[Discussion] TRON Quarterly Transaction Fees Review — Q3 2026

Open
#941 5 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Assessment

Difficulty
2/5
Estimated time
1-3 hours
Newbie friendliness
35/100
Issue type
Documentation
Clarity
Needs clarification
Activity status
Active
Tech stack
sql
Domain
data

Research direction

This is a quarterly analysis with tables, charts, and methodology; the payload does not name repository files, tests, or a data entry point to start from. Read the linked reviews #834 and #910 for context. The issue does not specify a code change or completion criteria, so it is not directly actionable as a first contribution.

Written by the indexing model from the issue text.

Description

Summary

This is the Q3 2026 follow-up on the energy fee reduction (energy unit price 210 → 100 sun, committee proposal 104, effective 2025-08-29). It continues #834 (Q1 2026) and #910 (Q2 2026) and extends every framework through 2026-09-30.

Headline: staking now covers more than 90% of on-chain energy, and I think this is the number everyone should be watching. Otherwise, Q3 continued the pattern #910 described: structural inflation deepened further while activation kept narrowing.

  • Staking: the staking share of energy rose from 86.3% before the cut to 90.2% in Q3 and 91.1% in September, so less than a tenth of energy is still paid by burning TRX. At Q3's demand and the 100-sun price, burn would only match issuance if about 17% of energy were paid by burning — a staking share of about 83%, below even the pre-cut level.
  • Activity: counted as daily USDT transactions in each amount band and measured against the pre-cut window, the $1–10 band strengthened again (+67% since the cut, +74% in Q3 alone), while the $10–1000 core band slowed (+6.4% since the cut, +1.5% in Q3 alone). Total energy demand eased to 190.8B/day (−3.7% vs H1 2026).
  • Burn and supply: in September daily burn covered 60% of issuance, and annualized net supply rose to +0.60%. That was the eleventh positive month in a row and the highest monthly reading since July 2021.
  • Why burn fell: Q3's TRX burned for energy (daily average) was 65% below the pre-cut window. In log terms, the unit-price cut accounts for about 70% of that change and the shift of energy payment from burning to staking for about 32%; slightly higher demand offset about 2%.

Quarterly cost table (2025-Q2 → 2026-Q3)

Period Avg TRX Price Energy Unit Price (sun) USDT Transfer Fee (Burn) SunSwap V3 Fee (No Subsidy)
2025-Q2 $0.26 210 $3.52 $12.68
2025-Q3 (Jul 1 – Aug 29) $0.32 210 $4.39 $15.80
2025-Q3 (Aug 30 – Sep 30) $0.34 100 $2.18 $7.85
2025-Q4 $0.30 100 $1.90 $6.85
2026-Q1 (Jan/Feb) $0.29 100 $1.87 $6.73
2026-Q2 $0.33 100 $2.14 $7.72
2026-Q3 $0.33 100 $2.14 $7.69

Fees were flat in Q3. TRX averaged $0.33 in both Q2 and Q3, so a USDT transfer paid by burning still costs $2.14 and a SunSwap V3 swap $7.69. That is 49% of the Jul 1 – Aug 29, 2025 level, the last period at 210 sun ($4.39 / $15.80), and 61% of the 2025-Q2 level ($3.52 / $12.68), when TRX was cheaper.

Method:

  • Formula: fee = reference energy × unit price (sun) × average TRX price (USD/TRX) × 1e-6.
  • Reference energy: the fixed values used in #910 — 64,285 for a USDT transfer to an address that already holds USDT, and 231,561 for a SunSwap V3 swap. I have not measured them. Actual energy per transaction varies with the contract's dynamic energy factor and, for swaps, with the route.
  • Average price: the arithmetic mean of TronScan's daily TRX price over the period.
  • Periods: 2025-Q3 is split at the cut, into Jul 1 – Aug 29 (210 sun) and Aug 30 – Sep 30 (100 sun). The 2026-Q1 row covers January–February, as in #834.

