USDT now travels Bitcoin on two different protocols. Taproot Assets went live on Lightning in March 2026. RGB launches this month, with UTEXO issuing. Same issuer, same dollar, same Lightning - but the architectures disagree about where the asset actually lives, who validates it, and what can fail. This note walks both rails.
This is not a protocol war between rivals. Tether shipped the Taproot Assets integration with Lightning Labs, then led the $7.5M seed round of UTEXO - the company issuing USDT on RGB. Both roads are paved with the same money. The fork is a portfolio decision.
USDT launches on the Omni Layer - the first home.
Tether discontinues Omni issuance. The original rail goes quiet.
Ardoino and Stark unveil USDT-on-Lightning at Plan B Forum, El Salvador.
USDT ships on Lightning via Taproot Assets after a 14-month integration.
UTEXO issues native USDT via RGB v0.11.1 - the module from No. 04.
Both protocols anchor to Bitcoin and ride Lightning. The disagreement is architectural: what goes on-chain, where the asset data sits, and who gets to check it. Switch the road and read the same four questions.
rgb-lib. Transfers hand over a consignment - a portable proof bundle - directly to the counterparty.rgb-lightning-node - LDK plus rgb-lib, the stack read at source in No. 04. Invoices carry RGB payloads; blinded UTXOs keep the receiver private.One payment, run down both rails. Watch what each protocol writes, and where. The speed class is the same - Lightning. The witnesses are not.
Payee issues an RGB invoice - the receiving UTXO is blinded, so the payer never learns it.
LDK finds a path across RGB asset channels. Intermediate nodes just move HTLCs.
The payer hands over a consignment; the payee's own rgb-lib verifies the full state history client-side.
The HTLC resolves; asset balances move along the channel instantly.
On channel close, a single cryptographic commitment lands inside a Bitcoin transaction.
Payee issues a Lightning invoice carrying the asset; a reusable static address works too (v0.7+).
Edge nodes swap USDT↔BTC at the boundary; core Lightning routes plain sats - no upgrades needed mid-path.
Ownership proofs check against Merkle-Sum Sparse Merkle Trees; universe servers store and serve the proof set.
The HTLC resolves; the edge node delivers the asset leg to the payee.
Asset state commits inside a Taproot output - indistinguishable from any other Taproot spend.
Dates and dollars, read off the record - not the roadmap.
Five axes, one dot each. The dot leans toward the road with the stronger claim today - tap a row for the reasoning on both sides.
History moves peer-to-peer as consignments; the receiving UTXO is blinded; the chain sees an opaque commitment. Nothing about the asset is broadcast to anyone who isn't a counterparty.
Far better than a public token ledger - but proofs are distributed through universe servers by design, so the proof set is meant to be syncable. Different posture, not an accident.
USDT live since 21 Mar 2026. Protocol at v0.8.0 with a stablecoin SDK (Jun 2026). Voltage, Speed, Joltz and Bitfinex already in the loop. Real dollars have moved.
v0.11.1 has run on mainnet since Jul 2025, and the wdk module shipped to npm - but USDT issuance starts this month. Mileage is about to begin, not accumulated.
tapd/litd are mature, Polar supports it for dev, and the v0.8 SDK targets stablecoin builders directly. Still labelled alpha, with breaking changes between versions - plan for migrations.
The stack I read in No. 04 - wdk-rgb-lightning over LDK + rgb-lib - is beta and moving fast (three footnotes resolved since filing). UTEXO ships APIs, an SDK and a live mint bridge on top.
Availability: if an issuer's universe goes dark and no mirrors exist, holders can lose the ability to verify or transfer. Multiverse federation mitigates - it doesn't yet eliminate.
Custody of history: your stash is your proof. Lose it without a backup and you can't demonstrate ownership. The burden moves from servers to clients - a different edge, equally sharp.
Tether spent 14 months integrating with Lightning Labs and put $8M into Speed, a Lightning-native payments firm, in Dec 2025.
Tether co-founded and led UTEXO's $7.5M seed in Mar 2026, alongside Fulgur and Boosty. Whoever wins this race, the same treasury collects.
Protocol elegance will not decide it - Tron's $83.9B incumbency says adoption will. Four signals worth more than any benchmark:
Which deposits go live first, and where. Bitfinex is already issuing on Taproot Assets; the RGB launch names "Tether Wallet and several exchanges." Watch the tickers, not the threads.
The rail a wallet picks when a user taps "send USDT" is the rail that wins. Self-custody support (Joltz on TA; Tether's own wallet on both) is the tell.
Each road's sharp edge, measured in practice: do TA universes stay mirrored and reachable; does RGB make consignment backups invisible to normal users.
UTEXO promises fixed, USDT-denominated fees with sub-second atomic settlement. If that holds under load, it is a genuinely new UX for Bitcoin - and pressure on the other road to match.
Things you only find by reading past the headline.
Tether integrated with Lightning Labs for road one and bankrolled UTEXO for road two. Framing this as RGB "versus" Taproot Assets misses the structure: the issuer bought both tickets. The real competitor is Tron's $1-4 fee and its embedded payment infrastructure.
read the cap table, not the discourseTaproot Assets carries production USDT and is still labelled alpha, with database migrations and RPC breaking changes between versions, and its seven BIPs still in draft. Production-grade adoption on pre-1.0 software is the normal state of this frontier - worth knowing before you build.
version discipline requiredTA's risk is a server going dark (universe availability); RGB's is a client losing state (stash custody). Neither is "worse" in the abstract - they assign the same burden to opposite parties. Your threat model picks your road.
symmetry worth namingThe RGB rail runs on the exact stack this series read at source level last issue - wdk-rgb-lightning over LDK + rgb-lib. Two of its footnotes have since been resolved upstream (versions realigned, bindings on npm), one is still open. Read the source walk ↗