A USDT transfer on Tron uses 64,285 energy when the recipient already holds USDT and 130,285 when their balance is zero. About 77% of both figures is a surcharge Tron adds to its busiest contracts, and the USDT contract is pinned at the maximum rate.
On 21 November 2026 I pulled the receipts for 39 consecutive confirmed USDT transfers, replayed transfers against live state with TronGrid's triggerconstantcontract, and read the chain parameters the network returned that day. Two of the pages ranking for "USDT TRC20 fee" still price energy at 210 or 280 sun, which doubles or nearly triples today's cost. None of the top results I read split the 65k and 130k figures into base cost and surcharge, and that split is the part you need if you budget payouts.
How Much Does a USDT TRC20 Transfer Cost in 2026?
Every TRC-20 transfer consumes two resources: energy to run the contract and bandwidth for the transaction bytes. If the sending account has neither, it burns TRX for both at the posted prices. Energy has cost 100 sun per unit since proposal 104 passed on 29 August 2025. Bandwidth costs 1,000 sun per byte. One TRX is 1,000,000 sun.
| Recipient's USDT balance | Energy | Bandwidth | TRX burned | USD at $0.3028 |
|---|---|---|---|---|
| Non-zero | 64,285 | 345 bytes | 6.7735 | ~$2.05 |
| Zero | 130,285 | 345 bytes | 13.3735 | ~$4.05 |
| Amount above sender's balance (reverts) | 8,624 | 345 bytes | 1.2074 | ~$0.37 |
Those are worst-case numbers, and almost nobody pays them. Of the 39 transfers I sampled, 37 burned no TRX for energy at all, because the sender had staked, delegated or rented energy. One burned the full 6.4285 TRX and one topped up a small shortfall with 0.0268 TRX. Bandwidth split differently: 22 used free or staked bandwidth and 17 burned 0.345 TRX. Six of the 39 went to zero-balance addresses.
Exchange withdrawal fees show the same gap. Binance charged 2 USDT to withdraw USDT on Tron on 21 November 2026, and Kraken lists 1 USDT. Kraken's fee is below the $2.05 burn cost of a single transfer, which only works if Kraken isn't burning.
Tron Energy vs Bandwidth: What Each One Pays For
Bandwidth is charged per byte of the signed transaction, and a USDT transfer call is 345 bytes. Every account gets 600 free bandwidth points per day. That allowance started at 5,000, fell to 1,500 in September 2021 and to 600 under proposal 87 in July 2023. Free and staked bandwidth both regenerate over a rolling 24-hour window.
The rule that catches people: if available bandwidth cannot cover the whole transaction, Tron burns TRX for all of it, not just the shortfall. With 600 free points, an unstaked account gets one free USDT transfer per 24 hours. The second costs 0.345 TRX even though 255 points are sitting unused.
Energy pays for execution in the Tron Virtual Machine and has no free allowance. Super representatives set its price by vote, not by auction. The on-chain proposal list shows it at 20 sun in 2018, then 10, 40, 140, 280 and 420 by December 2022. It has been halved twice since:
| Proposal | Passed | Energy price | Cost of a 64,285-energy transfer |
|---|---|---|---|
| 79 | 4 Dec 2022 | 420 sun | 27.0 TRX |
| 95 | 19 Sep 2024 | 210 sun | 13.5 TRX |
| 104 | 29 Aug 2025 | 100 sun | 6.43 TRX |
Account activation is a separate charge. Creating a Tron account costs 1 TRX, plus either 0.1 TRX or staked bandwidth. A TRC-20 transfer never triggers it. In java-tron's BandwidthProcessor, only account creation, TRX transfers and TRC-10 transfers count as creating an account; TriggerSmartContract does not. A customer can receive USDT at a brand-new address and then be unable to move it until someone sends that address TRX, which activates it.
Why Does Sending USDT to a New Address Cost Double?
It isn't about the address being new. It's about the recipient's USDT balance being zero.
USDT keeps balances in a storage mapping. Writing a non-zero value into a slot that currently holds zero costs 20,000 base energy, while updating a slot that is already non-zero costs 5,000. The 15,000 difference is then multiplied by the surcharge in the next section, which turns it into 66,000.
I tested the case that the phrase "new address" hides. A transfer to an address that held USDT earlier on 21 November and has since been emptied costs 130,285, the same as a random address that has never been used. Emptying a balance earns no refund either: a sweep of an entire balance and a sweep that leaves one base unit behind both cost 64,285.
That matters for deposit-address designs. If you sweep each customer deposit address to zero, the customer's next deposit into it pays the 130,285 rate. Leave 0.000001 USDT behind and the sweep costs you the same, while the next deposit costs whoever sends it 66,000 less energy.
What Is Tron's Dynamic Energy Model?