Section 1 — Energy demand eased slightly in Q3

Total on-chain energy averaged 190.8B/day in Q3 2026, 3.7% below H1 2026 (198.2B/day). The monthly averages were 194.3B (Jul), 188.4B (Aug) and 189.6B (Sep); August and September are the two lowest months since the cut. The 12 full weeks within Q3 (Jul 6 – Sep 27) ranged from 178.9B to 195.3B/day, all below the H1 2026 average.

Demand is still above where it was before the cut: Q3 is 2.4% above the pre-cut window (186.4B/day, Aug 1–29, 2025), and the average since the cut is 198.8B/day (+6.7%).

Daily and weekly total energy consumption

Chart: Daily and Weekly Total Energy Consumption, 2025-01-01 to 2026-09-30. Weekly points are ISO-week means; the first and last points are partial weeks (Jan 1–5, 2025 and Sep 28–30, 2026).

Section 2 — Staking above 90%, net supply +0.60%

Monthly structural metrics, pre-proposal through 2026-09 (TronScan energy-split and supply series):

Month Total Energy (B/d) Burn share % Staking share % Annualized net supply Δ %
2025-06 (pre) 184.3 15.3 84.7 −0.87
2025-07 (pre) 184.8 15.0 85.0 −1.00
2025-08 (proposal) 185.0 13.5 86.5 −0.70
2025-09 199.2 14.9 85.1 +0.09
2025-10 208.9 15.5 84.5 −0.03
2025-11 208.4 12.5 87.5 +0.19
2025-12 209.1 12.8 87.2 +0.16
2026-01 201.8 11.7 88.3 +0.30
2026-02 190.0 12.5 87.5 +0.32
2026-03 198.4 12.1 87.9 +0.29
2026-04 202.2 11.7 88.3 +0.29
2026-05 195.6 11.2 88.8 +0.37
2026-06 200.3 10.0 90.0 +0.44
2026-07 194.3 10.4 89.6 +0.46
2026-08 188.4 10.1 89.9 +0.52
2026-09 189.6 8.9 91.1 +0.60

Staking share = 100 − burn share. It includes energy supplied by contract deployers' stake (about 2% of energy in Q3). The 2025-08 row is the whole calendar month, which includes the first two post-cut days.

Daily and weekly proportion of energy consumption by staking

Chart: Daily and Weekly Proportion of Energy Consumption by Staking, 2025-01-01 to 2026-09-30. Weekly points are ISO-week means; the first and last points are partial weeks.

Staking now covers more than 90% of energy. The staking share rose from 84.7% in June 2025 and 86.3% in the pre-cut window to 88.5% in H1 2026 and 90.2% in Q3, reaching 91.1% in September, the highest month in the table. The direct-burn share is the other side of the same number: it held at 10.0–10.4% from June to August, then dropped to 8.9% in September.

Net supply kept rising: +0.46% → +0.52% → +0.60% annualized from July to September, and it has been positive in every month since November 2025. In September, daily burn (2.36M TRX) covered 60% of issuance (3.92M TRX), leaving +1.55M TRX/day of net issuance. For comparison, burn covered 71% of issuance in June 2026 and 152% in the pre-cut window (Aug 1–29, 2025).

Note (issuance): TronScan's Generated series was 3,915,712 TRX/day on 68 of the 92 days in Q3, and no day differed from it by more than 1%. That is 136 TRX per block (the 8 + 128 TRX reward rate set by committee proposal 102, in force since 2025-06-13) × 28,792 blocks. Q3 averaged about 28,790 blocks/day, and no day exceeded 28,792.

Note (staked energy vs the network limit): staked energy averaged 172.1B/day in Q3, about 96% of the network's 180B/day staking-energy limit (getTotalEnergyCurrentLimit, unchanged since committee proposal 97 in October 2024). Used staked energy is restored gradually over 24 hours, so this ratio is a pressure reading, not a strict utilization rate.

