its deterrence value, is the tell about the campaign’s frustration. Utilities recover stolen-power losses slowly if ever; prosecutions, as the parliamentary questioner pointed out, lag far behind seizures; and the machines themselves are cheap enough, especially aging models displaced from industrial farms elsewhere, that destroying them theatrically is partly a substitute for consequences the courts have not delivered. Six hundred twenty-nine arrests across four years of raids that seized seventy-five thousand machines is one arrest per hundred and twenty rigs. From a miner’s chair, that is not a deterrent. It is a cost of doing business, priced well below the electricity being stolen.

The arbitrage that never stops

To see why the raids continue finding rigs, stop thinking of mining as a crypto activity and start thinking of it as the economics underneath the raids: a machine for converting electricity price differences into money.

A Bitcoin miner earns the same revenue per unit of hashrate anywhere on earth; the network does not care where the hash came from. Costs, by contrast, are almost entirely electricity, which varies by an order of magnitude across the world depending on generation mix, subsidy policy, and enforcement. The entire global industry is therefore one continuous migration toward cheap electrons, and every source of cheapness gets arbitraged: stranded hydro in Sichuan until China banned it, flared gas in Texas, geothermal in Iceland, subsidized residential tariffs anywhere they exist, and, at the bottom of the ladder, electricity that costs nothing because it is stolen. Malaysia sits in the crosshairs for a specific reason: its residential and certain industrial tariffs are subsidized well below regional market cost, which means every kilowatt-hour a hidden mine draws is partly funded by the Malaysian treasury, and a stolen one entirely by TNB. Cheap-by-policy power plus tropical-grade enforcement gaps is the exact habitat the arbitrage seeks.

This is also why enforcement behaves like squeezing a balloon. China’s 2021 mining ban, the largest enforcement action in the industry’s history, did not reduce global hashrate for long; it redistributed it, to the United States, Kazakhstan, Russia, and a long gray tail across Southeast Asia. Kazakhstan welcomed miners until they strained its grid and the state cracked down; Iran has oscillated between licensing mining for export revenue and blaming it for blackouts; Kuwait, with some of the world’s most heavily subsidized power, ran its own seizure campaigns when hidden farms began distorting neighborhood load. The pattern repeats because the incentive is structural: wherever the local price of power is held below its market value, by subsidy or by theft, mining will arrive to harvest the difference, and it will keep arriving as long as the difference exceeds the expected penalty. Malaysia’s 629 arrests say the expected penalty is low. Bitcoin above $60,000 says what actually drives the mining revenue line is not.

The rigs themselves complete the loop. Every halving and efficiency generation pushes older ASICs out of profitability in fully-priced power markets, and those machines do not retire; they flow downmarket, sold by the pallet into exactly the jurisdictions where power is subsidized or stealable, because stolen electricity is the only input cheap enough to keep an S19 competitive in 2026. Malaysia’s seizure warehouses are, among other things, the terminal destination of the industrial mining fleet’s depreciation curve.

The other side of the same war

Now hold the Malaysian picture against what the same industry looks like where the electricity is legal, contracted, and priced, because the contrast is the thesis.

In the United States this month, CleanSpark, a Bitcoin miner, signed a 20-year triple-net lease with an investment-grade technology tenant for an AI data center campus in Georgia: $6.6 billion in contracted revenue, $11.6 billion with extensions, roughly $330 million a year in net operating income at nearly full margin, with first deliveries in late 2027 and an exclusivity agreement covering the company’s entire 885-megawatt Texas portfolio behind it. Days earlier, MARA agreed to acquire a 1,200-acre Texas site with access to up to two gigawatts, explicitly for mining and AI compute together. The market repriced these companies overnight, because it understood what had actually been sold: not hashrate, but powered land, grid interconnection, substations, and megawatt capacity, assets the AI industry is desperate for and cannot build quickly.

The comparison with Malaysia is almost unfairly clean. The asset is identical in kind, access to large volumes of electricity, and its legal status determines everything about its value. A Malaysian operator’s access to power is stolen, so its terminal value is a steamroller. An American miner’s access is contracted through interconnection queues and power-purchase agreements, so its terminal value is a twenty-year lease with a tech giant. The Bitcoin miners of Georgia and the mining syndicates of Johor are running the same machine economics on the same network, and the entire difference between a $6.6 billion balance sheet event and a parking-lot crushing is whether the electrons were bought or taken.

