← All research Token Supply Dynamics
Token Supply Dynamics

Ethereum activity expanded while network fee burn collapsed

Between January 2025 and July 2026, Ethereum L1 monthly transactions rose 111% while monthly USD fee burn contracted 97%. The relationship between activity and burn did not weaken gradually — and one month in the series breaks it further than capacity expansion accounts for.

Jan 1, 2025 – Jul 31, 2026 (19 complete monthly periods) snapshot
Data as of Aug 29, 2026
77.84ML1 transactions · Jul 2026+111.41%
8.75MActive addresses · Jul 2026+42.97%
$2.90MMonthly fee burn · Jul 2026−96.71%
$0.0373Burn per transaction · Jul 2026−98.44%
CategoryToken Supply Dynamics
WindowJan 2025 – Jul 2026
Primary sourceDefiLlama chain analytics
Reconciled againstToken Terminal, Artemis
Periods19 complete months

Key findings

  • Throughput roughly doubled.* Monthly L1 transactions grew from 36.82M in Jan 2025 to 77.84M in Jul 2026, a 111.41% increase, while active addresses rose 42.97%.
  • Fee burn collapsed.* Monthly burned value fell from $88.20M to $2.90M over the same window — a 96.71% decline.
  • Unit capture compressed further than the aggregate.* Burn per transaction fell 88.67% year-over-year in H1, from $1.71 to $0.19, because volume rose while burn fell.
  • One month does not fit.* The burned share of fees held between 68.8% and 74.7% for eighteen months, then registered 26.6% in July 2026. That break is flagged below rather than absorbed into the trend.

01 · Execution throughputVolume decouples from network burn

Indexed to January 2025, the divergence is unambiguous. Transactions closed the window at 211 against a base of 100; chain revenue closed at 3.3. Active addresses rose alongside transactions, so the growth is not a small set of addresses transacting more — participation widened with it.

indexed_activity_vs_burn.chart
INDEXED NETWORK ACTIVITY VS PROTOCOL FEE BURN · JAN 2025 = 1002501250JAN 2025OCT 2025JUL 2026ACTIVE ADDRESSES143+43%TRANSACTIONS211.4+111%FEE BURN3.3−97%
The read
Capacity expansions across the period increased available blockspace and reduced congestion, lowering the clearing price of execution faster than throughput increased. More activity settled on Ethereum, and each unit of it burned less ETH. This is an association between observable trends, not a demonstrated cause: this dataset records what was burned, not why the fee market cleared where it did.

02 · Unit economicsRevenue captured per transaction

Dividing burn by transaction count isolates what a single unit of activity contributed. Gross fees per transaction is retained as subordinate context — the gap between the two lines is the share of each fee not burned.

per_transaction_capture.chart
USD PER EXECUTED L1 TRANSACTION · MONTHLY$3.5$1.8$0JAN 2025OCT 2025JUL 2026FEES / TX$0.14BURN / TX$0.0373−98%

03 · Structural comparisonH1 2025 against H1 2026

Half-year aggregates are the more reliable frame than endpoint-to-endpoint change: they average monthly volatility and, in this case, exclude the July anomaly entirely.

MetricH1 2025H1 2026Δ
TransactionsL1 transactions 241.70M 415.50M +71.9%
Active addressesmonthly average 6.51M 8.52M +30.9%
Chain feesgross USD $558.6M $109.6M −80.4%
Fee burnUSD destroyed $414.0M $80.6M −80.5%
Burn per transactionUSD $1.71 $0.19 −88.7%

04 · The anomalyOne month does not fit the series

The burned share of total fees is normally stable. EIP-1559 burns the base fee, so the ratio moves only with the mix of base fee against priority tips and blob costs. Across the first eighteen months it stayed inside a 5.9-point band.

burn_share_of_fees.chart
FEE BURN AS A SHARE OF GROSS CHAIN FEES, %100%50%0%JAN 2025OCT 2025JUL 2026BURNED SHARE26.6%26.6%

July 2026 registered 26.6% — 27 standard deviations below the preceding eighteen-month mean of 72.8%. The month is internally inconsistent with the rest of the series: transactions rose while fees and burn moved by very different magnitudes, which a stable burn share does not permit.