The Dynamic Energy Model (TIP-491), switched on by proposal 83 in February 2023, charges extra to contracts whose energy use exceeds a threshold within one maintenance cycle. A cycle lasts six hours. The rule, with today's parameters:
if contract_usage_last_cycle > 5_000_000_000:
factor = min((1 + factor) * 1.2 - 1, 3.4) # +20% per cycle
else:
factor = max((1 + factor) * 0.95 - 1, 0) # -5% per cycle
energy_charged = base_energy * (1 + factor)
Proposal 95, in September 2024, raised the cap from 1.2 (a 2.2x multiplier) to 3.4 (4.4x) and the threshold from 3bn to 5bn. On 21 November 2026, getcontractinfo returned energy_factor: 34000 for the USDT contract, which is the cap expressed in units of 1/10,000.
Every receipt shows the split. energy_usage_total is 64,285 and energy_penalty_total is 49,635, which leaves a base cost of 14,650. Without the surcharge, a USDT transfer would burn about 1.81 TRX including bandwidth, roughly $0.55.
Two details matter for anyone writing a fee estimator:
- The factor moves slowly. If every cycle stayed under the threshold, it would take about 29 cycles (7.2 days) to decay from the cap to 1x, and about 8 cycles (2 days) to climb back. Hard-coding 64,285 is safe today. Hard-coding 14,650 is never safe.
- Tether does not share the cost.
consume_user_resource_percentis 30, which suggests the deployer covers 70%. But the deployer's energy limit is 9,343, already spent, and no receipt I pulled showsorigin_energy_usage. The caller pays everything.
Burn, Stake or Rent: Cheapest Way to Get Tron Energy?
There are three ways to pay for energy, plus a wallet-level workaround.
Burn TRX. No setup, no capital, posted price. Right for low volume and occasional transfers from cold storage. Stake TRX. Under Stake 2.0,freezebalancev2 locks TRX in exchange for energy. Your share of the network's 180bn daily energy is your stake divided by the total staked for energy, which on 21 November came to about 9.66 energy per TRX per day. One daily transfer to a non-zero balance needs roughly 6,653 TRX staked (about $2,000), and a zero-balance transfer needs about 13,483 TRX. Unstaking takes 14 days. Measured against burning, a fully used stake avoids 2,346 TRX of fees a year on 6,653 TRX locked, about 35% a year before voting rewards. You carry the TRX price risk for the whole time.
Delegate or rent. delegateresource lends staked energy to another address. This is how an exchange funds thousands of deposit addresses that hold no TRX, once each address has been activated. A delegation can be locked for a number of blocks: 86,400 (about three days) by default, up to 864,000 (about 30 days). Otherwise undelegation is immediate. Third-party markets sell delegated energy below the burn price.
GasFree. GasFree, published under the JUST DAO name, gives each user a CREATE2 address so that a relayer pays the energy and the user pays a fee in USDT. Sources quote its activation and transfer fees differently. Read them from the live interface before you depend on it.
What Happens When a TRC20 Transfer Runs Out of Energy?
fee_limit is the most TRX, in sun, that a call may burn for energy. The network caps it at 15,000 TRX. It is a ceiling, not a charge: a successful call, or one that ends in REVERT, pays only for the energy it used. A USDT transfer for more than the sender's balance reverts after 8,624 energy.
The expensive failures are the others. In java-tron's VMActuator, OUT_OF_ENERGY, OUT_OF_TIME and most exceptions other than a revert call spendAllEnergy(), which consumes the whole energy budget that fee_limit allowed. The transaction is still recorded on chain and its bandwidth is still charged. receipt.result can be SUCCESS, REVERT, OUT_OF_ENERGY, OUT_OF_TIME, ILLEGAL_OPERATION, TRANSFER_FAILED or one of ten rarer values. Alert on anything except SUCCESS.
A concrete failure: an estimator written before September 2024, with the old 2.2x cap, would budget 14,650 × 2.2 = 32,230 energy. Size fee_limit from that today, on an account that burns TRX, and every USDT transfer runs out of energy and burns its full limit for nothing. The opposite mistake, setting fee_limit to the 15,000 TRX maximum "to be safe", means one call that hits a non-revert exception can burn up to 15,000 TRX if the account holds that much.
Two smaller traps. Tron's USDT transfer returns false even when it succeeds; my simulations returned 0x00 for every successful transfer, the same quirk I covered in the USDC vs USDT blacklist guide. And the chain parameter getMemoFee is 1,000,000 sun, so adding a note to a transaction's data field costs an extra 1 TRX.