Note (break-even burn share): in Q3, issuance averaged 3.92M TRX/day and bandwidth and other burns 0.68M TRX/day, so TRX burned for energy would have had to reach 3.24M TRX/day for total burn to match issuance. At Q3's effective price of 100.25 sun that is about 32.3B energy/day, or about 17% of Q3's 190.8B/day — a staking share of about 83%. Q3's staking share was 90.2%. Before the cut it was 86.3%; with a 13.7% burn share at 210 sun, total burn (including bandwidth and other burns) covered 152% of issuance.

Section 3 — Network economic footprint: fiat burn cost

Total TRX burn cost averaged ~$0.85M/day in Q3, compared with ~$0.99M/day in Q2 and ~$2.06M/day in the pre-cut window (Aug 1–29, 2025).

  • Q2 → Q3: TRX was flat at about $0.33, so the decline (−14%) reflects less TRX burned (2.96M → 2.55M TRX/day).
  • Per transaction: at a constant TRX price, the energy fee of a given transaction is 52% lower than before the cut (1 − 100/210). Bandwidth is still 1,000 sun, so transactions that also burn bandwidth see a smaller cut.

Energy consumption by burning and burn cost (USD)

Chart: Energy Consumption by Burning (top, with TRX price) and Total Burn Cost in USD (bottom), 2025-06-01 to 2026-09-30. In the bottom panel, the solid line is TronScan's total TRX burn in USD: energy 73.5%, bandwidth 18.3% and other burns 8.2% of the Q3 total. The dashed line is a pre-cut baseline for post-cut days: the energy part is re-priced at 210 sun (same energy burned, same TRX price), and bandwidth and other burns are left as they were. In Q3 the baseline averaged ~$1.53M/day, against ~$0.85M/day actual.

Where the burn decline comes from

TRX burned for energy (daily average) is the product of three factors: energy demand × burn share × effective unit price. The log changes of the three factors add up to the log change of TRX burned for energy.

vs pre-cut window (Aug 1–29, 2025) TRX burned for energy (daily avg) Change Unit price Burn share Demand
Pre-cut window 5.29M — 207.9 sun 13.67% 186.4B/day
Since the cut (Aug 30, 2025 – Sep 30, 2026) 2.37M −55.2% 100.2 sun (×0.482, 91%) 11.90% (×0.871, 17%) 198.8B/day (×1.067, −8%)
2026-Q3 1.87M −64.6% 100.3 sun (×0.482, 70%) 9.79% (×0.717, 32%) 190.8B/day (×1.024, −2%)

In brackets: the factor's ratio to the pre-cut window, then its share of the log change; a negative share means the factor offset part of the decline. The pre-cut effective price is 207.9 rather than 210 sun because 2025-08-29, which the window counts as pre-cut, was a mixed-price day. After the cut the effective price is 100.2–100.3 rather than 100 sun because its two inputs come from TronScan series that do not reconcile exactly: TRX burned for energy (energy-burn revenue in USD divided by the daily TRX price) is 0.2–0.3% above burned energy × 100 sun. Total TRX burn, including bandwidth and other burns, was −47.8% since the cut and −57.3% in Q3.

Activity was still above its pre-cut level in Q3: demand was 2.4% above the pre-cut window. But to offset the price cut at an unchanged burn share, demand would have had to reach about 2.1× its pre-cut level. The payment mix then widened the gap: the burn share fell from 13.67% to 9.79%, which alone cut TRX burned for energy by 28%.

Section 4 — USDT activity by amount band

Growth in the daily count of USDT transactions in each amount band vs the pre-cut window (Aug 1–29, 2025). Each transaction counts once, and only transactions with an amount > 0 are included.