That difference also explains why the two stories are converging into one policy question. Grid operators everywhere, TNB included, are discovering that the scarce resource of the next decade is not generation in the abstract but deliverable, sited, interconnected power, and that a queue of buyers, AI data centers, industrial electrification, and miners legal and otherwise, is forming for every megawatt. In that world, hidden mines are not a quirky crypto crime; they are unauthorized withdrawals from the most contested resource in the economy, which is exactly why Malaysia’s enforcement is escalating from utility-fraud policing toward a whole-of-government committee with the central bank at the table.

The case that enforcement is winning

The optimistic reading of Malaysia’s campaign has real evidence behind it, and it starts with what the numbers do show.

Detection is compounding. The 300% rise in identified theft cases between 2018 and 2024 reflects better instrumentation as much as more mining: smart-meter rollouts, load-pattern analytics that flag the unmistakable flat, constant draw of a mining operation, and the shift the deputy minister described toward intelligence-led raids instead of reactive ones. A hidden mine is only economic while it stays hidden, and every improvement in grid telemetry shortens its life expectancy. The 75,000 seized rigs are, on this reading, the visible output of a detection machine that is genuinely improving, and the reason seizures keep rising is that discovery keeps accelerating, not that the problem keeps growing proportionally.

The institutional escalation matters too. Moving the issue from TNB’s fraud department to a standing committee with the finance ministry and Bank Negara reframes illegal mining as an economy-scale drain, not scattered utility theft, which unlocks tools policing never had: financial tracing of operators, pressure on the payment and exchange channels that monetize the mined coins, and reading the market the miners sell into more closely. The prosecution gap the parliamentary question highlighted is precisely what this machinery is built to close, and enforcement campaigns of this type, against organized electricity theft, unlicensed gambling networks, and similar infrastructure crime, historically bite when the financial layer gets involved.

And the regional context is not standing still. As grid demand from AI and industry grows across Southeast Asia, the tolerance for parasitic load shrinks everywhere at once, and the gray-market mining tail is being squeezed simultaneously in multiple jurisdictions. The balloon metaphor cuts both ways: when every neighboring country is squeezing at once, the displaced capacity has fewer places to go, and some of it simply exits, sold for scrap or shipped to the shrinking set of jurisdictions that still look the other way.

The case that the raids are theater

The skeptical reading concedes every number above and draws the opposite conclusion, because the numbers describe effort, not outcomes.

Begin with the ratio that will not go away: one arrest per hundred and twenty seized machines, across four years. Machines are the cheapest, most replaceable component of an illegal mining operation; the scarce inputs are the operators, the corrupt or negligent property arrangements, and the monetization channels, and the enforcement record on all three is thin. An operator who loses a container of depreciated ASICs to a raid and faces a 0.8% personal arrest risk per hundred machines has experienced a tax, not a deterrent, and the five-year growth of the loss estimates, to $1.1 billion and 14,000 sites, is the market’s verdict on the tax rate. The deputy minister’s own framing to parliament, that potential profits do not excuse the crime, is a moral statement precisely because the economic one runs the other way.

The subsidy engine also stays running. Malaysia’s below-market tariffs are the structural attractant, and they are politically untouchable in any near term; as long as the gap between subsidized or stolen power and market power exceeds mining’s margin, the habitat regenerates after every raid. Enforcement that removes machines without repricing the incentive is gardening, not eradication, and the global supply of displaced, nearly-free older ASICs guarantees the replanting stock. Every efficiency generation shipped by Bitmain adds to the pile of hardware whose only profitable home is somewhere with unpriced electricity.

And the sharpest version of the critique points at the counterfactual Malaysia is not pursuing. The same grid connections, industrial sites, and cheap generation that attract illegal miners are, as the American example shows, monetizable legally, through licensed mining zones with export-priced tariffs, through>What to watch

Three signals will show which reading is winning, in Malaysia and in the wider war.

Prosecutions, not seizures. The rig count will keep rising with detection either way; the number that would mark a regime change is convictions of operators and financiers, and asset recovery through the new committee’s financial channels. Watch whether the 629 becomes a curve or stays a rounding error.

Tariff and licensing reform. Any move toward licensed mining zones, unsubsidized industrial tariffs for compute, or target=”_blank”>where the compute race goes next.

The 75,578 machines in Malaysia’s warehouses are one national chapter of a planetary story: an algorithm that pays a bounty, everywhere and always, for the world’s cheapest electricity, and a hundred governments deciding one by one whether to sell it, tax it, or fight it. The Americans decided to sell, and their miners are becoming AI’s landlords. Malaysia, so far, has decided to fight, and its utility is becoming the arbitrage’s unwilling sponsor. The machines are indifferent. They will run wherever the electrons are cheapest, and the only question any government actually controls is whether that cheapness shows up on a contract or on a crime report.

Frequently asked questions