More Than Half of All Crypto Tokens Are Dead, and Most Died in One Year

The headline figure is 53.2%: more than half of all the cryptocurrencies CoinGecko tracks on GeckoTerminal have failed. It is the number every outlet agrees on, and it is the one to cite.

The scale underneath it is the part worth sitting with. In 2021, 428,383 projects were listed on GeckoTerminal. By 2025 that had grown to nearly 20.2 million (CoinGecko Research; CoinDesk coverage, 14 January 2026). The market did not get 47 times more valuable in four years. It got 47 times easier to enter.

The survival rate that follows the headline is 46.8%, which is our own arithmetic rather than a published figure: 100% minus 53.2%. A coin flip, at best, decides whether a token listed in this period is still alive.

The deaths were not spread evenly. About 86% of every failure in the whole period happened in 2025, roughly 11.6 million tokens, with 7.7 million of them dying in the fourth quarter alone. One year did almost all of the killing.

CoinGecko's year-by-year split makes the concentration unmistakable. Failures across 2021 to 2023 combined account for just 3.4% of the period's total. 2024 added 10.3%, roughly 1.4 million projects. 2025 alone carried 86.3%, and its fourth quarter by itself was 34.9% of every failure in five years.

Two things concentrated the damage. First, a market event: the $19 billion of leveraged positions wiped out in 24 hours on 10 October 2025, the largest single-day deleveraging in crypto history, triggered a downturn that took Q4's 7.7 million tokens with it. Second, a supply problem. Easy-launch platforms made spinning up a token nearly free, which is what turned 428,383 listings into 20.2 million.

When launching costs almost nothing, the graveyard grows faster than the living population. The relevant point for a marketer is that the flood is why attention is scarce. You are not competing with a handful of tokens for a buyer's notice. You are competing with millions of dead and dying ones.

The same dataset says three different things depending on which window you look at. Pick one below and the field redraws: every slab is one percentage point, and the ones lying flat are the share no longer traded.

Tokens overwhelmingly die from being unseen, not from broken technology. No distribution, no liquidity, and no community that outlasts the launch spike. The mortality data is a distribution story wearing a technology costume.

The clearest evidence is what happens to the tokens that were designed to extract money. Chainalysis identified 90,408 tokens on Ethereum in 2023 matching a pump-and-dump pattern, which was 24.4% of every token launched on the chain that year and 53.6% of the ones that made it onto a DEX at all. Together they produced $241.6 million in profit. Spread across the tokens, that is an average of just $2,672 each, and the whole set accounted for 1.3% of Ethereum DEX trading volume for the year (Chainalysis).

Read that again, because it is the whole argument. Even the tokens built by people actively trying to manufacture attention mostly failed to get any. Ninety thousand deliberate attempts, and the average one cleared less than three thousand dollars. If manufactured attention does not work at scale, an honest project that simply ships and hopes has no mechanism at all.

The low-effort flood of 2025 makes the same point from the other direction. Those projects did not fail because they were technically broken. Many were technically fine and completely ignored, which is a different problem with a different fix. The jump from roughly 1.4 million failures in 2024 to 11.6 million in 2025 did not come from code getting worse in twelve months. It came from the cost of launching falling far enough that anyone could add to the pile.

A 53.2% death rate, concentrated 86% in one year, is the clearest evidence available that distribution decides survival rather than the launch. Founders read the number as a warning. They should read it as a brief.

Here is the logic in one line. If launching is trivial and nearly free, then launching cannot be the differentiator, because everyone can do it, which is exactly why millions did it and died in a single year. The scarce, decisive input is attention.

That reframes every downstream decision. The budget question stops being "how do we launch" and becomes "how do we get seen once we have", which makes the how to market a token launch playbook matter more than the tokenomics deck. It does not mean marketing guarantees survival. Plenty of well-marketed tokens still fail. But the data makes the reverse close to certain: a token that is never distributed is a token that joins the 53.2%.

The tokens that survive are disproportionately the ones that reached an audience that already exists, because the audience is not the bottleneck. Distribution is.

An estimated 741 million people owned crypto in 2025, up from 659 million the year before, including 365 million Bitcoin owners and 175 million Ethereum owners (Crypto.com Research, February 2026). That figure is built from on-chain data with blended parameters rather than a survey, so treat it as directional rather than precise. The point survives the caveat: the buyers are there, and most dead tokens simply never got in front of them.

The economics of getting seen are why distribution is a solvable problem rather than a budget black hole. Forbes reported clipping-style distribution running at roughly $1 to $5 CPM against $20 to $80 for traditional paid social. In the same piece, Pudgy Penguins' Attari describes a crossover with a TikTok meme trend generating "around 250 million impressions in less than two weeks... all inspired of the edits that we had seeded" (Forbes, 23 July 2026). Both are figures given by the operators involved rather than independently audited, and the 250 million came from fan-made edits rather than a paid campaign, so read it as what seeding can unlock rather than a rate card.

The environment those clips land in is adversarial, which is part of why organic reach is hard to buy your way past. Chainalysis found impersonation tactics grew 1,400% year on year in 2025, with the average scam payment rising from $782 to $2,764 (Chainalysis). Attention is contested by people with worse intentions and better tooling than most founders.

Paid ads are also often closed or gated for crypto anyway, and where you can and can't run crypto ads maps which platforms and markets are even open. That pushes surviving tokens toward organic, creator-led distribution by necessity rather than preference.

· CoinGecko Research, "Dead coins: How many cryptocurrencies have failed?" — the study: 53.2% of tracked cryptocurrencies on GeckoTerminal have failed; listings grew from 428,383 in 2021 to nearly 20.2 million by 2025; 2021 to 2023 account for 3.4% of failures, 2024 for 10.3% (roughly 1.4 million), 2025 for 86.3% (roughly 11.6 million); Q4 2025 alone was 7.7 million, or 34.9% of all failures; $19 billion of leveraged positions wiped out in 24 hours on 10 October 2025, the largest single-day deleveraging in crypto history. Reported by CoinDesk, 14 January 2026.

· Chainalysis, pump-and-dump analysis (2023 data) — the study: 90,408 tokens on Ethereum met the criteria, 24.4% of all tokens launched on the chain and 53.6% of those listed on a DEX; $241.6 million in total profit; an average of $2,672 per token; 1.3% of Ethereum DEX trading volume for the year.

· Chainalysis, 2026 Crypto Crime Report — scams section: impersonation tactics grew 1,400% year on year, with the average scam payment rising from $782 in 2024 to $2,764 in 2025, a 253% increase.

· Crypto.com Research — global cryptocurrency ownership reached 741 million in 2025, up 12.4% from 659 million in 2024, with 365 million Bitcoin owners and 175 million Ethereum owners. Built from on-chain data with blended parameters rather than a survey, so directional only.

· Forbes, "One Sign Up: The $30,000 Bill That Ended Crypto's Influencer Era" — 23 July 2026: clipping distribution at roughly $1 to $5 CPM against $20 to $80 for traditional paid social; Pudgy Penguins' Attari on a meme crossover generating around 250 million impressions in under two weeks from seeded edits. The article quotes Lumina Clippers founder Rhys McKay. Figures in it are given by the operators involved rather than independently audited.

· On method. Every figure above was checked against the publishing organisation's own page before it was used here. Where a publisher's site could not be opened directly from our environment, the check was made against that publisher's own text as indexed, not against a secondary write-up of it.