MetricJun 2026Jul 2026Δ
Transactionsmonthly 74.53M 77.84M +4.4%
Chain feesgross USD $14.8M $10.9M −26.4%
Fee burnUSD destroyed $10.6M $2.9M −72.6%
Burned share of feespercent 71.6% 26.6% −45.0pp
What we can and cannot sayThis dataset cannot attribute the break to a cause. Candidate explanations include a genuine change in fee-market composition, a definitional or reporting change at the source, or an incomplete July aggregation. Separating them requires block-level base-fee and blob-fee series that a monthly rollup does not contain. The consequence is quantifiable. Had July burned at the eighteen-month mean share, burn would have been $7.94M rather than $2.90M, and the Jan 2025 to Jul 2026 change would read −91.0% instead of −96.71%. One unexplained month carries 5.7 percentage points of the endpoint headline, which is why the H1 comparison above is the figure to rely on.

05 · The full datasetEvery month, published

Nineteen complete monthly periods, as used in every figure above. Burn share is derived; all other columns are as sourced.

MonthTransactionsActive addrFeesBurnFee/txBurn/txBurn share
Jan 2536.82M6.12M$118.4M$88.2M$3.2156$2.395474.5%
Feb 2534.51M5.89M$102.3M$75.8M$2.9644$2.196574.1%
Mar 2538.24M6.45M$110.5M$82.1M$2.8896$2.147074.3%
Apr 2537.10M6.31M$86.2M$63.9M$2.3235$1.722474.1%
May 2545.62M6.98M$75.8M$55.8M$1.6616$1.223173.6%
Jun 2549.41M7.30M$65.4M$48.2M$1.3236$0.975573.7%
Jul 2552.14M7.55M$58.9M$43.1M$1.1297$0.826673.2%
Aug 2555.80M7.82M$54.2M$39.5M$0.9713$0.707972.9%
Sep 2551.23M7.41M$49.8M$35.9M$0.9721$0.700872.1%
Oct 2548.72M7.15M$44.1M$31.2M$0.9052$0.640470.7%
Nov 2546.35M6.94M$39.6M$27.5M$0.8544$0.593369.4%
Dec 2550.02M7.48M$42.3M$29.1M$0.8457$0.581868.8%
Jan 2670.01M8.82M$32.8M$24.5M$0.4685$0.350074.7%
Feb 2663.78M8.24M$22.5M$16.7M$0.3528$0.261874.2%
Mar 2666.61M8.56M$15.4M$11.3M$0.2312$0.169673.4%
Apr 2668.42M8.41M$11.6M$8.4M$0.1695$0.122872.4%
May 2672.15M8.69M$12.5M$9.1M$0.1733$0.126172.8%
Jun 2674.53M8.39M$14.8M$10.6M$0.1986$0.142271.6%
Jul 2677.84M8.75M$10.9M$2.9M$0.1400$0.037326.6%

06 · What to monitor nextFour signals

Methodology and material scopeMonthly aggregation, not a live dashboard, and not investment advice. Protocol revenue accounts strictly for ETH permanently burned via EIP-1559 and EIP-4844 blob mechanisms — it is economic destruction, not retained cash flow, and no entity receives it. Layer 2 sequencer margins and application-level take-rates are excluded. USD figures move with the ETH price, so dollar declines combine changes in burned quantity with changes in price; this dataset cannot separate the two. Active addresses count distinct senders and receivers including contracts, and are not people. August 2026 is excluded as incomplete. Indexed series share a single axis deliberately — no dual axes are used, because plotting different units against separate axes invites false visual correlation.
Primary metricMonthly L1 transactions, distinct active addresses, chain fees (total gas paid, including EIP-1559 base fees, priority tips and EIP-4844 blob fees) and chain revenue (the portion permanently burned). Per-transaction and indexed series derived; index baseline Jan 2025 = 100
Data sourceDefiLlama chain analytics, block-level ledger aggregations indexed daily and rolled to monthly totals. Trajectories and monthly counts cross-checked against Token Terminal and Artemis Analytics; definitional differences reconciled by anchoring to protocol-level burns rather than application-layer fees
Assumptions & boundsExcludes August 2026 as an incomplete period; reverted transactions that consumed gas without state change; L2 sequencer margins, rollup execution gas and application take-rates; and bridged mint/burn flows that would double-count cross-chain transfers

More breakdowns on our channels:

Need this level of detail on your own protocol?

This is the standard we hold public research to. Bespoke engagements get the same lineage and the same cross-checking, built around your questions.

hello@dashmode.io

Talk to our team