Band #834 window (Aug 30 – Dec 31, 2025) Through Jun 2026 (#910) Through Sep 2026 Q3 2026 only
$1–10 +56.5% +65.3% +67.2% +73.8%
$10–1000 +8.5% +7.8% +6.4% +1.5%
<$1 +3.0% −16.9% −29.5% −71.4%
All bands +8.4% +7.2% +6.1% +2.4%
  • $1–10 keeps strengthening. The cumulative reading rose from +65% (through June) to +67%, and Q3 alone came in at +74%.
  • The core band has faded to flat. In Q3, $10–100 was +5.8% and $100–1k −0.8% (+1.5% combined). Larger transfers were flat too: $1k–10k −1.4%, >$10k +1.6%. Overall, Q3's USDT transaction count was 2.4% above the pre-cut window, in line with energy demand (+2.4%).
  • <$1 fell sharply, to about 6,740 transactions/day in Q3 from about 23,600 before the cut.
    • On the same amount > 0 basis, direct USDT transfers in this tier fell by 71–73% (a range, because my direct-transfer count did not filter out zero-amount transfers), so the drop is not limited to contract-routed transactions.
    • Whether these small transfers moved into multi-transfer transactions would need a transfer-level count, which I have not done.
    • The tier is small: 1.1% of USDT transactions before the cut, 0.3% in Q3. Without it, Q3's all-bands count would be +3.2% instead of +2.4%.
    • I have not looked into the cause.

Address lifecycle (raw, all addresses):

  • Since the cut, 94.5% of addresses that initiated USDT transactions had not done so in the pre-cut window.
  • 46.2% of pre-cut initiators transacted again at some point in the following 13 months; 15.5% were active in Q3.
  • Both cumulative figures tend to rise as the window lengthens (retention can only rise), so the change from #910's ≈93% / ≈45% through June is not a trend in itself.
  • These are raw counts by transaction initiator. The classified user-layer view behind #834's 69.3% and 62.6% was not redone this quarter.

Recovery indicators

#910 suggested waiting for the burn path to recover before any further cut. Three readings track that path:

Indicator Pre-cut window (Aug 1–29, 2025) Jun 2026 (#910) Sep 2026
Daily burn ÷ issuance 1.52 0.71 0.60
Annualized net supply −0.79% +0.44% +0.60%
Burn share of energy (staking share) 13.7% (86.3%) 10.0% (90.0%) 8.9% (91.1%)

June and September are calendar-month values.

All three moved further away from recovery in Q3.

Insight / path forward

Q3 makes #910's two-sided picture sharper.

  • At today's price, staking is too high for burn to offset issuance. Staking covered 90.2% of energy in Q3 and 91.1% in September. At Q3's demand, burn would match issuance only with a burn share of about 17% (staking about 83%), i.e. with TRX burned for energy about 1.7× its Q3 level.
  • Activity is still net positive, but narrower. Measured against the pre-cut window, the daily USDT transaction count and energy demand were both up about 2.4% in Q3. The gain is concentrated in the $1–10 band (+74%), and the $10–1000 core is back near flat (+1.5%).
  • The burn path keeps shrinking. Q3's TRX burned for energy was 65% below the pre-cut window. On top of the price cut, about a third of that change came from energy payment moving from burning to staking; demand was slightly higher, not lower.
  • Net issuance keeps rising. Net supply has been positive for 11 consecutive months and reached +0.60% annualized in September, when burn covered 60% of issuance.

The governance read is unchanged from #910, and the case is now stronger: none of the recovery indicators moved toward recovery in Q3.

  • Staking share: in my view this is now the number to watch. With more than 90% of energy covered by staking, the burn price applies to less than a tenth of energy use. At today's price and Q3's demand, burn would match issuance again only if the burn share returned to about 17%, i.e. if TRX burned for energy grew to about 1.7× its Q3 level. This data does not show what would move energy payment back from staking to burning.
  • Further fee cuts: they should wait until the burn path recovers, rather than relying on volume to compensate.
  • Activity diversity: as in #910, I see broader on-chain activity (DeFi, dApps, execution-heavy use) as the lever that widens the direct-burn path, though this review has no measure of it.

