A scale that breaks intuition
Between July 2021 and December 2025, 25.2 million cryptocurrencies were listed on GeckoTerminal. By the end of 2025, 13.4 million of them were no longer actively traded. That is a 53.2% failure rate across the full five-year window, measured by tokens that recorded at least one trade before going dormant.
The headline figure understates the reality in two ways. First, the dataset excludes tokens that never traded at all, so the true universe of minted-but-failed tokens is larger than 13.4 million. Second, the failure rate accelerated so sharply in 2024 and 2025 that the 53.2% figure averages a slow start with an explosion. 96% of all five-year failures occurred in 2024 and 2025 alone. 2025 by itself accounted for 86.3% (11.6 million tokens).
The acceleration is the story. 2021 recorded 2,584 failures across the entire year. 2024 recorded 1.38 million. 2025 recorded 11.6 million. The 2025 figure is 8.4 times the 2024 figure, and the 2024 figure was already 534 times the 2021 figure. The market did not get worse linearly; it broke open.
What the data actually measures
Three methodology details matter when interpreting the CoinGecko dataset. The first is the trade-floor: only tokens that recorded at least one trade are counted as either alive or failed. Tokens that minted and never traded are excluded entirely. This means the dataset is conservative against the actual scale of failure.
The second is the Pump.fun rule: only Pump.fun tokens that have "graduated" (reached the threshold for full DEX listing) are counted. The vast majority of Pump.fun deployments never graduate and are therefore excluded. By mid-2025, Pump.fun was the source of 80%-plus of new Solana token launches; if non-graduated Pump.fun tokens were included, the failure figure would be multiple times larger.
The third is that "failure" here means "no longer actively traded." It does not require fraud, scam, or formal abandonment. It includes both intentional rug pulls and ordinary attrition (no users, no liquidity, no developer activity). The 11.6 million 2025 figure is therefore a measure of market-survivability collapse, not of fraud rate specifically. The fraud-specific data is on the next page.
The Pump.fun engine
One platform drove the volume. By mid-2025, Pump.fun was the source of more than 80% of new Solana token deployments. The platform reduced the technical barrier to deployment to a single click; no code, no team, no product, no audit. Daily deployment ran at approximately 10,400 new tokens per 24 hours through most of 2025. Daily token death ran at approximately 9,900 in the same window.
The "graduation rate" (the share of Pump.fun tokens that reach a real DEX listing with sustained liquidity) is the platform's own success metric. It dropped below 1% for the first time in February 2026. Of more than 7 million tokens deployed on Pump.fun since 2024, 98.6% are classified as rugs or pumps. Only approximately 97,000 ever held $1,000-plus in liquidity at any point in their lifetime.
Pump.fun is not a fraud platform in the legal sense; it is a deployment platform that exposed the structural reality of permissionless minting. Most tokens have no operating substance to back them. When the infrastructure to deploy a token cheaper than it costs to print a sticker, the bottleneck stops being the mint moment and starts being the verification moment. Without a verification moment, every weak project is funded indistinguishably from every strong one.
The October 10 cascade as inflection point
On October 10, 2025, a 100% China tariff announcement triggered the largest single-day liquidation event in crypto history. $19 billion in leveraged positions were wiped out within 24 hours. The market-wide stress flowed downhill to token survivability immediately.
Daily token failure rate before October 10 was approximately 15,000 per day. After October 10, it was approximately 83,700 per day. The cascade did not create new failure mechanisms; it stress-tested every weak project at once. Q4 2025 closed with 7.7 million token failures, more than the cumulative total of 2021 through 2024 combined, in 90 days.
The point of the October 10 reference is not that a leverage event explains the underlying failure dynamic. It does not. The underlying dynamic was already in place. The leverage event was the trigger that exposed how thin the substrate was beneath the issuance volume. A robust market would have absorbed the shock; a market built on millions of unverified tokens collapsed.
Continue the study
This page covers the scale of the wreckage. The remaining three pages cover the lifecycle of failure stage by stage:
Unverified mints. The mint moment in the current industry has no proof gate. The cost is measured in $17 billion of crypto scams in 2025 and $6 billion of rug pull losses.
No KPI evidence at fundraising. Presales raise tens of millions of dollars on whitepaper claims. The ICO era's 81% one-year failure rate is recurring in 2024 to 2026 with bigger raises and the same structural absence.
Upfront capital release. Even institutional-grade 2025 token launches are collapsing post-TGE because capital releases on day one and delivery is never gated. 84.7% of 2025 launches trade below TGE; tokens that launched at $1 billion-plus FDV have a 0% success rate.
