The Float Report · live balances · as of 2026-07-03

$21.6M is all that's sitting on crypto cards.

We read the live stablecoin balance of ~580K card wallets across 7 chains. Almost all of them are empty — crypto cards are payment rails, not places people keep money.

100 typical wallets15% hold funds

85% hold ~$0 · median balance $0 · the money that exists is held by a tiny few.

$21.6M
idle right now
580K
wallets read
85%
hold ~$0
33%
held by 267 wallets
~2.5d
of spend in float

The inversion

Half a million empty wallets — and a few hundred hold it all.

Every card wallet, grouped by how much it holds — richest tier first. The bar is each group's share of the total money. Notice the wallet counts next to it.

Balance heldWalletsMoney heldShare of the money
>$10K267$7.16M
33%
$1K–10K3,014$7.92M
37%
$100–1K16,430$4.39M
20%
$10–10049,921$2.05M
9%
$1–1017,485$0.08M
<1%
$0 (empty)492,809$0
<1%
85%
of wallets hold $0
$0
median balance across all wallets
267 wallets
hold 33% of all the money

People don't keep money on crypto cards — they spend through them. The float that exists is a handful of whales, not a habit.

Why the exception matters

People hold money the moment it pays them to.

Every card spends down to zero — except Plasma One, where 87% keep a balance. The difference? Plasma's balance earns yield. Give money a reason to sit, and behavior flips from rails to bank account. It's not that people reject holding crypto — a bare card balance is just a dumb place to keep it.

87%
of Plasma One wallets hold funds
~1%
of a typical no-yield card
Plasma's 10.8K wallets hold as much ($5.1M) as RedotPay's 344K.

By card

Funds held, card by card

Averages are whale-skewed, so the median holder is the honest "typical user". The bar shows the share of each card's wallets that hold anything.

CardWallets scannedTotal heldMedian holderHolding
Rain-family
Tria, Kolo, Avici, Ready & Tuyo — five cards on Rain's shared infrastructure. Their balances can't be split per brand (Rain keeps that mapping private), so we show the group.
25,814$9.04M$4171%
Plasma One
Yield-bearing balance you spend from — a checking account, not spend-down.
10,832$5.07M$9287%
RedotPay
344,040$3.73M$4013%
Karta
23,852$2.67M$845%
MetaMask
4,831$600K$3421%
Fiat24
65,104$350K$71%
Wirex
32,111$70K$991%
KAST
Custodial — this is funder wallets, not the card balance (KAST pools that off-chain), so it's directional.
8,331$70K$499%
Total514,915$21.60M

EtherFi and Gnosis are excluded from the table above — they are credit cards that spend against collateral (staked assets held elsewhere), so no stablecoin ever sits in the card wallet. A zero balance is their architecture, not empty users; their cardholders are among the most active spenders we track — they just don't hold.

By chain

Where the money rests.

Plasma punches far above its size — 10.8K wallets hold as much as Base's 169K, because its balances earn yield.

Base$5.17M · 168,812 wallets
Plasma$5.07M · 10,832 wallets
Arbitrum$4.19M · 9,772 wallets
Polygon$3.37M · 9,263 wallets
Tron$2.73M · 176,454 wallets
BSC$1.07M · 175,917 wallets

Methodology

How we measured it.

01

We've identified ~900K card wallets and read live balances for ~580K of them (the crawler runs highest-value-first, so every large balance is captured; the uncrawled tail is low-value). The table shows 515K — the 580K minus 64K EtherFi/Gnosis credit-card wallets that hold nothing by design.

02

Balances are the live stablecoin holding (USDC/USDT/USDT0/DAI) across Base, Plasma, Arbitrum, Polygon, Tron, BSC and Scroll.

03

Treasury, aggregator and settlement-relayer wallets are excluded: a wallet with a huge balance and near-zero card activity is infrastructure, not a user. This filter removed millions in fake 'user' balances, including 21 Rain settler addresses.

04

Whale balances were forensically verified as real users — diverse token holdings, independent funding, no exchange or treasury labels — not infrastructure.

05

Custodial cards (RedotPay, KAST) pool card balances off-chain, so their users' on-chain wallets are structurally near-empty; their low share is the model, not poverty.

06

Balances are a point-in-time snapshot, refreshed hourly by our crawler. Averages are whale-influenced, so we show the median holder too.

On-chain observational analysis; not investment advice. See also cardholder intelligence, card economics and the card spend tracker.