Data and method

  • Sources:
    • TronScan chart exports (Energy Consumption; Revenue from Burning; TRX Generated and Burned, which includes total supply and burn in USD), downloaded 2026-10-20. Used for the cost table, Sections 1–3 (including the decomposition) and the recovery indicators.
    • TRON on-chain transactions (public chain data: transactions that carry at least one USDT transfer, including transfers inside contract calls), every day from 2025-08-01 through 2026-09-30. Used for Section 4: the amount-band and direct-transfer counts (one count per transaction) and the address lifecycle (counted by initiating address).
    • Chain parameters and committee proposals from TronGrid (getchainparameters, listproposals), read 2026-10-20. Used for the unit prices, the reward rate, the staking-energy limit (which does not adjust automatically, since the adaptive energy model is off) and their dates.
    • The counts behind #834 (per-transaction and per-transfer USDT counts, and address counts on #834's classified user layer, for Aug–Dec 2025, also computed from public on-chain data), used only for the #834 checks below.
    • No data after 2026-09-30 is used.
  • Windows:
    • pre-cut: 2025-08-01 to 08-29 (29 days, as in #834);
    • #834 window: 2025-08-30 to 12-31 (124 days);
    • through Jun 2026 (#910): 2025-08-30 to 2026-06-30 (305 days);
    • since the cut: 2025-08-30 to 2026-09-30 (397 days);
    • H1 2026: 2026-01-01 to 06-30 (181 days);
    • Q2: 2026-04-01 to 06-30 (91 days);
    • Q3: 2026-07-01 to 09-30 (92 days).
    • Weekly figures are ISO-week (Mon–Sun) means, and "full weeks within Q3" have all 7 days inside the quarter.
    • 2025-08-29 is counted as pre-cut, as in #834. Because that day was a mixed-price day, TRX burned for energy is taken from TronScan's energy-burn revenue divided by the TRX price, not from the unit-price schedule.
  • Section 2 metrics:
    • Burn share is the ratio of monthly sums.
    • Annualized net supply is mean daily net increase × 365 / mean total supply.
    • Daily burn ÷ issuance uses TronScan's Burned and Generated series.
    • The long-run comparison ("highest monthly reading since July 2021") uses the full months January 2020 to September 2026 in TronScan's TRX Generated and Burned export.
    • Break-even burn share: the burn share at which Q3's TRX burned for energy (demand × burn share × Q3's effective price of 100.25 sun) plus Q3's bandwidth and other burns would equal Q3's issuance.
  • Figures carried over from #834 / #910:
    • Matching cells: cost-table rows through 2026-Q2 and Section 2 rows through 2026-06 match #910's printed values in 75 of 76 cells. The one difference is a display change: the 2025-Q3 (Jul 1 – Aug 29) price has a period mean of $0.3249 and is shown here as $0.32 (earlier posts showed $0.33, rounded from a 3-decimal value). The fee cells are unchanged.
    • The two Section 4 bands: #834 published +56% and +11%; on my basis, #834's window gives +56.5% and +8.5%. I could not reproduce +11% for $10–1000 on either the per-transaction or the per-transfer counts behind #834: the band is about +8.5% on those counts and on my on-chain count, and the $10–100 tier alone is +11.7–11.8%. So the counting-basis explanation given in #910 does not hold.
    • <$1: on my on-chain count, which the Section 4 table uses for every band, #834's window gives +3.0%. #910's "+15%" came from the per-transaction counts behind #834, which give +15.5% for the same window. Unlike the $10–1000 band, the two counts disagree on this tier: my on-chain count is about 11% lower after the cut but only 0.2% lower before it, and I have not found why.
    • #834's 62.6%: the share of users classified Med-Large before the cut who were still classified Med-Large after it, on a classified user layer. That is a tier-stay rate, not a retention rate (on the same classified basis, 78.4% of all pre-cut active user-layer addresses, across all tiers, transacted again after the cut), and it is not comparable to the raw lifecycle figures in Section 4.
Dominant language
No language data
Stars
341
Forks
347
PR merge metrics
No merged PRs in 30d

Getting set up

This project ships no dev container, Dockerfile or contributing guide, so setting up is up to you: start from its README, and see our first-contribution guide for the general steps.

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from tronprotocol/tips

All issues in tronprotocol/tips

Similar issues

More Data Engineering issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.