Cardano fees covered 0.7% of rewards as transactions drop 72%

A full-chain count by Bitquery found Cardano collected 3,300,000 ADA in transaction fees while paying 493,700,000 ADA in staking rewards across 73 five-day epochs from Sept. 1, 2025, through Sept. 1, 2026. Fees covered about 0.668% of the staking rewards and the reward total was roughly 149.6 times larger than fee revenue.

An official Cardano snapshot for epoch 655 provides an example for a single epoch. Epoch 654 recorded 108,500 transactions and 33,855 ADA in fees over five days, about 21,700 transactions per day and roughly 0.251 transactions per second. For epoch 654, fees equaled about 0.339% of roughly 9.998 million ADA distributed as rewards; that percentage uses a different reward denominator than the 12-month comparison and is not directly interchangeable.

On-chain activity fell sharply. Average daily transactions declined from 90,294 in 2022 to 24,869 for January through August 2026, a 72.46% drop. Using a classification that marks wallets sending at least 3,000 transactions in a month as bots unless their behavior resembles an exchange, the share of transactions attributed to bots rose from 11.5% in 2022 to 32.8% in 2026, with batchers identified as the largest bot subgroup. The counts report on-chain actions rather than unique users: a single automated actor can submit thousands of transactions, a decentralized exchange batcher can process many customers while appearing as one sender, and a wallet that stakes ADA without sending transactions does not appear in sending-wallet counts.

Staking participation changed during the study period. The share of circulating ADA staked fell from 75.6% at the end of 2022 to 58.3% in the last epoch covered by the analysis. The data do not include a direct measure of network security outcomes.

Cardano’s reserve balance in epoch 655 was 6,126,859,027 ADA, equal to 13.62% of the protocol’s 45 billion ADA maximum supply. Under Cardano’s monetary rules, each epoch places transaction fees and 0.3% of the remaining reserve into a virtual pot. Twenty percent of that pot goes to the treasury and the remainder can be used for stake rewards. The reserve is programmed to shrink by a fixed percentage each epoch, producing exponential decay with a documented half-life of roughly four to five years. As the reserve contribution declines, nominal reward outlays can fall.

Holding average fees constant and using the 2026 average of 24,869 daily transactions with the observed 149.6 reward-to-fee ratio produces an illustrative requirement of about 3.72 million transactions per day, or roughly 43.1 transactions per second. Other estimates place the transaction rate needed to replace reserve-funded rewards in a similar range, from the mid-to-high 30s up to the low 50s transactions per second, with a central estimate near 45 TPS. These scenarios are simplified and do not account for the 20% treasury allocation or potential changes to fee parameters.

Work on capacity includes the Leios specification and related proposals. Public testnet updates in August reported roughly sixfold improvement in synthetic throughput in Leios tests. The proposed CIP-164 models sustained capacity above the simplified 43 TPS illustration. A planning document for the Intersect implementation targets code completion in the fourth quarter of 2026; preview, pre-production testing and governance remain outside that schedule and a mainnet hard-fork date has not been set. The reported tests used synthetic traffic and do not show whether real-world user activity will rise to match increased capacity.

The measured data include the size of fee revenue, the scale of staking rewards, transaction volumes, the share of automated transactions, staking participation and the reserve balance. Protocol governance can change fee parameters and network upgrades are designed to raise capacity. Adoption and user activity will affect future fee revenue and transaction volumes.