USDT on Tron vs Ethereum vs Solana: Payout Fees Compared
Every chain here charges more for a recipient's first credit, though not always for the same reason. Here is what each one charged on 21 November 2026 (TRX $0.3028, ETH $1,622.80, SOL $67.14):
| Chain | Recipient already funded | Recipient's first credit | Surcharge | Who sets the price |
|---|---|---|---|---|
| Tron (burning TRX) | 64,285 energy, ~$2.05 | 130,285 energy, ~$4.05 | +66,000 energy | Super representative vote |
| Ethereum (0.16 gwei) | 46,109 gas, ~$0.012 | 63,209 gas, ~$0.016 | +17,100 gas | EIP-1559 fee market |
| Arbitrum (0.02 gwei) | 45,071 gas, ~$0.0015 | 62,171 gas, ~$0.002 | +17,100 gas | L2 base fee plus L1 data |
| Base (0.006 gwei) | 45,059 gas, under $0.001 | 62,159 gas, under $0.001 | +17,100 gas | L2 base fee plus L1 data |
| Solana | 5,000 lamports, ~$0.0003 | plus 0.00148844 SOL rent if no token account, ~$0.10 | Token account deposit | Fixed base fee plus rent schedule |
The Ethereum and rollup figures are medians from mainnet receipts. The 17,100 gap is EIP-2929 arithmetic: a cold write from zero costs 22,100 gas, while a write to an existing balance costs 5,000. Tron's gap is nearly four times larger in raw units because its surcharge multiplies the storage write as well.
Solana charges for space instead. A recipient without an associated token account needs one, and its rent-exempt deposit is 1,488,440 lamports today. SIMD-0437 cut rent from 6,960 to 5,080 lamports per byte, yet Solana's own token-account page and many guides still quote 0.00203928 SOL. On Solana the charge depends on whether the token account exists, not on its balance. When the account is closed, the deposit goes wherever its owner chooses, so rent you fund for a payout recipient is effectively theirs.
The headline is uncomfortable for anyone who picked Tron because it's cheap. At 0.16 gwei, a mainnet Ethereum USDT transfer costs less than 1% of a Tron transfer paid by burning TRX. Ethereum stays cheaper until gas reaches about 27 gwei at today's ETH price. What Tron offers in exchange is predictability. Its price changes only by vote, eight times in seven years, while Ethereum's changes every block.
One caveat: EIP-8037, part of Ethereum's Glamsterdam upgrade, would reprice the zero-to-non-zero write at about 97,920 state gas, roughly five times today's surcharge. It has no mainnet date yet.
Will USDT on Tron Get Cheaper?
My view: the multiplier will stay at the cap for as long as USDT dominates Tron's traffic, so any cut will come from the unit price, by vote. That is how the last two cuts happened, 420 to 210 to 100 sun, and nothing in the dynamic energy parameters points the other way.
The posted price already matters less than it looks. 37 of my 39 sampled transfers burned nothing for energy, so for operators it is a ceiling, not a bill. What hurts a payout team in 2026 is a stale estimate, not a price rise. You support Tron because your recipients ask for it, not because it's the cheapest rail.
What This Means for Payout and Exchange Engineers
1. Price each transfer from the recipient's balance. Call balanceOf first and budget 130,285 energy if it returns zero and 64,285 otherwise. Or run triggerconstantcontract with the exact transfer and read energy_used; in my tests it matched real receipts exactly.
2. Read parameters, don't hard-code them. Fetch energy_factor from getcontractinfo and getEnergyFee from getchainparameters on a timer.
3. Size fee_limit from the estimate. 20 TRX covers a zero-balance USDT transfer at today's price with about 50% headroom. Never default to 15,000 TRX.
4. Leave dust when sweeping deposit addresses. One base unit costs you nothing and halves the energy of the next deposit.
5. Delegate energy to deposit addresses with delegateresource instead of topping them up with TRX after the one-off activation, and undelegate when the sweep is done.
6. Alert on every non-SUCCESS receipt. Treat OUT_OF_ENERGY as a bug in your fee estimate, not as a network problem. Don't rely on the boolean that Tron USDT's transfer returns.
7. Plan bandwidth separately. Free bandwidth covers one transfer a day. Beyond that, stake about 213 TRX per daily transfer or accept 0.345 TRX each.
8. Warn recipients using fresh addresses. Receiving USDT doesn't activate a Tron account, so someone has to send the address TRX before it can move anything.
Key Takeaways
- A USDT transfer on Tron costs 64,285 energy, or 130,285 to a zero balance: 6.77 or 13.37 TRX if burned.
- 49,635 of the 64,285 is the Dynamic Energy Model surcharge, and USDT sits at the 4.4x cap.
- Energy has cost 100 sun since proposal 104 (August 2025). Pages that quote 210, 280 or 420 sun are out of date.
- At today's gas prices, Ethereum, Arbitrum, Base and Solana all undercut a Tron transfer paid by burning TRX by 40x or more.
getchainparameters before you trust any figure on this page, including mine.