138 TWh
Estimated annual Bitcoin electricity consumption
A sourced examination
From grids strained by proof-of-work mining to proof-of-stake collapses that wiped out billions, from NFT bubbles to rug pulls and pig-butchering scams, the harms are documented, measurable, and repeatedly repeated. This page collects the evidence.
This is an educational critique, not financial advice. Claims are linked to public sources below.
01
Bitcoin is the clearest case. Its proof-of-work protocol requires miners to burn electricity solving cryptographic puzzles that add no practical value beyond securing a speculative ledger. The Cambridge Centre for Alternative Finance estimated in 2025 that Bitcoin consumed 138 terawatt-hours annually, about 0.5% of the world’s electricity, producing 39.8 megatonnes of CO₂ [2].
That is not a rounding error. It is a national-scale energy load competing with hospitals, homes, industry, and climate mitigation. A 2025 study found that the 34 largest U.S. Bitcoin mines consumed 32.3 TWh in one year — 33% more than Los Angeles — and that fossil fuel plants generated 85% of the increased demand from those mines [2].
The cost is not only carbon. Bitcoin mining also creates electronic waste, consumes water for cooling and power generation, and occupies land. The Cambridge report estimated 2024 Bitcoin e-waste at 2,300 tonnes, while earlier research estimated annual e-waste above 30,000 tonnes under different assumptions [2] [3].
The core problem
The energy is not paid for by the people who suffer the climate, grid, and community impacts. It is embedded in the price of a speculative asset and externalized to the public.
02
Bitcoin’s protocol adjusts mining difficulty roughly every two weeks to keep block production near ten minutes [4]. That creates an arms race: when Bitcoin’s price rises, more miners deploy more hardware, hashrate increases, difficulty increases, and energy consumption rises with it [5].
There is no protocol-level cap on total power demand. Security is purchased with electricity. The network does not ask whether the grid can afford it, whether the electrons displace clean energy available to the public, or whether the asset being secured has productive value.
The result is a system in which the cost of “trust” is continuously paid in fossil-fuel-heavy electricity, specialized hardware, and electronic waste — while the benefit is the ability to move speculative value pseudonymously across borders [2] [3].
Difficulty adjustment means the network automatically absorbs more computing power as profitability rises. The energy bill scales with speculation.
Miners must constantly replace ASICs as efficiency improves. One study estimated average mining hardware lifespans of 1.3 years before becoming unprofitable [2].
03
Proof-of-stake systems reduce mining energy, but they do not remove the underlying incentive structure: wealth and governance become concentrated among early token holders, founders, validators, and affiliated firms [6].
The Terra/Luna collapse is the canonical example. Terra was a proof-of-stake ecosystem whose stablecoin UST was backed by LUNA tokens. When confidence broke in May 2022, the feedback loop destroyed roughly $45 billion in market value in about a week [7].
Critics had warned that Anchor Protocol’s 19.45% annual yield was unsustainable and effectively paid retail investors with new issuance. Founder Do Kwon’s share pool expanded to 91.7% of the company, illustrating how control can concentrate before the collapse. Kwon was later sentenced to 15 years in prison, and the SEC pursued a $4.5 billion judgment against him following a fraud verdict [7] [8].
The pattern
Proof of stake can make a system more energy-efficient while making insider extraction, governance capture, and cascading collapse more dangerous.
04
FTX was not a protocol; it was an exchange. But it became one of the clearest demonstrations that cryptocurrency’s promised transparency often masks ordinary financial fraud. FTX collapsed in November 2022 after reports that billions of dollars in customer assets had been lent to its affiliated trading firm Alameda Research [9].
At its peak, FTX was valued at about $32 billion. Investigations found roughly $8 billion in missing customer funds and about $10 billion lent to Alameda. A separate $473 million SIM-swap hack showed how fragile the security of “decentralized” finance really was [9] [10].
The lesson is not that one founder was evil. The lesson is that the industry’s culture — anonymous insiders, opaque related-party lending, exotic derivatives, and weak regulation — makes large-scale fraud easier to build and harder to detect.
05
Non-fungible tokens turned digital files into speculative assets by certifying a token as “unique” or “authentic” on a blockchain. The economic result was a bubble. NFT trading volume grew from about $82 million in 2020 to $17 billion in 2021, then collapsed in 2022 as sales fell by more than 90% [11].
By September 2023, more than 95% of NFT collections had zero monetary value, and 79% had never been sold at all [11]. The “digital ownership” narrative did not survive contact with reality: ownership of a token is not ownership of the underlying image, and off-chain links rot.
NFTs also amplified existing harms. When minted or traded on proof-of-work chains, they inherit Bitcoin-like energy costs. Even on lighter chains, the pattern remains: artificial scarcity, celebrity hype, plagiarism, scams, money-laundering risk, and a final transfer of value from late retail buyers to early insiders [11] [2].
Sold for $2.9 million in 2021. By 2023, the top bid had fallen to $280. A perfect illustration of speculative price discovery without underlying cash flow.
Digital files can be copied infinitely. NFTs create scarcity not in the asset, but in the token pointing to it.
Minting, trading, and gas fees consume blockchain resources. On energy-intensive chains, the environmental cost is direct.
06
A rug pull is a confidence trick: promoters create a token, generate hype, attract liquidity, and then abandon the project or drain the funds. In an industry with weak regulation and pseudonymous insiders, rug pulls are not an edge case — they are a recurring business model [12] [13].
The scale is documented. In 2017, Satis Group estimated that 80% of ICOs were scams. By 2019, estimated losses from exit scams exceeded $4.3 billion [12]. Academic research has since catalogued token spammers, sniper bots, and coordinated rug-pull patterns on Ethereum and BNB Chain [12].
Why it keeps working
New tokens are easy to create, marketing is easy to fake, wallets are irreversible, and victims often have no legal recourse. The system optimizes for speed and anonymity, not accountability.
07
Cryptocurrency’s most human harm is not abstract. It is the victim who loses savings to a scam. Pig-butchering scams — also called romance investment scams — are among the most damaging. Fraudsters build trust over weeks or months, often through romance or friendship, then steer victims toward fake trading platforms tied to cryptocurrency [14].
These scams target people who are isolated, financially stressed, romantically hopeful, or new to investing. The cryptocurrency angle is not incidental: it provides the illusion of a modern trading opportunity, the pressure of “limited-time” returns, and a rails system that is hard to reverse once funds are sent.
When a technology’s growth depends on recruiting newer participants to pay earlier ones, and its marketing exploits trust, loneliness, and greed, it behaves less like finance and more like a predatory parasite.
08
A cancer is not merely inefficient. It consumes resources, distorts incentives, damages healthy tissue, and spreads by exploiting weaknesses in the host. Cryptocurrency shares those traits:
None of this requires believing that every participant is malicious. It only requires observing that the system’s dominant, documented outcomes are waste, extraction, and harm.
09
The following public sources support the claims above. Wikipedia entries are used where they aggregate primary reporting, court documents, academic studies, or official filings; primary reports are linked directly where available.